DE10015858A1 - Natural language translation method for communication of related semantics involves using different declarations and symbols to pass suitable speech terms - Google Patents

Abstract

The method involves using the different combined declarations and grammatical and semantic contents corresponding to the number of speech examples provided by actions to pass suitable speech terms, uniform definitions and generation rules sufficient for indicating any language to be used. Diversity of manifestations and possible symbols in languages are considered in conducting multilingual communication of semantic contents. An Independent claim is also included for the processor-supported device used in communication of related semantic contents.

Description

The invention relates to a method for a semantic-related, one-way as well as multilingual communication in natural languages according to the preamble of claim 1 and a device for computer-aided implementation of this method according to the preamble of Claim 25. Details and refinements are in the subclaims specified which the variety of manifestations, combination and thereby determined possibilities of meaning of those in such languages introduced symbols and can be combined in different ways Utterances as well as the grammatical and semantic to be connected with them Consider content and use a repertoire adapted to it from a corresponding number of language examples, provide measures, which are to be carried out with new, suitable language terms, which uniform, in the specified way, universally in any language the definitions and production regulations to be used are sufficient so that the Communication task with a device available to the capacities customizable storage and processing effort is solvable.

So far, no satisfactory methods of the type according to the invention have been found Solutions are presented. As explained further below, there is a realization diverse, partly not yet sufficiently recognized and defined Problems, which above all also the complex organization of the natural languages to be used. This state accordingly exist, as well as references // 2 //, p. 218 and // 12 // can be seen that there are currently no uniform terminological conventions that here are some of the essentials for the following description of the used method according to the invention, partly also for a precise, proper assessment of the current status and the description of the The necessary terms to be used to solve the problem are: Under natural Languages L or languages L for short, // 1 //, p. 14 following, become institutions  understood, with the help of which, according to the current status, natural, People communicate with each other and using habitually used symbols interact, for which according to // 2 //, p. 235 finite or if necessary, unlimited amounts of symbol complexes are used that are formed over a finite vocabulary of symbols. in the In line with this definition, such interactions may also include multilingual communication activities, such as those of people are generally exercised when exchanging information about facts and, for example, when talking or exchanging letters, for example when granting and Receiving orders, asking and answering questions, Translating, interpreting, typing or even correcting statements attack. The common feature of these diverse activities is that their Carrying out uniformly both a mastery of the ability to generate understandable, sufficiently correct and clearly formulated language texts as well to understand if necessary also in coherent or not standardized form of uttered statements, which according to // 3 //, p. 156 not manageable without the above-mentioned mastery of creating language texts is. The cited language definitions leave the current state in detail, however, what kind of symbols are to be used, which ones Rules for their habitual use and assembly Complexes as a whole have to be observed and what connection between such complexes and the interactions of the communication partners is.

For the method according to the invention, symbols are generally referred to as Gs designated terms with properties H (Gs) used, which often with Probabilities p below one, further terms A as well as others Show symbols Gs and thus symbolize. Certain symbols will be Gs by a disjoint class of, correspondingly alternatively occurring, Complex Kp of speech signals T with properties H (T) displayed or also as certain class-forming or complex-forming superizations of others Symbols Gs formed. According to their properties H (Gs) are used for the Symbols Gs also use other terms such as Gr, Gt, G, Vt and Z etc. The above mentioned, for displaying symbols Gs in the available media for example, acoustically, optically, geometrically or electronically generated signals // 12 // are irrespective of their size and therefore differ from // 12 // as Texts T, not referred to as symbols. Cited in this description Examples of texts T are given by strings, in particular also Sequences of letters, represented or approximated, between simple Slashes are used; on phonetic alphabets and other notations is omitted for simplification. Certain properties of symbols like Gs symbol complexes Kp = GsGs are also considered to be their contents or meanings. Those as producers and senders as well as recipients and processors of texts T people communicating in one or more languages or Devices used in these roles according to the invention are also considered  Communication partner or partner Pa for short, the producers also as Authors At, the recipients also as listeners Hr and the users of such Devices also as users, where appropriate also authors At or listeners Hear what communication content translates into actions, act as users. A self-contained, possibly also several common sentences Comprehensive text T of an author At is called utterance Tp, one out of a or several utterances Tp composed by one or more partners Pa Communication process also as a communication act. As in // 3 //, p. 243 The oral and written versions of each are proposed in one Language community X generally use language as different treated natural languages L. Clear transfers of a text T = T1 a language L in other media, for example an acoustically demonstrable one Text T1 from the oral German language into an electronically coded one Text T2 of this language, with correspondingly unique assignments of Text features H (T1) to features H (T2) are usually the same text T1 = T2 treated.

According to the language skills of people, the one according to the invention To provide communication in one or more languages and thus closes the Translation of texts T into other languages L, by definition for example also from oral to written. As with Human language use is such a communication method Mastery of a number of functions, including those related to mentioned provision of content incoming, if necessary also in coherent or incorrect form of expressed texts T also the content-based drafting and, if necessary, correction of sufficient Texts to be understood understandable. That of an inventive Process both the ability to generate understandable texts T and Determination of the meaning of texts T is also to follow from the above cited insight that listeners hear their task of understanding the content of Texts T are only able to comply if they focus on that for the Regulations used during the generation of these texts can be used Conversely, authors make their texts understandable, even while avoiding them misleading ambiguity, can only ensure if they follow the procedure the listener takes Hr into account when understanding the content, so do they too is known.

As in // 3 //, p. 410f. and p. 445 traditionally, but without further Proposed definitions, contents of texts T are closer to below specified type differentiated into grammatical and semantic content, both also called meanings. Terms with from are also used to represent them // 12 // different definitions used. A text T that, even if it is as The utterance Tp stands alone in relation to a semantic assigned to it Contents of experts, defined as connoisseurs of the languages L and the the semantic content addressed therein, which is judged to be permissible, is regarded as  correct text called T = Tf. Issues differ from // 12 // regardless of whether they are addressed by texts T or not, as referred to pragmatically; accordingly are addressed by texts T. semantic content generally on pragmatic facts, in short also things called referred. Based on // 2 //, p. 219, certain relationships between Texts or certain parts of text T or grammatical content as well Parts of it as grammatical, references between parts of text T, also under Consideration of their grammatical content and pragmatic facts as semantic, between pragmatic facts, as they may also be be influenced by certain ideas of the communication partners Pa can, as pragmatic, and between certain parts of text T as syntactic designated. A guided by considering the grammatical content of texts T. Communication becomes grammatical or grammatical and one according to one Process of the type according to the invention with additional, if appropriate also communication based solely on pragmatic facts pragmatic or semantic-related, also called semantic for short. Corresponding is related to semantic content according to the invention Communication in natural languages L generally one prior to transmission also understood speech communication serving pragmatic facts. This definition takes into account that by a, a text T assigned semantic content, depending on the situation, also different, pragmatic issues, also referred to below as Wk can, such as in the following under the term context summarized situation, for example through objective facts, but also content to be assigned by adjacent texts T as well as by subjective ideas of the partners Pa etc. can be co-determined. Generally the facts also include objective, usually unambiguous facts one that can be used for further material purposes, for example also a combination with other facts, such as triggering, Changes, termination or control of certain operations.

To differentiate from the terms to be used according to the invention, the for written languages L intuitive, often also arbitrarily defined as words, terms marked by distances from other such words as they were also adopted for the description of oral languages L, strictly as called natural words or script words, according to the performance certain forms of inflection thereof in lexica also as lexical words. Accordingly, other units used in language descriptions such as Sentences, questions, commands, exclamations as well as units used for their construction such as Characters, letters, punctuation, syllables, sounds etc. natural sentences etc., called natural characters etc., with rules of spelling respectively the pronunciation summarized under prosody // 12 //. How outlined below, the ones to be used according to the invention differ Symbols Gs usually from the usual ones, here also under the designation natural symbols summarized, natural words, characters, etc.  

Languages L, whose texts T one-dimensionally lined up symbols fixed type of how words or characters or ideograms etc. are to be assigned, are called serial. Unless otherwise stated, this is done Description of the invention without restriction of generality using the example such serial languages L; an application also on multidimensional, is, for example, symbol complexes arranged in two dimensions basically also provided. For example, oral or handwritten texts T also by certain characteristics H (T) as well as at the same time resulting repertoires R (H (T)) of these, which are also considered to be corresponding multidimensional feature vectors are characterized.

Certain combinations of text T and content A are generally called Denoted news TA. The combination of a text with a grammatical content A = B is accordingly called grammatical message TB noted the combination with a semantic content C as semantic TC news. The ones commonly found in natural languages L. Combinations of semantic content C, each with a grammatical Message TB are also called semantic messages, noted as TBC, briefly also as a TC. According to the invention, semantic content C and pragmatic facts Wk as combinations of other pragmatic ones Describe terms K with properties H (K), which are referred to below as Technical terms K, also briefly referred to as things K.

Methods of the type according to the invention are also characterized in that according to // 2 //, p. 217f. as a prerequisite for the realization of technical Devices for human-machine communication in natural languages L are to be seen as they under function and device names like automatic translation, acoustic understanding of speech or hearing Typewriter, electronic credit agency and generally intelligent machines and robots with language skills were discussed.

As well as the descriptions of the current status given in // 2 // and // 4 // of communication systems designed for technical use can be seen are useful implementations of the functions required for this and devices and corresponding procedures to be assumed type according to the invention so far not before. Unsolved questions will be particularly some can be summarized under ambiguity, vagueness and contextual dependence Problem areas of natural languages // 12 // assigned. For example, an as Symbol understood lexical word, especially in the case of oral or handwritten lecture often through a large variety of speech signals, that is, texts T, so that, conversely, one is represented by the Letter sequence / tas / approximate sound sequence T different script German Words like / that / or / that / or parts of words like / cup / or / key / etc. can be assigned. Furthermore symbolizes, depending on one by other words of an utterance or a situation specific context  Scripture often identified different grammatical as well semantic content; the latter even if there is a clear grammatical for it Function was identified. So the English word / like / as a verb or Function with correspondingly different semantic content, that is grammatically clear, for example as a noun nominative singular identified German word / sound / semantically a mineral, a Music impression or, as in the text / sound. /, A microphone system symbolize, whereby, as defined above, a possibly to be observed Context not only through the other natural words of the utterance in question Tp, but also by the content of previous or subsequent Comments, the ideas of the communication partners Pa, their situations etc. can be determined. For example, an author At with the script word / large / an often vaguely defined spatial range, one by special achievements of emerging members of a professional group or each ironically address the opposite of this depending on the situation. Misunderstandings can also result from the fact that the partners Pa in the formulation and Interpretation of texts T start from different contexts, whereby People are often intuitively correct in a way that is not specified to decide which contextualization // 12 // when uttering Tp should be made appropriately. From the partners Pa is here requires a communication content from an by definition unlimited Diversity of symbol complexes as well as of contexts too open up.

Font T also uses // 12 // for punctuation non-alphabetic punctuation marks both grammatical and assigned semantic functions, i.e. content, without clarifying whether these characters as symbols equivalent to lexical words or as, parts to be assigned to them, what content they should be treated and how they could be combined with other content. Given the diversity and ambiguity encountered in previous ones Terms used in language descriptions remained unclear by what type of Symbols and terms mean // 12 // of texts T as well as of Contexts expediently express what properties such symbols assign according to which rules existing texts and text sections T with to relate to them and to be divided among them and in what way one the meanings of symbol complexes from the possible meanings of the Combine individual symbols and the influence of further context data could take into account. Given this variety of problems, according to // 3 //, p. 411 no one even outlines a satisfactory and comprehensive semantic theory offered. It remained open which basic building blocks, with which Interrelations, a method of the type according to the invention at all could be built.  

It is known that according to // 1 //, p. 227f people are intuitive expiring first language acquisition usually to one, also intuitive expiring, mastery of the listed and others in languages L are capable of encountering individual problems, but according to // 1 //, p. 16 u. P. 47f. neither specify how your language acquisition takes place nor what data, terms and rules they acquire and use them in communication, so that also a realization of a method of the type according to the invention by the desirable transfer of human skills to working principles of machines, such as those demonstrated for mathematical problem solving was previously not possible with language problems.

As in // 1 //, p. 128f. stressed, in particular the question of the The meaning of language has so far not been satisfactorily answered, whereby it is even doubted whether language is a bridge between sound and meaning beats and meaning can be described at all by terms. Also proposed formal, developed according to the principles of logic Semantic theories, what criteria like the "truth value" // 12 // of sentences are considered unusable according to // 1 //, p. 164. So it seems // 2 //, p. 239 the problem of ambiguity mentioned is so high Differentiation of the symbols to be used requires that a computational controllable processing is no longer considered possible. This turns out specifically the problem of exponential with the number of symbols increasing combinatorial explosions of data to be considered as they both in the description of the symbols including their diverse mutual reference options as well as the corresponding variety Decision trees to be processed for their assignment to given texts T and accrue to each other. Furthermore, the according to // 3 //, p. 413 appears as "reference" designated relationship between text forms T and thereby designated, also called "Relata", objects or content for a computer-aided Treatment not formulated specifically enough. From those expressed in // 1 //, p. 219 Doubts about the usability of terms can also be concluded that the According to // 3 //, p. 411 the usability of offered semantic systems can be determined is also because it has not yet been clarified from which components with what properties, and in what way, by the text T as well as by the addressed semantic content to be taken into account in contextual situations as well as the pragmatic facts to be connected with it could be. So it wasn't possible after // 12 // or // 3 //, p. 138, one to be clarified as a prerequisite for methods of the type according to the invention Methodological distinction between grammatical and semantic content Specify the type with the corresponding symbols to be assigned. You go after // 3 //, p. 170f. assume that, but only for written languages fixed, lexical words the role of said habitually According to // 2 //, p. 227, criteria used so far are missing, according to which they can be classified satisfactorily. For example, hold further purely formal criteria such as "well-formedness" or  "Formality" according to // 1 //, p. 47 did not stand up to criticism. Also were for the as hierarchical relationships or dependencies deemed significant // 12 // between proposed categories of natural words according to // 2 //, p. 233 and P. 238 contradictory ideas developed. So conventional ones turned out to be Language terms such as verb, tense, person, predicate etc. as well additionally proposed constituents such as verb phrase, noun phrase etc. // 12 // for the development of communication processes and systems as not being sufficiently strict, for example the Use discontinuous, ie in texts T by sections T of others Symbols of interrupted symbols were declared inadmissible // 12 //. Suggested phrase structure, also called context-free // 12 // Grammar systems // 12 // did not allow the influence of, for example contexts given by previous utterances Tp at the Meaning analysis of texts T to be taken into account sufficiently // 13 //.

After // 12 // general, it was particularly difficult to draw the border between the semantic meaning of a "context invariant" linguistic statement and what is meant by it in each individual case, often pragmatic intended for clear further use Situation, since this mostly depends on the specific, above under Context summarized use situation depends. Without one However, drawing a line is the linguistic transmission from, to the respective one Situation-related, pragmatically clear facts Wk that another material use can not be used.

For terms used tentatively as language symbols, also because of the listed problems and deficits, furthermore not sufficient Context is given between the question of what information texts T regarding grammatical and possibly displayed contents semantic type and information theory, as described for example in // 5 //, P. 87f. was roughly formulated. According to current usage, Information is often the subject matter itself and not a specific one understood the quantitative property of such a transmission. For example was disputed in // 6 // that a natural language is information at all transmit; it rather calls only an interaction - not specified in more detail between what was said and the listener. Furthermore has been on so far proposed symbols criticized on the one hand that they are not based on empirical allow traceable characteristics to be traced // 7 //, p. 559, while on the other hand their According to // 1 //, p. 16, authors do not want to understand themselves as empiricists and the Even deny the need for such observability // 8 //, p. 124. Da However, it can generally be assumed that technical processes are complete generally have to deliver observable results which result from the Interaction of objects result, their properties also in principle with a level of accuracy and consistency usual in technology the observation must also be accessible for procedures  type according to the invention in addition to the fundamentally given observability generated text T also one for the symbols and thereby transmitted contents composed of this, if necessary also for triggered thereby Actions to demand. If one also assumes that such processes in addition to Transmission of texts T also the transmission of information about such Symbols and content have to serve, so can previously proposed methods with deficits in the areas of observability and information allocation generally not be suitable as solutions.

Given the problems encountered in setting for description of texts T suitable language terms and categories, which also in the Uselessness of experiments developed according to such principles technical devices found their expression // 4 //, // 9 //, under consciously accepted renunciation of an "understanding" of meanings Purely statistical language models are used, which are based on those for written languages specified lexical words and the "hidden Markov-probabilities" designated frequencies of their due to the preceding words Sequence based, with up to 3 consecutive words, also called trigrams, could be taken into account // 10 //. An extension this number is due to the problem of combinatorial explosion observing data, for example with a repertoire of 20,000 Words can already contain more than a trillion trigrams // 10 //. there but is correct for word sequences like / what he said yesterday / for that Occurrence of the last word only then a usable, from the average for any words different frequency to be expected, even if the 5th preceding word is included, so that also hexagrams and each after text T, even more extensive complexes would have to be taken into account. Methods based on such available language models are also available at necessary limitation to a few 1,000 words depending on the intended function still with unsatisfactory error rates of approx. 15% to 30% // 11 //.

According to the described, characteristic of the current state Unresolved problems are known inventions deal with some sub-aspects of natural language Communication processes. So refer to DE 30 42 408 C2, DE 31 42 540 C2, DE 35 03 233 C2, DE 36 15 972 A1, DE 36 16 011 A1, DE 30 32 664 C2 and DE 31 51 106 A1 specified methods or devices Questions of electronically operable encyclopedias that work as a translator To give people, from him already from the rest of the present text T divided and correct also with regard to grammatical categories to be assigned identified natural words or phonemes of a Source language L some words or phonemes of a target language L ' offer, but still the intelligence and language knowledge of people assume to be acoustic from a continuous flow respectively written text signals at all the words or even phonemes respectively  Identify morphemes of the source language L and the problem of from to resolve the respective text context dependent ambiguity. An in DE 41 35 261 C1 presented "translation system" serves to reduce the Amount of data required for the lexical assignment of foreign language words identified compound words, i.e. composites, of the source language is needed. In the one proposed in DE 43 11 211 A1 for translation work Text analysis system should also be specifically taken into account that German verbs different forms of Dative objects can be assigned. DE 42 32 482 A1 suggests translations under Intermediate an artificial standard language without performing Solutions to the problem of ambiguities in the output and Specify target languages. Process for the "speech recognition" designated task, as in DE 32 11 313 C2, DE 37 33 391 C2 and DE 35 14 286 A1 described, also offer no useful answer the question of which of the alternatively possible assignments of natural Words and content of a grammatical or semantic type depending on Prefer textual context and what pragmatic issues with it depending on the situation.

The invention has for its object a method of the aforementioned Type for one according to the existing specifications in one or more natural Languages, also multilingual, to indicate which communication to lead with, to be written in the media provided for this purpose texts to be recorded, which are thus carried out with semantic content Reference is made to the fact that this makes transmission more pragmatic, that is also for further material uses of suitable facts, also under Taking into account the respective context, including one Facts preserving translation into texts of other natural languages, too from oral to written and vice versa, is possible, in such Cases where given data or inaccurate or incorrectly written texts do not allow sufficiently secure and clear text-content assignments, corresponding markings such as text or content corrections are also possible, and a facility for describe computer-aided implementation of the method, so that when current problems solvable and mechanical devices which have so far been possible in natural language communication from People take on functions of the type mentioned and thus such communication between man and machine and between Make machines possible.

This object is achieved by the claims set out in claim 1 further implemented measures, in one embodiment also using the in Device described in claim 25, within the scope of a designated as Mp Procedure resolved. Process Mp provides communication in natural Languages L between communication partners Pa taking into account existing  brought in by users of the method, possibly also partners Pa, To perform specifications Vr with reference to semantic content F, by which pragmatic facts formed as complexes Kp = KK of technical terms K Wk are transmitted. This communication is done by, from authors At Speech signals of the intended for the natural languages L concerned Media to be formed and issued and by their recipients Mr. Texts T to be evaluated, which by assigning grammatical Contents N considering grammars p (TN) Messages TN are supplemented, which continue by assignment semantic Contents F considering semantics the probabilities p (TNF) TNF messages are supplemented, being pragmatic to be transmitted Issues Wk = Wk2 by changing, from the communication partners Pa prerequisite, pragmatic facts Wk = Wk1 according to the semantic content F of these semantic messages TNF are generated. These semantic messages (TNF) are listed as complexes Kp = EmEm Consideration of the data D, if any, as well as other data Components Em conditional, probabilities p (Em) from components Em put together, which are described by data Wp technical terms K and, in Adapt to natural languages L including symbols Gs. These each comprise a disjoint class K1 = Vt of text sections T = Ts superimposed text symbols Vt = (Ts), whose texts Ts each refer to certain, agree on tolerances dH of more precisely defined text properties H (T), and from this to disjoint classes K1 = Gr superimposed grammatical symbols Gr = (Vt), each a repertoire R (Pr) of, closer to probabilities p certain, possibilities of TN them in grammatical messages to assign assignments Pr, which them and each of them assigned text symbols Vt a Gr a rank Ra and one or more Assign positions P, which are assigned to these text symbols Vt Include text sections Ts a Vt in texts T. These text symbols are Gt each a disjoint class K1 = (Fp) of, the possibilities of their arrangements Pr taking into account semantic information Fp, each one Technical term K and a repertoire R (K1) of disjoint classes K1 = (q) from Instructions for use q for which, taking into account the arrangement Pr assignment to semantic content F of semantic content News TNF including its association with subject terms K others Text symbols Vt present in the relevant semantic message and with Subject terms K of a prerequisite prerequisite Wk1 including its modification to a matter of WW2. Here are pragmatic facts Wk2 to be transmitted by such texts T. taking into account existing data D including the specifications Vr in each case generated by one or more grammatical symbols displayed thereby Gr and a permutation of these associated text symbols Vt below Attention to the arrangements Pr combined into a grammatical content N. which, through a permutation of these symbols Vt assigned positions P given by these arrangements Pr,  optionally also discontinuously arranged text sections Ts combined text T is added to a grammatical message TNF, and that the subject terms K of a permutation of these text symbols continue Vt associated semantic clues Fp according to a permutation of these Instructions Fp assigned usage instructions q taking into account the by given a pragmatic prerequisite Wk1 Probabilities p (K) 1 as well as other ones present in the data Wp Probabilities p (Em), as a complex Kp = KK of these technical terms K with Probabilities p (K) 2 combined semantic content F formed through which this grammatical message TN taking into account the Probability p (TNF) is added to a semantic message TNF and the assumed pragmatic situation Wk1 in the to be transmitted pragmatic facts Wk2 is changed, which as from the, until then led, communication resulting situation Wk2 one of the specifications Vr appropriate further use is to be fed.

Method Mp provides to use technical terms K, each as one, also as Repertoire R (A) referred to, set K = R (A) one or, as a rule, several other terms A can be described. Complexes Kp = BB of terms B. generally defined by the fact that they each have one or, as a rule, several Contain terms B, in the latter case certain references between at least 2, but possibly also several or all of them Terms B a Kp exist. With the pragmatic facts mentioned Wk, the are formed as complexes Wk = Kp = KK from technical terms K according to their definition as sets K = R (A) between them, as they generally used for linking sets and below at the detailed description of the linking instructions q in detail are. As noted above, the association of a term A with a set B and thus also to a repertoire R (A) of a class K1 (A) or a complex Kp (A) generally characterized by A a B, A a R (A), etc. In different The invention provides refinements for the natural ones to be used Languages L each next to the repertoire R (Vt) of text symbols Vt also use other repertoires R (Gt) of other text symbols Gt whose Texts T a Gt also determined by a repertoire R (H (T)) Text features H (T) are marked. The following are general Text symbols Vt, with Vt a Gt, also the text symbols Gt called Gt attributed and also referred to as such.

The invention provides for the possibility of a communication act There is a disjoint class K1 = (Wk1x) of possible pragmatic ones Facts Wk1x = Wk11, Wk12 etc., of which according to the provided that the probabilities p (TNF) are preferred Situation Wk1x = Wk11 is assumed, with which, taking into account the Data Wp results in a semantic content F, which this message TN with highest probability p (TNF) that does not fall below a threshold value ps  is assignable. According to the invention, this takes into account that the Communication partners different according to the possibilities Knowledge and different opinions also from different ones Facts Wk1x can go out. As far as possible, the Communication of a situation Wk1x to be assigned to partner Pa provided. The class K1 = (Wk1x) required for communication of facts Wk1x becomes the context of each Communication act attributed.

A resulting pragmatic situation WW2, also expressed in a simplified way the part of Wk2 deviating from Wk1 as well as the part responsible for it semantic content F, is also present as a message of a grammatical Message TN as well as its associated text T in the relevant Referred to contextual situation. The combination of grammatical and NF semantic content N and F of a TNF message is also called NF content referred to the relevant text T.

In a preferred embodiment, text symbols Vt are superimposed on a grammatical term Gr, for their connectivity in arrangements Pr, which can also be demonstrated by the transformations Qs below, or non-connectability with other symbols Vt in a semantic message TNF, possibly also by the The presence of further symbols Vt-related probabilities p (Vt) apply, which correspond within predetermined tolerances dp, the tolerances dp being adapted to the capacities of the available devices and the communication files to be performed in a preferred embodiment. Deviating from known transformation grammars // 12 //, generative grammars and semantics // 12 // and constituent structure grammars // 12 // as well as unification grammars // 12 // including those used therein, also as "Generalized Phrase Structure Grammar" // 12 // or also "Lexical Functional Grammar" // 12 // designated representations of functional including syntactic relationships in natural languages L are used in the method Mp according to the invention Mp corresponding to the transformations Qs by text as well as text complexes T represented text symbols Vt, which may also be only parts of lexical Words or, if necessary, also discontinuously arranged complexes Kp of several lexical words, which in written languages L also include punctuation marks, spacing as well as operators o defined below for the modification of assigned text sections T and thus differ from the grammars cited in the quotations // 12 // used Ph rasen // 12 //, as in // 12 //, p. 816f. or also in // 1 //, p. 116f. be discussed, differentiate. This has the consequence that the symbols Vt can be combined into a comparatively small number of grammatical symbols Gr, which makes it possible to describe the assignability of the symbols Vt to semantic messages TNF with an advantageously small amount of data assigned to the symbols Gr. The rank Ra defined in more detail below also differs from the dependency relations introduced for known dependency grammars // 12 //, for example by substitute samples // 12 // defined in more detail between the grammars // 12 // cited above as constituents of terms adopted by natural sentences.

The dependencies of the dependencies to be taken into account according to the invention Probabilities of certain quantities A of, generally noted as p (A) other sizes such as B or B and C or B, C and D etc. are conditional Probabilities of the form p (A / B), p (A / B; C), p (A / B; C; D) etc. expressed. Accordingly, taking into account a pragmatic situation Wk1 probabilities of p (TNF) semantic messages to be taken into account TNF is also strictly noted as p (TNF; WW1). When adding a given Text T for a grammatical message TN is next to the above Value p (TN) also has a conditional probability p (N / T) to consider for its Supplement to a semantic message TNF in addition to p (TNF) also the conditional Value p (F / TN), strictly also p (F / TN; Wk1), is required. As below are generally used in the formation of sizes A due to with Security, ie p (D) = 1, available data D conditional probabilities p (A / D) to be observed.

The assignment of semantic content F to texts T of grammatical Messages TN intended consideration of conditional probabilities p (F / TN; Wk1) has the advantageous consequence that for the respective context preferable, by terms K with probabilities p (K) closer certain meanings of symbols Vt both contexts are taken into account as determined by the grammatical message TN, Textual contexts are given as well as contexts as given by those who respective situations as well as previous statements Tp Issues Wk1 are described, from the intended use quantitative probability data p advantageously also follows that a as Facts to be preferred communication issue Wk2 often also in for natural languages L characteristic ambiguous communication situations can be achieved, both by the presence of an ambiguous text T, which both disjunct several grammatical contents N and semantic Contents F and corresponding disjoint repertoires R of news TN and TNF are assignable, as well as by the existence of ambiguous, by a disjoint repertoire R (Wk1x) of situations Wk1x described situations can be marked.

Deviating from the ones previously described in // 1 // and // 13 //, for example The result of a communication act is not based on semantic models an already context-related semantic or pragmatic content of a Expression Tp is limited, as it is inventively determined by the semantic content F a message TNF of a text T is given; rather it becomes a rule  much more extensive than complex Kp a much larger number of Terms K formed facts Wk2 as communication result Wk2 used, in which not only the probabilities and Probability relationships p (K1) properties H (K1) of the in the content F represented technical terms K = K1, but possibly also the Properties H (K2) further, with these through further Probability relationships p (K2) of related technical terms K2 from those in Situation Wk1 differentiate the properties H (K1) and H (K2). Correspondingly from the use of such provided according to the invention extensive communication result Wk2 as a one if necessary the following statement Tp, in turn, presupposes facts Wk1 furthermore advantageous that this is always assessed in a context in which the messages F of previous statements Tp are taken into account appropriately become.

As well as from the embodiments listed below and one of the variety number of text examples adapted to linguistic expression possibilities T appears, is due to the intended use of language symbols Vt formulated independently of context and thus applicable to any contexts and these properties, which are taken into account accordingly and which are characterized by certain assignments one by an assigned grammatical Symbol Gr given, repertoires R (Pr) of possibilities of their grammatical Arrangements Pr, a disjoint class K1 of text sections Ts and one accordingly semantically ambiguous class K1 of semantic references Fp are characterized by probabilities and Probability relationships p (K) more specific terms K as well, Instructions for use, also known as linking instructions or parameters, q for this purpose, advantageously include the following for communication files Partial task can be solved: A text section T = Ts, which, as with natural spoke L usual, for example in a lexical listing standing alone as semantically, if necessary also considered grammatical, ambiguous, lets itself in a statement Tp due to the intended consideration of, if necessary, several contexts to be assumed disjointly, as described by further, also semantically ambiguous sections of text T these Expression Tp as well as, possibly also several disjoint, facts Wk1x, which of encountered situations as well as possibly previous utterances depend on Tp, are given one with one quantifiable level of security, using the ones described below non-linear operations Bb often also clearly, identifiable part a result of the relevant communication act, as a rule also assign clearly preferred facts Wk2, as by People because of their largely intuitive handling of properties of languages L and contexts is expected. The invention offers Use symbols Vt that are both grammatically related to their Linkage options Pr as well as semantically related to them  Technical terms K and the applicable instructions for use q are ambiguous, the advantage that there is also a very large variety through communication extensive facts Wk2 through a limited, the storage and Processing capacity of available devices is customizable, according to the Language definition mentioned above also to be called vocabulary, Repertoire R (Vt) of symbols Vt can be expressed and communicated by means of texts T. becomes. Deviating from that used in previous language systems Lexical words assigned lexical meaning // 12 // The symbols Vt according to the invention include one each as a class of Elements // 12 //, p. 449 class K1 = (K) of terms K in addition, disjoint classes K1 = (q) of Instructions for use q.

For further description of the invention, the following, partly already above Symbolism used: Any set of sizes A is called Repertoire R (A) denotes such a size A as an element A = E1 of this. A repertoire R of sizes A with assigned probabilities p (A) is also called field R (A). Lowercase letters and v, w, x, y and z are reserved for running indices according to Ax = A1, A2 etc., Axy = A11, A12,. . . A21 etc., where indices like in Ax.y = A1.12 can also be separated by points, here with x = 1 and y = 12. The Relationship B a A classifies B as a subset of A. A set of according to the designation B1 o B2 o B3 o. . . to be used alternatively Sizes Bx a A are considered a disjoint repertoire R (Bx) or a disjoint class K1 = A = (Bx) of its elements E1 = Bx. An assembly of components Em = Bx, which also depends on their mutual assignments, is called Complex called Kp = A = BxBx, also noted as Bx, Bx, where Compilations of components Bx, partially or in special cases also all in the form of disordered repertoires as well as fields R (Bx), can also be defined as complexes Kp, just as it is for complexes Wk = KK defined facts Wk is provided. In detail, too Notations K1 (Bx) = (Bx) and Kp (Bx) = BxBx are used. The definition of im As a rule, repertoires or fields R represented by several sizes, Masses K1 or complexes Kp only include the membership of if necessary one size or none at all for these amounts. With nE it will Non-existence, the non-use or also the one defined below Negation of a term B is displayed. The characters +, - and = are in the conventional meaning, multiplications are by B1 × B2, Divisions indicated by B1 / B2; Slashes also serve the mentioned, in each case evident from the context of conditional probabilities p and text examples t. A complete one Repertoire R (Bxy) of one unit each Bxy a Ax of a disjoint class Kl = Ax from a repertoire R (Ax) of such classes Ax becomes a permutation designated. By A1 ne A2 the inequality of the quantities or sizes A1 and A2 are displayed. The meaning of ue is used as further reference symbols  "about the same", g ("greater than"), gg ("much greater than"), k ("smaller"), kk ("much less "), ge (" greater than or equal to "), ke (" less than or equal to ") and nne (" not necessarily the same) used. The positions P of components arranged in series Em in complexes Kp = EmEm are given according to their order Em1 Em2 is also explicitly noted according to P = Em1: Em2, whereby, as for example also with the arrangement of discontinuous text sections Ts in texts T, not it is excluded that other components Emx as well as parts thereof in between. Accordingly, according to the notation P (a), too possible distance range a between Em1 and Em2 is displayed.

As stated above, the method according to the invention provides Mp Communication using, as sets of any one another assigned objects A formed, hereinafter also briefly things mentioned, to carry out subject terms K, which accordingly also as things K = R (A) or as complexes K = Kp with components Em = A. Taking into account the arrangements Pr of the grammatical symbols Gr Use of things K a Fp a Vt in the semantic content F, in particular links with other things K, are according to Instructions for use q a Fp a Vt a Gr made, their mentioned adjustment to the language L in question preferably in a manner such as that provided by the to be carried out on texts T of this language L described below Transformations Qs is determined. Notes q are taken into account accordingly, of which q = d is an average, also noted as K2 d K1 or K2 d1 K1, and q = s the union according to K2 s1 K1 of two things or also Subject complexes K2 and K1 calls, whereby the same indexes indicate that that, as here, q1 and K1 belong to the same semantic clue Fp1. Furthermore, q = m removes a thing K4 from a superset K3 according to K3 m4 K4 and by q = f the replacement of a thing K5 by a negation K6 = nK5 of which according to K5 f6 K6, which according to K5a K9 and K6a K9 is to be defined in more detail by another thing K9. Note q = c looks general the replacement of a technical term K7 or a complex K7 with a thing or a subject complex K8 according to K7 c8 K8. Parameter q = c includes replacing a property H (Kx) 1 such as one Probability p (Kx) 1 due to a different property H (Kx) 2 as a probability p (Kx) 2 of the matter Kx. Farther with q = k according to a notation K9 k K10 by K10 a Fp10 and the related thing K10 a Fp10 the replacement of thing K9 by a complement Set of K9, based on a superset K11 = K9 s K10 with K9 d K10 = 0, required. With q = o, with K14 = K12 o K13, the formation of a thing K14 displayed, the mutually exclusive parts of the sets K12 and K13 belong to. In adaptation to the languages concerned, L may be used further instructions for use q are also used. For example, q = o with K15 = K12 o K13 o. . . also the formation of a disjoint repertoire R = K15 two or more alternative items K12 or K13 etc. are displayed.  

The proposed modification of a pragmatic situation Wk1 corresponds to the resultant semantic content F. Probabilities p (Kx) of things Kx a F in a situation Wk2 will be like this using q = c also noted as Wk2 = Wk1 c F. The variation a content Wk1 according to a one thing K1 = F1 Content F, as for example a head symbol Vt1 defined below with K1 a Fp1 a Vt1 is assigned, is accordingly as Wk2 = Wk1 c Kx with Kx = K1 written down. For example, Wk1 contains one thing in addition to other things Kx K1 with p (K1) = p1 and another thing K2 with p (K2) = p2 = 1-p (K1), and follows from a semantic message T1N1F1 according to F1 for thing K1 Probability p (K1) = p3 with p3 ne p1 results in a situation Wk2, in which not only has K1 a probability p (K1) = p3, but also K2 a probability Probability p (K2) = p4 with p4 = 1-p3. Performs content F 1 For example, if p3 = 1, then because of p (K2) = 0, the matter disappears K2 result in the communication result Wk2.

According to the description given above, q = o defines Terms Kx according to K2 o K1 o. . . as disjoint sets K1, K2 etc. one corresponding to disjoint repertoires R (Kx) are displayed.

By citing a repertoire R (K1 (q)) of one or several classes K1 (q) of linking parameters q to a semantic Note Fp = Fp1 with Fp1 q Vt1 a Gr1 it is taken into account that Gr1 in content N several arrangements Pr1, Pr2 etc. with several other grammatical ones Symbols Gr2, Gr3, etc. can be connected, whereby a thing K1 a Fp1 is to be linked with several other things K2 a Gr2, K3 a Gr3 etc. Through the assignment of a disjoint class K1 = (Fp) from semantic hints Fp, which are not only the things K a Fp, but Distinguish K for the same thing by the instructions for use q a Fp can take into account that with such an assignment Pr1 of a symbol Gr2 to Gr1 correspondingly different connections q of the permutations of Things K1yz a Vt1y a Gr and K2yz a Vt2y a Gr2 are possible, being as below explained using examples, for example (q = d) o (q = s) applies.

Instructions for the method Mp according to the invention are Vr also regulations of users of the method and data derived from it considered. These generally include information about the proportions of the To take over communication and what hints here note. This includes, for example, instructions as an author Partner Pa certain facts Wk through corresponding messages TNF to communicate assigned outgoing texts T as listeners Hr incoming texts T Assign messages TNF and facts Wk2 and these if necessary to use it further, to correct incoming texts T, to translate to use certain languages L, messages F incoming texts assess, confirm, question or comment, questions about  answer, etc. The specifications Vr also include, for example, information about which or which proportions of the existing facts Wk1 the basis is whether and how far incoming texts are assigned to T Issues WW2 are to be accepted as further requirements Vr, whether and to what extent, for example also depending on the resulting probabilities p, The resulting facts Wk2 associated with utterances Tp by an author At take over as well as facts Wk1 one to be released subsequently Expression Tp should be used to determine whether author At and listener Hr from different issues Wk1, the interlocutor Pa assigned issues Wk1 as well as Wk2 to check whether they are plausible or contain errors or untruths, etc., if applicable deny or correct, as well as advice on what Communication results Wk2, for example also depending on the resulting Probabilities p (K) 2 or changes p (K) 2 - p (K) 1 thereof, the Notify users or implement them in actions, etc. Here, the Specifications Vr also stipulates whether and also how existing data D, how for example in the form of electrical, acoustic, optical or mechanical Data accrue, must be considered in the facts Wk.

If necessary, Vr is also required by corresponding specifications, of which assume that communication partner Pa when assigning semantic Contents TNF on outbound utterances, written by them as authors Atp or to incoming texts T received by them as listeners Hr pragmatic Understand terms K, their probabilities p (K) for their assignment too Symbols Vt or other things K, for example in circumstances Wk, of values known as objective in a subjective or incorrect way deviate, such deviations, for example by Requirement of corresponding classes K1 = (Wk1x) of facts Wk1x or also taken into account in accordance with modified data Wp in the communication become.

According to the invention, a probability p (A) is generally used to describe both understood the relative frequency with which an event or object A in a Numerous cases were encountered, as well as the expected value for one too treating individual case. Through p (A), in general, too Probability relationships to other variables B such as conditional Probabilities p (A / B) are displayed. Such dependencies one Probability p (K1) of one thing K = K1 of other things K2, K3 etc. as well as other terms such as symbols Vt etc. are in accordingly explicit notation as conditional 2nd order probabilities O (2) according to p (K1 / K2) or 3rd order O (3) according to p (K1 / K2; K3) or higher order O (n) in complex KK, facts Wk and contents F taken into account, whereby also Influencing factors such as an arrangement Pr through notations such as p (K1 / Vt1; Pr) etc. are displayed. As another, also simplifying as Probabilities p (K1) are designated probability relationships  also by algorithms such as p (K1) = a-p (K2) etc. or from Quantitative approaches such as union, average, membership, complement, etc. resulting references taken into account. According to the definition above, Probabilities p (K1) of first and higher order O (n) in general, too Probabilities p understood that a thing K1 in an utterance Tp is to be expected, so strictly one at a time, to simplify the Description of conditional probability p (K1 / Tp) not specifically noted, p (K1 / K2; Tp) etc. is indicated for the fact that K1, as appropriate derived from a body of previous utterances Tp, in any further statement Tp is expected. Be in an embodiment of the invention these conditional probabilities according to p (K1 / Tps) also on one on the Length of text T standardized, standard utterance Tp = Tps related. According to the Relationship p (A1) × p (A2 / A1) = p (A2) × p (A1 / A2) is assumed to be with specified conditional probabilities p (A2 / A1) also the inverted values p (A1 / A2) are known; the same applies to higher-order values. In others In the case of probability information p, training is also provided, for example, by Expressed standard deviations, uncertainties dp taken into account or provided. Due to the probabilities p (K) the occurrence of the, generally defined in more detail by further data A, Things K including their references in complexes Kp = KK to other things K, as in pragmatic facts Wk, content F as well as complexes Kp other things K described things K occur, set, where According to the above definition, it is not excluded that the simplistic as Complex Kp referred to facts Wk or content F in addition to complexes KK related things K also fields R (K) disordered things K with Probabilities contain p (K).

The invention provides, preferably for the communication facts Wk to use with terms K, their occurrence and mutual references predominantly through data, especially also conditional probabilities p, low order O (n) can be described. In a preferred embodiment are mostly data of the order O (1) and O (2) and only in order O (n) greatly decreasing scope also uses 3rd and higher order data, thus the risk of being disproportionately large with the number of items K. Increase in the data to be provided for performing the method Mp Wp, combined with combinatorial explosions during their processing decision trees, remains avoidable. In an apprenticeship a repertoire R (K) of subject terms K is taken into account, the mutual references described by conditional probabilities p (K) be, whose order O (n) a predetermined, the performance available device adapted value n = n1 does not exceed. This will, consistent with the nonlinear information processing described below Bb, also achieved by conditional probabilities p k ps below Thresholds ps that adapt to the capacity of available computers to be determined are neglected. Furthermore, this is preferred by  Use of terms K defined as orthogonal achieves which no or negligible influence on the connection between have already defined things K. For example, in one thing K2, their existence with a conditional probability p (K2 / K1) of a thing K1 depends, other things K3 etc. defined such that p (K2 / K1; K3) = p (K2 / K1) applies.

The semantic information given by Fp = Fp1 with K1 a Fp1 and q1 a Fp1 linking of things K1, K2 provided in the formation of the contents F etc. according to K2 q1 K1 etc. according to the invention takes into account the the resulting probabilities p (K2 / K1) according to the instructions q = q1 etc. and those in the data Wp and facts Wk Probabilities p (K). In line with the circumstances of the languages L certain links q1 a Fp1 etc., if necessary Factors qr for the change, depending on the type of link resulting values p (K2 / K1), for example to reduce them by certain, factors qr k 1 assigned to the use of the languages L. In a further development, the instructions for use q a Fp a Vt are also the assigned the probabilities p (K / Vt) of the things K a Fp a Vt, which in determining the probabilities to be considered p (F / TN; Wk1) can be used.

The introduced links of a thing Kxyz a Fpxyz a Vtxy a Grx with other such things according to, for example, K211 q111 K111 are under Consideration of the respective indices for simplification also as Fp q Fp, Vt q Vt and Gr q Gr notes, including links according to permutations of the symbols Gr etc. as disjoint classes K1 assigned indications Fp a (Fp) a (Vt) a Gr are understand. Corresponding notations apply to the variations Wk q F from Facts Wk according to disjoint classes K1 = (F) assigned content F.

For the things K of the TNF according to the invention to be changed by messages Issues Wk1 are in addition to data that are to be specified by users, according to Requirements Vr also takes into account other data given by contexts, which at one or more partners Pa from previous TNF messages such as possibly also from the situations corresponding to non-linguistic Data transfers occur, such as the status of Devices communicating according to the invention and further hereby connected devices such as computers and other electrical as well as mechanical devices etc. There, as stated, Authors At and Hörer Hr also assume different issues WW1 , it is not excluded that they are also different Issues Wk2 generate, but according to the Pa Messages F to be formed according to the invention often also reduce the Differences between two resulting facts Wk2 compared to the differences between two initial facts Wk1  is achieved. The according to the content F and the linking parameter q = c Changes to be made to initial issues Wk1 concern in particular the probabilities p (K1) etc. of those present there Things K1 etc. and conclude with probability relationships p (K) higher Order corresponding changes of probabilities p (K2) from with K1 etc. related things with K2 etc. Depending on resulting values becomes, for example, resulting values p (K1) k ps a distance of things K1 from Wk1 below threshold values ps to be determined performed. With corresponding resulting values p (K1) there is also a Added items K2 linked to K1 that are not available in WW1 provided if this results in sufficiently high data according to the data Wp Probabilities p (K2) result. According to this procedure too an addition of new things K2 to a situation called learning Wk1 as well as for the data Wp, whereby thing K2 as a complex Kp of according to facts Wk or data Wp things K3 etc. is formed, which K2 with sufficiently high probabilities Show p (K2 / K3) etc. Any other necessary including grammatical, data for linking existing or possible text sections Ts, symbols Vt, arrangements Pr etc. in a preferred embodiment also using the one described below grammatical communication method M provided.

From the given definition of probabilities p (K) also 1st order of things K as well as possibly through the opinion of a partner Pa as well the expected occurrence in an utterance caused Tp = Tps Probabilities p follows that even in cases where a thing K1 by the message F = K1 of a grammatical message TN is addressed with probability p (K1 / TN) = 1, at which intended modification of a precondition 99999 00070 552 001000280000000200012000285919988800040 0002010015858 00004 99880 in which K1 for example with a probability p (K1) kk 1, into one resulting situation Wk2 this value does not usually increase to p (K1) = 1 is increased, but to one, possibly from a corpus C of texts T also to be determined empirically, value p (K1) k 1. Here in Wk2 for not in Message F addressed, but with K1 through probability relationships, for example according to p (K1) + p (K2) = 1, related things correspond to K2 Changes in the probabilities p (K2) to be carried out, in this Example p (K2) g o follows. In a simplified embodiment, for things K2 linked to K1 by probability relationships p, for which in Wk1 there are low values p (K2) kk 1, regardless of that by message F for K1, value p (K1) achieved these values p (K2) unchanged for less effort Maintain situation WW2. However, this approach closes in particular in the case of objectively verifiable facts Wk does not rule out that in cases where Situation Wk1 as KuKu complex of a manageable number of each other Ku is represented according to strictly assigned matters, in WW2 for K1 also values p (K1) = 1 and for the other things Ku values p (Ku / K1) etc.  according to the given probability relationships p (K), with the above For example, values p (K2) = 1-p (K1) = 0 can also be used. Corresponding the possibility to be considered according to the invention that the message F1 If an utterance Tp1 receives an untruth, method Mp provides as Initial facts Wk11 of another utterance Tp2 in addition to that according to F1 Issue Wk2, modified from Tp1, also disjuncts the original one Issue Wk1 to be used as the communication result of the utterance Tp1 and this, if necessary, as the starting point Wk1 of further statements Tp2 presuppose.

The invention generally provides to use data Wp, their repertoire R (Em) of components Em the communication files to be covered and thus content to be addressed is adapted. In an apprenticeship, especially for those Communication by human authors At an according to procedures Mp acting as a listener Hr is suitable, which these messages according to the invention a further use by others, them as users associated machine devices, for example, data Wp used, in which a comparatively extensive, the language use of Human-repeating repertoire R (Vt) of symbols Vt and corresponding lexical words a limited, the task areas of the assigned User-adapted repertoire R (K) of subject terms K is compared. This repertoire R (K) becomes useful for such a task adapted that the facts occurring in it, according to procedure Mp transmitted statements attributable to Tp and as combinations of such things K formed, semantic contents F and facts Wk are to be described. In an education of such matters Wk also by according Specifications Vr clearly coded signals, for example electrical or represented mechanical type, which may also be suitable by Comments Tp assigned to the relevant subject area Trigger user operations. For example, for the Communication with those who act as listeners Hr according to the method Mp Devices which are assigned to users as electronic computers, which in Usually by means of clearly coded commands Programming languages are put into operation by such commands, if necessary also the associated electrical ones, if necessary also Objective electromechanical signal sequences, arithmetic operations as Facts Wk of a corresponding subject area considered. For as a user The machine tools used accordingly become a part of their range of services adapted repertoire for example electromechanically induced commands as Subject area determined. In a preferred embodiment, the listeners are Mr. Acting devices according to the invention by appropriate specifications Vr caused by not being with sufficient certainty and uniqueness assignable facts Wk the author At on the respective, for example on unsuitable choice of words or factual ambiguities or errors draw attention to the problems to be attributed. This limitation of one  Repertoire R (K) to be considered by a machine listener Hr Subject terms K to those required for the subject area to be covered Scope leads to advantageous savings in those to be made available for this Storage and processing capacities.

As data Wp, as described above for the invention Communication methods Mp to be considered are general Compilations of the to form the message complexes Kp = TNF using components Em with those to be taken into account here Properties H (Em) used. These properties include in particular also, for example also expressed by probability relationships, References to other components Em as well as to addresses used as addresses existing terms which, in addition to components Em, for example as data D Include present features H (T) of texts T. Both Arrangements of the components Em are also carried out, as explained below Ordering principles such as sequences or complex-forming ones Connections, such as those that apply to technical terms K, are taken into account.

The probabilities p (A) of Terms A are also expressed by information j = log (1 / p (A)), in which log stands for the logarithm to base 2 and j the amount of information in bits indicates that, to the safe knowledge, with p = 1, that A is present, where conditional probabilities p through corresponding conditional information j are displayed.

The consideration of the probabilities p (TNF) provided for method Mp of the semantic messages TNF to be formed as well as p (Em) of the corresponding ones using components Em such as technical terms K and symbols Vt and Gr, which may also be present as a result of further components Em as well In a preferred embodiment, data D are conditional using of the known Bayesian approach, referred to here as Ba, as it is, for example was explained in // 9 //. This is generally considered to be certain Assuming data D which, for example, given specifications Vr, texts T like also include features H (T) and complexes Kp thereof and in general ambiguous an assignment of language content such as symbols Vt and Gr as well complexes formed therefrom such as N, F, NF as well as complexes Kp thereof allow. Components Em, which are generally represented by data D including other components used as addresses Ad Em as well as text features H (T) ambiguous, i.e. with probabilities p k 1 are displayed as disjoint repertoires R (Em), for complexes Kp = Ks one according to disjoint repertoire R (Ks), which each by definition one of these components Em, optionally also in addition further components Em previously assigned to this complex Kp contain further repertoires R (Em) and referred to as competitive hypotheses Ks become. The required data D are also referred to as evidence D.  

The same applies to a grammatical message available as evidence D. TN corresponding to the addresses Ad indicated by their symbols Vt a Gr disjoint repertoires R (Fp) of semantic clues Fp a Vt a TN both this semantic content Fx = F1, F2,. . . Fm like also the disjunctions that arise as intermediate results in their construction Complex Kp = KK of technical terms K as competitive hypotheses Ks here by means of evidence D in a complex formation step a number m of Competitive hypotheses Ksx = Ks1, Ks2,. . . Ksm with sufficiently high Probabilities p (Ksx / D) are displayed, then one applies to Ks1 according to Ba Probability p (Ks1 / D) = p (Ks1) × p (D / Ks1) / Ne, where the denominator Ne is for the probability sum Ne = (p (Ks1) × p (D / Ks1) +... + p (Ksm) × p (D / Ksm)) stands. The probabilities p of the other hypotheses Ksx become by corresponding permutations of the counter according to p (Ks2) × p (D / Ks2) etc. displayed.

In the application of the below provided in a preferred embodiment explained grammatical communication method M is next to a Text T also simplifies the disjoint repertoire R (N) according to M associated with this text T with probabilities p (N / T) Messages N as well as the corresponding repertoire R (S) of as complexes Kp = VtVt sentences S a N formed by symbols Vt and assigned to the contents N as Evidence D used. In the frequently encountered case of grammatically clear texts T applies according to the assigned content N as well as the corresponding one grammatical message TN as evidence D. According to the invention shows here Symbol Vt a Gr a N taking into account its arrangement Pr as address Ad each have a disjoint class (Fp) a Vt of semantic clues Fp whose Technical terms K a Fp a correspondingly disjoint repertoire R (Em) of each for a competitive hypothesis Ksx = Fx components to be considered Em form. In the case of grammatically ambiguous texts T corresponding to the additionally, symbols Vt a Gr a N to be taken into account accordingly Several disjoint repertoires R (Em) are given. Examples of that application of the approach Ba according to the invention follow below.

In a part of the process particularly covering the role of the author At Mp suitable training is also given as evidence D by one Text T a TNF to be transmitted facts Wk2, whereby accordingly Probabilities p (T / Wk2; Wk1) to be formed according to approach Ba for this if appropriate, disjointly suitable texts T are taken into account.

Corresponding to the corpuses C of texts T for the languages L concerned Frequencies p (Em) of components Em determined for the approach in Ba Probabilities to be used p (Em) also uncertainties dp took into account which are generally set higher, the higher the respectively smaller of the values p (Em) or 1-p (Em) for this component Em fails.  

In a preferred embodiment, the invention provides for the Determination of probabilities p (Kp) of the to be formed Apply complex Kp = TNF nonlinear processing operations Bb. Here, taking into account their probabilities, p (Em) links to be made from disjunct classes K1, Components Em a K1 to complexes Kp = TNF the probabilities p (Em) of the components Em as well as p (Kp) of the competition hypotheses Kp = Ks treated complexes Kp nonlinear depending on the amount of each changed existing or determined values p (Em) and p (Kp), taking into account the disjoint repertoires R (Em) and R (Ks) applicable probability sums Ca probabilities p k pr below fixed reference values pc, possibly also to p = 0, and Values p ge pr, possibly also to p = 1, are increased. Process part Bb is accordingly also called nonlinear probability, information and generally also called data processing Bb. In a preferred one Training is determined using the Bayesian Ba approach Probability values p (Ks) used as a basis.

In an advantageously simple embodiment Bbd, process part Bb provides Reference values pr = ps in relation to certain probability values p (Em) and p (Ks) of certain components Em and complexes Ks of the respective Repertoires R (Em) and R (Ks) and components Em and Complex Ks with probabilities p k ps for the concerned like if necessary also several further complex formation stages such as if necessary also to be neglected for good. Process part Bb relates in particular also on technical terms K as components Em and contents F based on them Complex Kp = Ks, with the components Em a R (Em) can be called competitive hypotheses Ks.

In the case of non-linear data processing Bb, it is generally taken into account that for the given disjoint repertoire R (Ksx) of competitive hypotheses Ksx each a probability sum Ca = p (Ks1) + p (Ks2) +. . . applies, where in As a rule, Ca = 1 must be used, but other cases with Ca k 1 also apply in such cases where a repertoire R (Ksx) is not possible through any of the proposed hypotheses Ksx may be represented as a complex Kp or thereof it can be assumed that in addition to the hypotheses Ksx a R (Ksx) Repertoire R still further, but not explicitly known hypotheses Ks can belong.

In the configuration Bbd of Bb, R (Ks1y) of hypotheses is used for a repertoire Ks1y set a threshold ps1 = b1 × p (Ks11) with b1 k1, where Ks11 is the Hypothesis with the highest probability p (Ks11) of this repertoire R represents, and all hypotheses Ks1y with probabilities p (Ks1y) k ps1 for certain advanced complex formation steps or also  finally neglected. In a preferred embodiment of this, b1 becomes all the more chosen higher, the lower the available working capacity or working time of the devices available. As a result of this only those that save effort, but at least one competitive hypothesis Ks1y is taken into account, for whose occurrence one defined as sufficiently low Information set j k js1, with js1 = log (1 / ps1), is missing.

In a preferred embodiment, here with Ksxy = Emxy, because of one low, possibly also by Kp already inserted in a complex other components caused by Emxy, assignment probability p (Em1y) k ps1 for the complex formation Kp initially neglected components Em1y a R (Em1y) saved and then again for the Complex formation Kp is taken into account if there are further Complex formation steps with components of further repertoires R (Em2y), R (Em3y) etc. in relation to an absolutely defined threshold psa1 k ps1 as assignment probabilities considered too low result in p (Em2y) k psa1 etc. In a further embodiment, data Wp are used in which Probability references p (Em2 / Em1) below the threshold values to be determined ps1, ps2, etc. are neglected, in an education of which also Data blocks Wpx a Wp with gradually reduced threshold values ps and correspondingly increasing amounts of data are used and in this order, more extensive blocks Wp2 to form complexes Kp can be used if Wp1 no satisfactory solutions were found.

According to a further embodiment Bbr of non-linear data processing Bb, which is particularly important when complex terms K = K1y Components Em for semantic content F is used for a disjoint repertoire R (K1y) a threshold ps3 = b3 × p (K11) with b3 g 1 and (1-ps3) kk 1 is determined, ps3 being chosen such that at least one thing K11 aa R (K1y) with a probability ps1 k p (K11) k ps3, and there will be all things K1y a R (K1y) with p (K1y) g ps3, again for certain Complex levels or final, neglected. As a result, the Writing semantic content F a TNF, largely redundant Complex formations, especially of things K, contain, by an author At as well as the interpretation of incoming texts T as such by an author At unlikely redundant messages TNF by a listener Mr. avoided in favor of more information-rich news TNF. The according to the Training Bbd and Bbr undertaken non-linear data processing Bb sees so generally ago, both TNF messages with very unlikely Complex formations, the availability of which is correspondingly high amounts of information j g js1 are missing, as well as those with trivial complex formations, the extremely minor Provide information amounts j k js3, possibly in favor of possible Neglecting news TNF whose offer of information j in one middle range js3 g j g js1.  

The possibilities of non-linear methods provided for method Mp Data processing processes Bb are initially approximated using some treated, as example 1) summarized text examples T from the German Written language explains for which initially clear assignments of grammatical messages TN with correspondingly clear texts T = Ts assigned symbols Vt = Ts or complexes VtVt = TsTs thereof be accepted. The case of incoming texts T is used as an example treated, whereby, as stated above, the assigned items K and Subject complexes KK, as far as possible, on the opinion of as Communication partner Pa acting author At are obtained. Here represent an incoming text T1 = /Black./ as utterance Tp = T1 = Ts11 a grammatical message T1N1, the grammatical content N1 of which is a single one, symbol Vt11 a Gr1 a N1 assigned to a grammatical symbol Gr1 as Evidence D includes, in addition to other, neglected, options 3 semantic hints Fp11z are each associated with a thing K11z as they are here as well as in the following examples by using quotation marks set colloquial expressions are approximated. For this K111 applies: "Author At expresses himself about black colored objects, which generally also includes living beings, people, etc. ", K112: "... about people who have a conservative policy" and K113: "... about ... Negro ". According to a prerequisite situation Wk1 = Wk11, with numerical values c kk 1, a ue 1 and c k b k a, probabilities p (K11z) used as a basis for the reference to the occurrence in a Expression Tp comparatively low absolute values corresponding to p (K111) ue p (K112) ue p (K113) = c can be specified. According to procedure Bbd becomes a Threshold ps 1 k c set, with text T1 with the 3 messages F11z = K11z accordingly remains semantically ambiguous. Is against, for example because previous messages, a context with a situation Wk1 = Wk12 assuming that p (K111) = b and p (K112) = p (K113) = c hold, and will Correspondingly, a threshold value c k ps1 k b is set, followed by a practical one message F111 clearly related to the "color property" K111. Here was based on a grammatical symbol Gr1, whose repertoire R (Pr) by Possibilities of its arrangements Pr relative to other symbols Vt a Gr further possible symbols Gr those of certain nouns in the nominative plural correspond. Note that text T1 may also have a grammatical message T1N2 with a grammatical content N2 = Vt41 with Vt41 a Gr4 a N2 is compatible, in which the grammatical symbol determined Gr4 Accusative forms of nouns symbolized in the plural, which in Example 1) not be considered further. In line with those listed below Transformations Qs as well as Q are used in all text examples of the example 1) the symbols Vt1y and Gr1 are used as head Vtk1y and Grk1. Corresponding to one of the configurations explained below using further examples the semantic references Fp11z a Vt11 were the same here, from in numerical symbolized in this example also by the letter c  Value to be distinguished, use note q11z = c defined above assigned. According to the notation expressed by K7 c8 K8 above the replacement of K7 by K8 given by c8 takes place in this example characterized in that the Wk11 or Wk12 in the presumed facts as Property H (K11z) 1 given a-priori probability p (K11z) = p (K11z) 1 of a technical term K11z by the semantic content F resulting probability p (K11z) 2 as a new property H (K11z) 2 of otherwise unchanged, for example given as a set of terms A, Technical term K11z is replaced.

Example 1) is expanded by a grammatical symbol Gr2 with 2 symbols Vt21a Gr2 and Vt22 a Gr2, with which the texts T2 = / blacks are red./ and T3 = / blacks are black./ are considered as further expressions Tp, whereby , again clearly, Ts21 = Vt21 = / are red / and Ts22 = Vt22 = / are black / is set. Here there is a possibility of arranging Pr from Gr2 to Gr1, which is determined by a rank Ra with a dependency Gr2 = An (Gr1) explained below and a position P, in which text Ts11 = Vt11 is used discontinuously, the text parts of which T = / black / and the point T = /./ are each separated by a text Ts2y. It should be noted that the texts Ts2y and Vt2y each begin with a space //, which indicates a spacing from the preceding word, as it also corresponds to the text transformations Qs to be taken into account according to an embodiment of the invention. A further character T, which functions as an operator o and which is to be used as an operator o to capitalize further the symbol Vt11, was neglected in the formulation of Ts11, as was a space leading the text Ts11.

Each symbol Vt2y is assigned 2 semantic references Fp2yz, which the same link hint q2yz = d for averaging the respective thing K2yz a Fp2yz with one thing K11z a Fp11z a Vt11 assigned. With Vt21 = / are red / 2 things become about in the sense of K211: "the objects indicated by subject Gr1, again according to author At, are red colored "and K212:". . . are politically socialist oriented "indicated by Vt22 = / are black / corresponding to things K221: "... Black colored" and K222: "... politically conservative". According to facts Wk11 is conditional Probabilities p (K221 / K111) = p (K222 / K112) = a and p (K211 / K111) = p (K212 / K112) = c to go out, while for the further assignments medium Values p (K2yz / K11z) ne b apply, with an index like here z at different Positions can also represent different numerical values. Will be here a threshold value c k ps1 k b is determined according to Bbd, so for text T2 the Subject complexes F = K111K211 roughly in the sense of an author's opinion "Black colored objects etc. are red colored" and F = K112K212 roughly according to "Conservatives are socialist oriented" negligible. In contrast, remain the considered contextual references for text T2 content F in the sense of a Author's opinion K113K211: "Negroes are socialist oriented" and for T3 in  Meaning of K111K222: "are black people, for example dressed in this way conservative "and K113K222:" Negroes are conservative "even with low values b kk 1 cannot be excluded.

The configuration Bbr of the non-linear data processing Bb provides as Avoid unnecessarily redundant content formations F = KK by Assignments of Kb to Ka with probabilities p (Kb / Ka) g ps3, with (1 - ps3) kk 1, to be avoided. Accordingly, for example 1) Threshold b k ps3 k a predefined, whereby the quantitatively by p (K221 / K111) = p (K222 / K112) = a certain complex formations KK are due to a g ps3 become negligible with little information. In favor of the above According to Bbr, content formations are also those according to Bb for text T3 possible contents K111K211 in the sense of an author statement "black colored Objects etc. are black colored "and K112K222:" Conservatives are conservative ".

According to the possibility considered according to the invention, also disjoint Allowing different initial issues Wk1x is shown in Example 1) in addition to the considered situation Wk11, the situation Wk12 provided that instead of p (K111) = c a value p (K111) = b applies, where, according to the above, the other probabilities p (K) here unchanged from Wk11 and the introduction of a Threshold value ps1 g c is dispensed with. Using the above Bayesian approach Ba results according to Bb that for texts T2 and T3 the same messages F = KK are sufficiently probable as for WW11; however, context Wk12 causes text T3 to have content K111K222 p (F / TN) is more likely to be assigned than a content K113K222.

Example 1) also explains the case of outgoing texts T, the above Things Kxyz then according to the presupposed opinion of the as Communication partner of the functioning handset Pa = Hr are to be obtained.

In a preferred embodiment, the method Mp according to the invention is however without restriction of generality, on the basis of an apprenticeship of this also using a grammatical Communication method M carried out, which in patent application P.. . is described. Method M sees communication with grammatical Messages TN before, each consisting of a text T and this one to be assigned, there also designated, grammatical content N, which is described in detail as complex Kp with structure J from below grammatical symbols Gr is formed as components Em. In a preferred training symbols Gr are used for this purpose, which in P.. . described transformations Q to be applied to texts T suffice, whose Results by, with the participation of experts in the languages concerned L Observations of certain invariances I to be carried out can be demonstrated. This  Transformations Q are with the, in one of the mentioned configurations Method Mp based on the invention, explained below Transformations Qs are compatible, but differ from those for known grammars // 12 // included Transformation grammars // 12 // characteristic, also as meaningless // 12 //, p. 800 designated transformations, so that the Symbols Gr with their properties H (Gr) from the previously used terms how to distinguish phrases // 12 //. According to method M, a text T a Tp of an utterance Tp one and possibly also several such messages TN, usually disjoint, assigned, which according to the The method Mp according to the invention furthermore each by one or disjoint several semantic contents F for corresponding semantic messages TNF are to be completed. Method M uses the grammatical symbols Gr Symbols Vt of method Mp corresponding, there also as objective words or in short words Vt, symbols Vt and the symbols Gr of Mp corresponding class-forming, also referred to there as units G. Superations of objective words Vt to disjoint classes G = (Vt) used. Here, a word Vt in texts T is replaced by the same disjoint class Vt = (Ts) of sections of text represents like the corresponding symbol Vt of Method Mp. The units G = (Vt) are each a repertoire R (Pr) of Possibilities of, also in M so called, arrangements Pr of them associated words Vt a G in grammatical messages TN and accordingly assigned to semantic messages TNF, which are determined by probabilities p are defined in more detail for the words Vt a G of a unit G within predefined tolerances dp match. These arrangements Pr are as in move Mp, through the positions P of the text sections Ts a Vt and a rank Ra given and are described by valences order, which also mentioned Transformations Q are sufficient and accordingly different from those for conventional ones Language procedures and these stored grammars proposed and as valences not considered to be reliable // 12 //. According to method M grammatical symbols Gr are also used to save data as well as complex-forming superations of such symbols Gr as words Vt and Units G as well as certain parts thereof can be created. By a Valence Um1.1 of a unit G1 is generally determined that this unit G1 in a grammatical content N with a maximum of one, in an arrangement Pr assigned unit G2a (G) 1 from a disjunct class (G) 1 can, this valence Um1.1 is considered saturated after such an assignment. Here, the mutual rank Ra of two over their respective valences Um units G assigned to one another are determined by the probabilities p, with to which a saturation of these valences is to be expected. By definition the arrangements Pr include both the possibilities of, possibly also sole, occurrence of symbols Vt in messages TN regardless of other symbols Vt as well as theirs by referring to further such symbols Vt related occurrence. They will also be simplified  Arrangements of the symbols Vt a G including the text sections Ts a Vt referred to as Pr.

The proposed, by the underlying definitions of the symbols Vt a G made possible description of their arrangement possibilities Pr in messages TN and TNF by valences around the units G has the advantageous consequence that unlike lexical words, the multitude of possible syntactic references between the texts Ts a Vt objective words Vt data saving by a lot lower number of superizations G = (Vt) of these words Vt with properties H (G) becomes writable, where a valence Um1.1 of a unit G1 = (Vt1y) the references of the words Vt1y a G1 to words Vtxy a Gx are expressed that in addition to a further superation K1 = (Gx) of superizations Gx = (Vtxy) are summarized, and furthermore the reference data largely on combinatorial explosions avoiding data of order O (2) limited stay.

The data to be observed in method M, including data on complex formation of the messages Kp = TN to be used with their components Em Properties H (Em) are summarized under the name Ws and in this embodiment of the invention, the data Wp of the invention Procedure Mp assigned. Unless otherwise stated, the rest is done Description of the method according to the invention Mp without limitation General validity based on the grammatical method M and the symbols Gs = Vt and Gs = defined for this in accordance with the Mp method G and its properties H (Gs).

The contents N of the texts T to be formed according to the grammatical method M. contain complexes Kp = GG of units G = present in arrangements Pr (Vt) and correspondingly assigned complexes Kp = VtVt of according to present method Mp objective words Vt to be used as symbols Vt, however, close the linkage to be made according to method Mp semantic references Fp a Vt to be assigned to these symbols Vt semantic content F and the formation of appropriate pragmatic Facts wk not a. As in method Mp, as mentioned above, in the Instructions for use q a Fp arrangements to be observed Pr are in this Training to be found in the news TN according to method M. Arrangements Pr of the units G and corresponding words Vt are taken into account, such as according to the valences around these units G in the news TN surrender. 2 is assigned to one another in an arrangement Pr Units G1 and G2 the higher-ranking unit G1 = Ar (G2) as carrier Ar of G2, the lower-ranking unit G2 = An (G1) referred to as the trailer An of G1. Accordingly, Vtxy a Gx and text sections apply to the associated words Tsxyz a Vtxy the relationships Vt11 = Ar (Vt21) and Vt21 = An (Vt11) and Ts111 = Ar (Ts211) and Ts211 = An (Ts111) etc. A chain hierarchically one another Subordinate units with G1 = Ar (G2), G2 = Ar (G3) etc. form a branch  As (G1, G2, G2...), With the highest-ranking unit G1 of a content N also as a header G1 = Gk1 of this content N as well as the branch As to which it belongs. In such a branch As (G1, G2, G3 ...), G2 is also referred to as G1 with respect to G1 hierarchically closer or closer to G3, in relation to G1 and G3 also ranked in between. Such are simplified Rankings also as G1 g G2 g G3 as well as Vt1y g Vt2y g Vt3y and corresponding to G3 k G2 k G1, in detail also G1 g G3 etc. noted. On Technical term Kxyu a Fpxyu a Vtxy a Gx of a symbol Vtxy a Gx, which one in is assigned to a content N unit Gx = Gkx arranged as head Gk also, as explained below, as the head of the message in question TNF assigned semantic content F, as it is also by the transformations Qs can be determined. Using the grammatical method M take place with these transformations Qs, such as shown below, each on a text T1 = Tf1, with T1 a TN, to be carried out Shortening to a text T2 = Tf2 by one text section each Ts a Vt11 a G1 of a unit G1 a N a TN, which according to G1 = An (Gx) and G1 ne Ar (Gx) as Trailer An, but not as the carrier Ar of another unit Gx a N.

In accordance with the measures provided for method M, Design based on that a text T = Tp an incoming as well of an outgoing utterance Tp to be formed as part of each grammatical Message TN or in the case of ambiguity of a disjoint class K1 = (TN) of these, such as a complex TN, TN, if necessary, for serial languages that is, a sequence, one of these or, in the case of ambiguity, a disjoint Class K1 = (TN, TN) of such consequences is present or when intended of the method M is also generated so that text Tp is divided accordingly Form also as complex Kp = VtVt of according to their positions P arranged symbols Vt = Vtxy as well as a complex Kp = TsTs of these assigned text sections Tsxyz a Vtxy or in the event of ambiguity as a corresponding disjoint class K1 = (VtVt) and (TsTs) thereof. According to Method M is a permutation VtVt of the symbols Vtxy a Gx all in a grammatical content N according to their valences order units Gx also referred to as the sentence VtVt = S, the permutation of the text sections Ts a Vt all symbols Vtxy a S text T a S usually as one with regard to its content F correct text T = Tf is assessed if the Deviations of the text sections Ts from their standard texts Tsa Do not exceed tolerances dT. Accordingly, a text T = Tp also lies as a sentence S or sentence sequence SS or in the case of ambiguity as a disjoint class K1 = (S) or (SS) of these. Note that according to the definitions of correct Texts Tf and the transformations Q as well as Qs a sentence S is not necessary one at least as a combination of two as subject and predicate or as Nominal phrase and verb phrase of designated parts of the sentence formed natural sentence such as / plums are green / must represent, but that for example, also predicate-free utterances such as / Green Plums / or / No! / Apply as T = Tf = Tp and thus represent sentences S. Furthermore closes  Method M as Mp the use of discontinuously arranged Text sections Ts a Vt in texts T and thus of discontinuous words Vt in sentences S as used above in Example 1).

For the translations to be performed according to the task, Mp sees process in one advantageous embodiment, the use of further grammatical Method M introduced symbols referred to as vocabulary Vqq which each have a complex, also called vocabulary, Kp = VtVt = Vq of objective words Vt of a source language L, a disjoint class K1 = (Vq ') of vocabulary Vq' is assigned, each as a complex Kp = Vt'Vt '= Vq 'are formed from words Vq' of a target language L '. With these words Vqq is translated in that a given text T corresponds to the language L. a sequence of grammatical messages TN with assigned sentences S = VtVt is added and the words Vt of each sentence S completely to vocabulary Vq be summarized and this as a complex S = Kp = VqVq = VqqVqq is present. The words Vt 'a permutation of vocabulary Vq' a Vqq der Target language L 'of this sentence S become a grammatical content N' a T'N ' a grammatical message T'N 'from L' put together, the text T 'a T'N' the translation of the text T a TN of the language L applies, according to the vocabulary Vqq also has a disjoint class K1 = (T'N ') of such messages T'N' according to different translation texts T 'may be incurred. In addition to the measures provided for in grammatical procedure M take place in Mp inventive method, as set out below, an addition to Messages TN and T'N 'to semantic messages TNF and T'N'F' and the Selection of a message T'N'F 'with translation text T' a T'N'F ', whose semantic content F 'a F'N'F' best with the given content F a TNF matches.

According to a preferred embodiment provided for method M, the symbols Vt are also formulated as complexes Vt = ZZ of further, also referred to as characters Z, components Em = Z, which in texts T each have a disjoint class K1 = Z = (Tr) of text sections Tr be represented. According to an embodiment of this, these text sections Tr a Z each satisfy certain text features H (T) as well as repertoires R (H (T)) as well as complexes Kp thereof which lie within tolerances dT to be defined. It is preferred to also refer to ordered repertoires R (Z) of characters Z, also referred to as alphabets, as well as correspondingly ordered repertoires R (Vt), R (G) and R (Ad) of the symbols and addresses Ad to be used for their arrangement in the data Ws used. In a further embodiment, to represent the words Vt in the data Ws and messages TNF, in addition to the languages L in question, adapted characters Z such as letters, syllable characters, punctuation marks and word spacings or, in oral languages L, corresponding prosodic elements such as stresses, pitch changes or pauses are also determined, Operators o defined below in detail are used which bring about certain changes to further characters Z and their text sections Tr a Z assigned to them according to their positions P in messages TN, with the result that, depending on the text context, the words Vt in messages TN also occur through Character complexes ZZ2 are represented that differ from the character complexes ZZ1 noted in the data Wp.

The in process M and correspondingly Mp as components Em of the messages TN and TNF symbols to be used are commonly referred to as Gs. Of these, text sections T are to be formed as disjoint classes K1 = (T) Symbols Gs such as Vt and Z also text symbols Gt, certain class-forming Superizations K1 = (Gt) like the units G also grammatical symbols Gr and components to be used for the formation of semantic contents F also called semantic symbols Gp. As further components will be Em according to the possibilities of disjoint arrangements, including certain ones Subwords denoted parts vt of words Vt and disjoint from them Classes super subunits g a G used.

Of known, dealing with natural languages L, procedures and these underlying grammars and semantics, as also quoted above , the present differs in connection with the grammatical Method M also the design of the semantic method Mp by using objective lenses to be used as symbols Gs Words Vt is performed, many of which are for the written languages specified lexical words of the respective languages L differ, i.e. in Texts T, for example, by non-contiguous or contiguous sections of text T a Vt are represented or in Written languages L also several lexical words or only parts of them include and possibly include spaces and punctuation marks. As As a result, they become, corresponding to the transformations Qs and Q. Superations Gr to be used according to the invention, such as units G as well certain classes K1 and complexes Kp can be summarized, which, as above for the valences order shown in previously known language procedures not or not sufficiently taken into account, in particular also quantitatively properties to be determined H (G), each for a large number such symbols Vt and text sections Ts a Vt that can be re-assigned in each case apply so that the symbols to be used according to the invention Gs, Gr, such possibly also Gt while avoiding combinatorial explosions by only slightly disproportionate to the number of objective words considered Vt and thus the lexical words to be found in it growing Amounts of data become writable.

In a less advantageous embodiment of the invention, in which a Application of the grammatical method M is dispensed with and texts T is assumed, for example, as a sequence of lexical words with this there are more conventional grammatical terms the method Mp envisages using data Wp in which the method Mp to be used, preferably corresponding to the transformations Qs  Symbols Vt assigned such lexical words and word combinations in a first step of Mp according to this data Wp these texts T are shown completely as corresponding complexes of symbols Vt become.

The proposed adaptation of the symbols Vt and Gr and their properties H (Vt) and H (Gr) to the relevant natural Languages L is achieved in a preferred embodiment in that symbols Vt can be used, which is given in texts T by a disjoint class K1 = Vt = (Ts) are represented by texts Ts that follow the above, below described transformations Qs suffice, the results with participation by experts, by definition experts in the languages concerned L like also the semantic and pragmatic content it addresses, Universally observable on texts T in any language L and usually also with the results of the grammatical communication method M underlying transformations Q are compatible. Of the conventional ones Transformation grammars ((12)) proposed transformations the transformations Qs as well as Q differ with respect to planned measures, observable properties as well as results.

According to the transformations Qs, the following applies: For one in texts T by one Text section Ts11z a Vt11 displayed symbol Vt11 with assigned disjunct Mass K1 = (Fp11u) of semantic hints Fp11u a Vt11 exist or several texts T, which can also be used as utterances T = Tp, with the following Characteristic: A text T = T1, written by experts regarding one of them assigned semantic content F = F1 as correct text T1 = Tf1 is shortened by the shortest possible, not necessary connected text section T11, if T11 = Ts11z and if not, as in subsequently described case, T1 = Tsk applies, in a text T2 with an assigned, content F2 modified from F1, which also relates to F2 as correct text T2 = Tf2 is assessed, with each of these texts T2 Linking the respective content F2 according to one of the semantic references Fp11u a Vt11 with an associated term K11u a Fp11u in accordance with a usage note q11u a also assigned to this Fp11u gives the content F1 in a sufficient approximation. The Transformations Qs therefore require changes to the proposed changes Texts T around T11 compliance with a given by T1 = Tf1 and T2 = Tf2 Invariance I on such text pairs T1 and T2, which are indicated by experts is, the statistical significance and thus objectivity of the achieved Results by observing a sufficiently high number of experts any observer of these reactions can be determined.

This transformation Qs is a necessary condition for the fact that the Symbol Vt11 with subordinate rank R one by another text part Ts' a T2 of the text T2 is assigned to the symbol Vt ', as well as being sufficient  Condition for the fact that none is indicated by a further text part Ts "a T1 further symbol Vt "is present, which the symbol Vt11 with a parent rank Ra is assigned. For the disjoint class K1 = (Ts11z) the Text sections Ts11z a Vt11 of the symbol Vt11 apply a correspondingly disjoint Class K1 = (Qs) of transformations Qs, in which by using the the same semantic note Fp11u a Vt11 the same content F1 is restored.

Depending on the natural languages L, a symbol Vt11 obeys the Transformation Qs even when applied to other different ones texts marked accordingly different semantic contents F1 = Tf1, which is obtained by removing a text section T11z a Vt11 in texts T2 = Tf2 with correspondingly different contents F2 be modified, each with different semantic references Fp11u a disjoint class (Fp11u) a Vt11 to restore the respective Initial contents F1 apply, which are both with regard to the Technical term K11u a Fp11u as well as the usage note q11u a Fp11u can distinguish from each other. Depending on the L for the language in question Achievable result is the class (Fp11u) a Vt11 of the semantic clues Fp11u of this symbol also as a disjoint class K1 of the permutations of this Symbol Vt11 assigned disjoint class K1 = (K11v) of terms K11v and also assigned disjoint class K1 = (q11w) of Instructions for use q11w noted. In another representation, the Things K11u and Vt11 also each have a disjoint class K1 = (q11uw) from Assignments q11uw assigned, resulting in a corresponding for Vt11 Permutation of semantic clues gives Fp11uv, in which Fp1111 by K111 and q1111, Fp1112 by K111 and q1112 etc., furthermore Fp1121 by K112 and q1111 etc. is determined.

Text sections Ts a Vt of symbols Vt, for which texts T = T1 exist only are constructed from this text according to T1 = Ts and when using Qs are reduced accordingly to T2 = 0 with content F2 = 0 that deviates from F1, are also used as text sections Ts = (Tsk a Vtk) of symbols Vt = Vtk referred to, which are also called head Vtk, as in method M, and also for this, invariance I is defined as fulfilled. By definition they are Changes from texts T1 to texts Tsk also in the case of a resulting one Text T2 = 0 with corresponding changes to the presumed facts Wk1 connected in facts Wk2.

According to the transformations Qs is from 2 in a grammatical message T1N1 in an arrangement Pr symbols Vt11 and Vt21 assigned to one another gave the symbol Vt21 a higher rank Ra than Vt11 if Vt21 did not can be generated by applying the transformations Qs to the text T1 a T1N1, but probably by applying the Qs to a text T1 'a T1'N1', in which by, possibly also multiple, application of the Qs to T1 or  texts T ", T '' 'etc. formed therefrom, the text section Ts11z a Vt11 and thus Vt11 also no longer exists. Vt11 is accordingly considered the symbol Vt21 in T1N1 ranks lower or assigned depending on Vt21. When using the grammatical method M, in which the semantic symbols Gp symbols Vt also represent objective words Vt a G, such rankings Ra also correspond to the rankings as they stand for the symbols G formed as classes K1 = G = (Vt) of symbols Vt from the M underlying transformations Q result in such texts T1 a T1N1, where with Vt11 a G1 and Vt21 a G2 according to G1 = An (G2) and Vt11 = An (Vt21) the Symbols G1 and Vt11 as trailers An from G2 and Vt21, respectively to be expelled. Accordingly, with G2 = Ar (G1) and Vt21 = Ar (Vt11) Unit G2 or symbol Vt21 as carrier Ar. Based on this the notes and technical terms assigned to the symbol Vt11 also become as trailer An according to Fp11u = An (Fp21u) and K11u = An (K21u) and according to Fp21 = Ar (Fp11) etc. referred to as carrier Ar, one of the Vtk a Gk assigned technical term K a Fp a Vtk accordingly also as head Kk a F des Content F a TNF acts.

From the transformations Qs continues to follow, also in line again with the transformations Q that that by the removed text part T11 represented as trailer Vt11 = An (Vt21) or as head Vtk11 in one grammatical message T1N1 present symbol Vt11 not at the same time as a carrier Vt11 = Ar (Vt31) of another, in the text T1 by a text section Ts31z a Vt31 represented symbol Vt31 acts.

In a preferred embodiment, only such texts T11 and T1 are passed through a semantic note Fp111 a Vt11 marked symbol Vt1 a G1 assigned to a separate unit G1, for which at least 2 different ones Texts T1 = T1a and T1b with different contents F1 = F1a and F1b exist in which the carrier unit G2 = Ar (G1) by different symbols Vt2a and Vt2b are represented, the contents F2 = F2a and F2b of Texts resulting from text section T11 shortening Ts = T2a and T2b by combining it with the same note Fp111 a Vt11 again in their Output contents F1a and F1b can be converted back. If, however, in this Language L only one symbol Vt21 a G2 exists, which in texts T1 symbol Vt11 as Tag Vt11 = An (Vt21) can be assigned, T1 is preferably assigned Symbol complex Vt21Vt11 as idiom Vtxy a Gx of its own unit Gx assigned, depending on the results of the further symbols Vt a G of the texts T1 etc. transformations Qs to be applied, not excluded becomes that Gx = G2 is set. This, also explained below using an example, Idiom formation condition, also referred to as transformation Qsa a Qs.

The aforementioned, each as a disjoint class K1 = Gr1 = (Vt1y) of symbols Vt = Vt1y grammatical symbols Gr = Gr1 suffice for another Class K1 of, in compliance with the invariance condition I mentioned  transformations to be performed Qs. Thereafter, T1 a exists for each text T1N1, in whose messages by, if necessary also by the named multiple, applying the transformation Qs a symbol Vt11 a Gr1 a text (T3 = Tf3) a T3N1 with deviating from F1 was proven semantic content F3 ne F1, which instead of the result of the transformation Qs, or by means of the named sequence of this, verified text section Ts11z a Vt11 a Gr1 contains a text section Ts1yz a Vt1y a Gr1 for which by Application of the transformation Qs, or consequence of it, that is, by Shortened by a text section T31 = Ts1yz, or after Reduction by further text sections Ts 'a Vt' etc. more in message T3N1 from Vt1y dependent symbols Vt 'etc., again text T2 a T2N2 with the semantic Content F2 is generated, which is identified by a semantic reference Fp1yu a Vt1y, or additionally by further references Fpu 'a Vt' of the symbols Vt ' etc. is modified again in content F3.

The mentioned rankings Ra as well as arrangements Pr differ of those proposed for conventional dependency grammars // 12 // used, for example, for dependency relations between natural words Reference information. For the further description of the invention Unless otherwise stated, the method Mp is used without restriction of the The applicability of the method M is generally applicable, particularly in the case of Description of the grammatical messages TN a TNF the semantic TNF messages used as a basis, with the arrangements corresponding to the Pr Symbols G = (Vt) the valences Um and their assigned Saturation properties are taken into account.

As a result of the compatibility of the transformations Qs with those also based on observations of invariances I, which are determined by T = Tf given given based transformations Q, through which the grammatical symbols Gr of the grammatical communication method M are defined and determinable, the transformations Qs provide in connection with the transformations Q a universal to any natural language L Applicable procedure for the definition of and adapted for these languages L. machine communication method suitable symbols Gs with properties H (Gs), which here also in a grammatical and a semantic area can be distinguished. From the transformations Q and Qs it follows that for both areas only formal language criteria are observed in the case of utterances such as attributes such as "acceptability" or "Well-formedness" // 3 //, p. 157 without reference to their respective requirements pragmatic content other than the current state Language theories and procedures were used // 12 //, // 1 //, p. 234, not is sufficient. Here the properties H (Gs) determined by Qs and Q make it possible, the given communication task, if necessary also under Application of the above-mentioned non-linear processing operations Bb, with a  adapted to the storage and processing capacity of available computers Effort to solve.

In the extension of the names introduced above, references between Text symbols Gt as syntactic or also grammatical, between symbols Gt and Gr including G as grammatical, between symbols Gp and content F as semantic and between technical terms K as well as derived from them Complexes like semantic content F or also facts Wk as called pragmatic. The symbols Gs = Vt used as semantic, grammatical and text symbols Gp, Gr and Gt. Characters Z with text features H (T) are also preferred as text symbols Gt. used in standard expressions Tps with sufficiently high Probabilities p (H (T) / Tps) occur, with H (T) also as characteristics certain repertoires R as well as complexes Kp thereof are used which feature vectors are also called. As addresses Ad for text symbols Gt Characteristics H (T) are preferred, which these symbols with sufficient show high probabilities p (Gt / H (T)), which also simplify p (Gt / T) can be noted. The texts T a Gt like Tr a Z and Ts a Vt are also determined by the fact that they are less than predetermined tolerances dT of Standard texts T = Ta such as Tra and Tsa differ.

The use of symbols Gs = Vt, the texts of which Ts a Vt through, as a rule also multiple, use of the transformations Qs such as possibly also Q on texts which can be used as utterances Tp = Tf = Tp defined and in the texts T assigned to such grammatical messages TN are characterized by corresponding positions P and hierarchies Ra with the result that the communication method Mp by means of such messages TN associated semantic messages TNF is carried out, which in principle in, usually several, stages from messages TuNuFu = T1N1F1, T2N2F2, T3N3F3 etc. are to be built, which are subsets T1N1F1 and T2N2F2, Represent T2N2F2 and T3N3F3 etc. of the higher levels T2N2F2 etc. in the In the case of semantically ambiguous grammatical messages TN is here a message TuNuFu possibly also a disjoint class K1 = (TuNuFuv) to understand several semantic messages TuNuFuv; this Case also covered according to the invention was in the above formulation Simplification neglected. It exists for every semantic message TNF a message T1N1F1 a TNF, the grammatical message T1N1 from the Header symbol Vt11 = (Vtk11 a Grk1) of the message TNF with assigned text Tsk11z a Vtk11 is formed. The assigned semantic content F1 a T1N1F1 With probability p (K111 / Vt11) = b forms a thing K111 a Fp111 one Vt11 associated semantic hint Fp111 a Vt11, whose Usage note (q111 a Fp111) = c, as stated above, the modification the predefined situation Wk1 for K111 Probability p (K111) = a and possibly more dependent on K111 Probabilities p (Kw / K111) of other things Kw according to the resulting  Value p (K111) = b calls, which leads to a corresponding situation Wk2 = Wk21 leads. Another message T2N2F2 a TNF is added by adding a Head symbol dependent symbol Vt21 = An (Vt11) in one by this Ranking Ra and a positioning P = P2 assigned to Vt11 and Vt21 Text sections Ts11z and Ts211z generated in text T2 by their semantic Content F2 the situation now assumed for T2N2F2 as situation Wk1 Wk21 is changed into a resulting issue Wk22 etc. Also because of the possible way of creating pragmatic ones Issues Wk2 and these corresponding TNF messages assigned utterances T = Tp, the link structures differ of the complex formation of semantic messages TNF, messages F and pragmatic facts Wk2 used components Em of the for conventional, for example for constituent structure grammars // 12 //, proposed structures.

As is also explained below with the help of examples, the is in principle in the order to be carried out described structure of the invention semantic messages to be used TNF is not always a Order specified, according to which complex formations for such messages TNF and determinations of facts Wk according to procedure Mp in the respective communication files are to be made. As for procedures of Computational linguistics desired // 12 //, p. 817, the Mp method rather offers the Possibility of following different processing strategies and sequences to proceed, but adapted to the languages in question L, in detail the instructions relating to the instructions for use q a Fp a Vt of the symbols Vt Consideration of sequences of complex formations Kp = KK to be considered are. Such orders for the modification of facts Wk1 in Issues WW2 usually follow from the order of the grammatical Messages TN in texts T to be assigned complexes TN, TN such News TN present. Furthermore, there are conditions for compliance Sequences as a rule also in those cases where F is a content formation Change from a set A of things K to those sets A excluding Set B is to be made as it is for the usual language, discussed further below Negations with q = f is required. As part of this language-related The Mp procedure makes room for different processing strategies including complex formation sequences, which are effective executions according to the communication files to be performed, the available data and enable the capacities of the devices to be used. In an advantageous Training provides for the Mp procedure, taking into account the aforementioned Restrictions on the complex formation of a TNF message with any due to available data D with sufficient probability p (Em / D) displayed component Em1 to begin and with any, through the data D and components inserted into the complex Kp and used as addresses Ad Em1 with sufficient probabilities p (Em2 / D) and p (Em2 / Em1) continue to display other components Em2. See more strategies  For example, before first creating grammatical messages TN, which in semantic contents F are assigned to a second phase, in each case Complex formations either towards falling or rising ranks Ra are taken, and each with a range of evidence D such as a text length T must be taken into account, which is at least 1, in another Training also show at least 2, components Em, as a head symbol Vtk a Gk of a content N can be used.

According to the definition of things K as given by the possibly also differing opinions of the Communication partner Pa coined pragmatic terms K, which are not necessarily agrees with objectively demonstrable facts Invention before, in the above-mentioned modification of pre-existing facts Wk1 according to Wk1 q K = Wk2 with q = c the possibilities of inaccuracies as well Errors to take into account. In particular, in addition to one according to the non-linear data processing Bb provided certain neglect Contents F, resulting probabilities p (K) = 0 or p (K) = 1 by values a or 1 - a replaced with a kk 1. For example, at a Wk1 or the data Wp given probability relationship p (K1) + p (K2) = 1 in the presence of a text T1 with assigned semantic message T1N1F1, their semantic content F1, especially when using the nonlinear Procedure Bb, for thing K1 indicates a probability p (K1) = 1 for the resulting situation Wk2 probabilities p (K1) = 1-a and p (K2) = a selected. Such inaccuracies will also arise in a further training by corresponding tolerances dp of the probabilities p in the assumed facts Wk and data Wp are taken into account, whereby general also disjoint classes K1 = (Wk1x) different, the partner Pa Wk1x may be assumed to be assumed can.

As stated above in the description of the instructions for use q a Fp a Vt, are in the data Wp next to the symbols Vt assigned terms K = K5 with K5 a Fp a Vt also uses other technical terms K = K11, which according to K5 a K11 represent generic terms of the K5 things, whereby the advantageous Generic terms K11 are defined so narrowly that for subsets K5 sufficiently high probabilities p (K5 / K11) g pr above to be determined Reference values pr apply. Here in particular generic terms K11 used as the union K11 = K5 s nK5 of K5 with its negation K6 = nK5 are formed, the average of which is K5 d K6 = 0. The above according to the properties of languages L according to K5 f6 K6 Performing negations of technical terms K5 are accordingly also in accordance K6 = K11 m5 K5 by removing things K5 from such generic terms K11 reached.  

To explain this use of generic terms K, that discussed above Example 1) with a symbol Vt31 = / not /, with Vt31 a G3, which, in addition to other possible arrangements Pr is used here in an arrangement Pr, which according to an order of rank Ra Gr2 = Ar (G3) and a position specification P = gr21: Gr3: gr22 is determined in more detail, whereby the grammatical symbol Gr2 = gr21gr22 as a complex Kp of two, Sub-symbols gr a Gr introduced above with partial words vt a gr manifest which here vt211 = / are /, vt221 = / red / and vt222 = / black / are, with Vt21 = vt211vt221 and Vt22 = vt211vt222. A semantic note Fp311 a Vt31 include one thing K311 and a usage note q311 = f, which together a negation of a thing indicated by the carrier Ar, here Gr2 Provide K2yu a Vt2y, this matter possibly also by the semantic hints Fpxyu further in one or more branches As lower-ranking symbols Vtxy k Vt21 may be modified. At a text T4 = / blacks that are additionally possible with Vt31 in example 1) are not rot./, in which in addition to Vt11 and symbol Vt21 now discontinuous for text formation contributes, the text T2 = above / blacks are red / among other things assigned content F = K113K212: "Negroes are socialist oriented" under Use of another thing defined as a generic term, for example in the sense of K2121: "... are politically oriented" in another content F = K113K2122 with K2122 = (K2121 m K212) in the sense of "Negroes are not political modified socialist ", whereby K2121, as specified above, as narrow summarized generic term of K212 a K2121 according to (K212 s K2122) = K2121 and (K212 d K2122) = 0 was set.

Note that when Example 1) is expanded by another than Tag Gr4 = On (Gr2) of Gr2 symbol Vt41 = / always / with Vt41 a Gr4 further text formation T5 = / blacks are not always red / possible becomes, for the content N according to the properties H (Gr3) and H (Gr4) one Arrangement Pr with Vt31 = An (Vt41) applies and according to Fp311 a negation of one Case K411 a Fp411 a Vt41 and not as in T4 of case K212, why use a corresponding other generic term K4111 = (K411 s nK411) is.

Also from the following example 2) comes from the oral German language shows that symbols Vt formed according to the transformations Qs and Q often from the lexical words chosen arbitrarily for a language L. differentiate. To do this, the following are first followed by following Lowercase letters considered without the use of approximated text T: T1 = / one pass /, T2 = / no pass /, T3 = / one pass pass /, T4 = / no pass pass /, T5 = / an invalid passport / and T6 = / his invalid passport /. Here text T5 is used as judged sufficiently synonymous to T4. According to the transformations Q are, respectively with Ts = Vt and Vtxy a Gx, these and other texts T = S using the following objective words Vt can be represented: Vt11 = / fit /, Vt21 = / k /, Vt22 = / m /, Vt23 = / d /, Vt24 = / s /, Vt31 = / valid / and Vt41 = / un /. According to the Q,  each with Vt11 = Vtk11, T1 = Vt11, Vt21 = An (Vt11) in T2 and T4, Vt31 = On (Vt11) in T3, T4, T5 and T6, Vt41 = On (Vt31) in T5 and T6 and Vt24 = At (Vt11) in T6.

These symbol definitions which can be achieved as the results of the transformations Q are supplemented by the transformations Qs in the following way: The objective Word Vt31 receives a usage note q311 a Fp311 a Vt31 = d for one Averaging one, possibly by trailer An (Vt31) modified thing K311 a Fp311 with a carrier Ar (Vt31) assigned to it, optionally by further symbols Vt dependent on this carrier Ar thing modified or still to be modified K. For the words Vt22, Vt23 and Vt24 also apply such usage instructions q221 = q231 = q241 = d for such averaging with one thing K a Ar (Vt2y) each Carrier Ar (Vt2y). The head symbol Vtk11 comes with a note q111 a Fp111 a Vt11 zu, with q111 = c replacing the properties H (K) 1 of the things K a Wk1 by the properties H (K) 2 of these things K, as they are in the Complex formations of the things K a Vt all in the grammatical concerned Content N of Vt11-dependent symbols Vt a N among each other and with of the matter K111 a Fp111 a Vt11. Q411 a Fp411 a Vt41 also applies from Vt41 = / un / the negation q411 = f one to the carrier Vtxy = Ar (Vt41) assigned thing K a Vtxy, which, as stated above, may also be by other things K a Vt more than dependent on this carrier Vtxy arranged symbols Vt can be modified. In text T5 this is not the case so that by q411 a Fp411 the negation of one with the property "validity" circumscribed thing K311 a Fp311 aa Vt31. The objective word Vt21 = / k /, on the other hand, 2 usage instructions q211 and q212 are assigned. Of this demands, with a qr reduced by the factor mentioned above Probability p, hint q211 = f the negation of the carrier Ar (Vt21) as he is required to generate the content F of the text T2. In addition, calls with correspondingly higher probability p, hint q212 = f the use of language followed by the negation of a technical term K a Vt of a word Vt that the wearer Ar (Vt21) is optionally assigned as another trailer to (Ar (Vt21)), such as it is present in text T4 as Vt31 = An (Vt11), where q212 reference data of the order O (3) represents which the observance of a corresponding Request link order. Looks at the transformations Q and Qs Method Mp alternatively, however, the use of data Wp before, in which Texts such as T3, a separate word made up of sub-words Vt11 and Vt31 Vt111 can be assigned, that in text T4 as well as in the T7 below is supplemented by Vt21 = An (Vt111) to form a semantic message TNF.

In Example 2) it should be noted that the negation mentioned in T4 corresponds to the As a rule, use of language does not semantically express that "at all no passport, also no valid "is addressed, but, as in T5," a invalid passport ", ie only the property" validity of the passport "is negated. Due to the Mp provided for the inventive method  Formation of the narrowest possible general terms Ka to terms Kb a Ka with Probabilities p (Kb / Ka) g pr that are as high as possible become one Content formation F of negations Kc f Kb in accordance with Kc = Ka m Kb, where Ka as a union Ka = Kb s Kc the one to be rewritten by "valid passport" and "invalid passport" Subject complexes Kb and Kc is formed.

As also to be shown on the further text T7 = / no invalid passport / of example 2) includes the usage instructions for q211 and q411 Negation also the possibility of displaying in languages L not unusual double negations, whereby here too from too narrow defining generic terms K is to be assumed.

Further options for taking into account common phrases are given by the cited use of numerical factors qr zu, the use of language reproducing, changes of the resulting for the complexes Kp = F Probabilities p (F) and given by idiom formations.

In a transfer from Example 2) to the German written language L, arbitrarily defined writing rules sometimes require the text T6 to be reproduced, for example, by a text T6 '= / Its invalid passport /, which is composed of lexical words that a possessive pronoun, a form of inflection Adjectives and a noun include, with modifications by capitalization and punctuation added. As explained above and explained below using further examples, these lexical words do not allow superpositions comparable to the method Mp according to the invention for grammatical terms such as the symbols Gr and G according to the invention, which can be used in a data- and effort-saving manner. However, application of the transformations Q and Qs also results in the case of written texts T, corresponding to the above oral symbols Vt a Gr, symbols Vt, which according to the invention can be represented in accordance with the spelling rules to be observed with the help of the above-mentioned operators o detailed below. Analogous to the oral language L, the symbols Vt11 = / g result in a pass. /, Vt24 = / _s = /, Vt31 = / _valid / and Vt41 / = / _un = /, which correspond to the corresponding written language symbols Gr1, Gr2, Gr3 and Let Gr3 superimpose, which again enables data-saving generation of font texts T with grammatical and semantic contents N or F assigned according to the invention. In this case, for example, the text Ts = / _valid / can also be assigned to other symbols Grx in addition to Gr3, but on the other hand, symbol Gr3 includes the properties of a large number of other texts Ts = Vt introduced to be used as symbols Vt to save data.

In addition to the above procedures for building semantic messages TNF, which in the context of the listed, also by using instructions q a Fp  a Vt conditional restrictions on any order in the merge of the components Em a TNF, the invention sees in advantageous Embodiments also several of the following, also taking into account the procedures to be performed:

When determining text transmitted by incoming partner partner T Pa Facts Wk2, these texts T are initially completed by the listener Hr by assignment, possibly also disjoint, of several grammatical contents N added to grammatical messages TN. The structure starts from Head symbols Vt a G towards falling range. Alternatively, the Construction starting with p (Em / D) with sufficiently high probabilities the data D displayed components Em = (Vt a G) in any order. The subsequent addition of the TN messages to semantic messages TNF is based on the branches As with the lowest rank Ra assigned symbols Vt a G rising along these branches As in the direction Ranges Ra. The content of F a TNF is structured in the order of one Text T assigned to messages TN. Building a content F a TNF one out several branches As built up grammatical content N a TN takes place in the Sequence in which the branches As follow one another in an embodiment thereof also in the order of the deepest followers on these branches As. In a Further training is first the shorter branch As with branched branches As edited from this. A prerequisite situation Wk1 is gradual by the content F a TNF in the order of the messages TN in one Situation Wk2 modified, which in each case as the situation Wk1 the a-priori Probabilities p (K) for the content F 'a to be formed in the following stage T'N'F 'and the resulting situation Wk2' etc. etc. provides alternative this is done, especially in the case of extensive branches consisting of several branches merged messages TN, the modification of the facts Wk1 gradually for each branch As a TN in the order in which the branches are arranged As, where appropriate modified according to the resulting Facts Wk modified a priori probabilities p (K) for that in each case the following Ast As assigned things K a Fp a Vt a G a As used become. According to the invention with the modification of a fact Wk1 given probability p (K1) of a thing K1 also the Probabilities p (K2) through with thing K1 Probability relationships p (K2 / K1) related things K2 accordingly modified.

Adaptation to the communication files to be performed and the available ones Devices and data Wp takes place at disjoint to be considered semantic references Fp a Vt a G, the existence of which may also be due disjointly different symbols Vt a G correspondingly alternatively possible News TN is due to a reduction in corresponding Decision trees resulting in complex formation opportunities for everyone Complex formation step using the nonlinear  Data processing Bb, whereby things with K, by existing data D as possibly caused by one or more items K already used, Probabilities p (K) k ps1 temporarily below threshold values ps1 or are finally neglected. Further reductions in the repertoire R (Ks) of the subject complexes KK possible as competitive hypotheses Ks nonlinear after complex formation levels KK, as they take into account the Notes Fp a Vt of symbols Vt a N of content N or as intermediate results also the symbols Vt a As a N of a branch As can be reached, whereby each the complex KK = F with the highest, through the evidence D and data Wp conditional probability p (F) 1 ge ps2 to generate the as Communication result serving Wk2 is used. At There is more than one content F with probabilities p (F) 2 close to the maximum p (F1), several issues Wk2 are also taken into account disjointly and the further use determined by specifications Vr.

When transferring facts Wk2 through texts T to a partner Pa by an author At a disjoint repertoire R (F) of as complexes F = KK determined from contents K put together contents F, each one Facts Wk1 to be assumed at partner Pa, possibly also disjoint several such facts Wk1x, in the facts to be transmitted Wk2 modify, taking into account the data Wp things K a Fp a Vt, the words Vt as well as disjoint repertoires R (Vt) of words Vt according to Specifications Vr to be used are assigned to language L as well as the mentioned include these generic terms K to be assigned in the specified manner. The K things a Vt of a content F are defined in an order that an independent symbol Vt to be used as head Vt1 = Vtk1 assignable, thing K begins and each with things K a Vt2 of each as Tag Vt2 = An (Vt1) of the item K a Vt1 indicating the preceding thing Symbols Vt1 is continued, with the independent valences Ums of the symbols Vt1 used as addresses Ad for the symbols Vt with assigned things are and according to an address Ad also disjoint with several symbols Vt suitable items K are selected for the complex formation of a content F. The order of selecting things K a Vt according to the addresses Ad is in Direction of falling probabilities p of those indicated by the valences Um Symbols Vt a G. From things K a Vt of symbols Vt a G are preferred merged contents F a R (F) are used, which result from these symbols Vt a G grammatical messages to be formed TN with sufficiently high Assign probabilities p (F / TN) and uniqueness. It will preferably uses content F, which consists of the smallest possible number of Things K a Vt of symbols Vt that can be put together to form a message TN a Gr put together, according to the requirements also General terms K of these are used. When it comes to translations, repertoires become R (F) of content F used, which turns out to be complexes F = KK from the vocabulary Join Vq 'a Vqq the things K a Vq' assigned to the target language L ' leave, with a repertoire R (F) a correspondingly disjoint repertoire  R (R (Vq ')) of repertoires R (Vq') of vocabulary Vq 'a Vqq is assigned, their vocabulary Vqq in a previous process step translating text T were assigned to the source language L. For those too visual message T'N'F 'of the target language L' is a to be transmitted Situation Wk2 'on which the TNF of the message Source language L communicates facts Wk2 sufficiently accurately.

In adaptation to the properties of the languages L are in another Embodiment of the invention, in particular in the case of Hr Assignment of semantic contents F to texts T associated grammatical messages TN, the symbols Vt a TN of those As branches first used for content formation, the lowest-ranking followers Vt = On to the head Vtk corresponding to the positions P of the text sections Ts a Vt and ts a Vtk have the smallest distance, being equally distant Symbols Vt preceding symbols Vt before the trailing are taken into account become.

In a further training, the complex formations TNF are in sequence made according to the availability of the content for the formation of F probability relationships p (K) to be taken into account of the addressed Straighten things K in the data Wp.

In a further embodiment, the assignment of semantic Messages TNF on incoming texts T by listener Hr the components Em in an order after falling, by probabilities p (Em / D; Wp) given certainties with which they are available through available evidence D taking into account the data Wp including the addressing Ad by already considered components Em are displayed, whereby by the Instructions for use q given orders are taken into account.

The proposed non-linear data processing Bb advantageously has the result that the one to be processed both in the role of the listener Hr and the author At Diversity of disjoint possibilities in the complex formation of grammatical Messages TN and semantic messages TNF are disjointed accordingly Repertoires R (Em) of components Em that are basically too Decision trees with approximately exponential with the number of each Complex Kp joined components Em growing branches would lead to early manageability at every branching option low, for example each comprising only one or two alternatives Solutions considered as competing hypotheses remain limited. Here it makes the provisional provisional provision of initially less probable partial solutions Ks possible in such cases where advanced complex formations are not sufficiently probable Complex formations Kp can be achieved, initially deferred solutions Ks reactivate and thus to overall sufficiently probable final results  TNF to come. According to the invention, in a further embodiment Approach Bbr by Bb also reset competitive hypotheses Ks, at to which the component Em a Ks of the complex formation step in question little information j k js contributes to the semantic content F. The result of the Symbols Gs underlying transformations Q possible description of their Properties H (Gs) by predominantly low-order data O (n) make it still possible, the capping of possible solution branches is already possible to carry out n-1 complex formation steps, ie with data of order O (2) after one step at a time.

As already shown above for the case of negations, method Mp generally before, in the data Wp in addition to that in the semantic references Fpxyz a Vtxy of the symbols Vtxy contained things Kxyu a Fpxyu other things To take Kv into account, the Kxyu high enough Have probability references p (Kv / Kxyu) g p1. This is an example again explained in 2 units Gx of the oral German language, in which G1 the symbols Vt11 = / people /, Vt12 =, expressed in German for this purpose / Fruits /, Vt13 = / books /, Vt14 = / glasses /, Vt15 = / carpenter / and Vt16 = / Certificates / and G2 includes the symbol Vt21 = / good /. In addition to these symbols Vtxy assigned things not explicitly described here Kxyu a Fpxyu a Vtxy one thing Kv = Ka is taken into account, like Ka: "one for people important function or property "can be expressed. Things become Kxyu also assigned things Kxyu1, which approximately as K1111: "moral quality", K1211: "taste", K1311: "worth reading", K1411: "optical quality", K1511: "manual skill" and K1611: "tested ability" are to be designated. Note Fp211 a Vt21 includes one Thing K211: "a thing Ka indicated by a carrier Vt1y = Ar (Vt21) is in to a satisfactory extent "and a link q211 = d for averaging with a thing K1y11 a Vt1y of the carrier Vt1y. Comparatively high conditional probabilities are assumed p (Kxyu1 / Kxyu) g p1, p (Ka / Kxyu) g p1 and p (K211 / Ka) g p1. For those with G1 = Gk1 = Ar (G2) oral sentences S = Vt21Vt11 = / good people /, Vt21Vt12 = / good fruit /, Vt21Vt13 = / good books /, Vt21Vt14 = / good glasses /, Vt21Vt15 = / good carpenter / and Vt21Vt16 = / good testimonials / are the result intuitively assigned content F as complexes K1yuK1yu1KaK211 of the relevant Things like K in the sense of "morally good people", "tasty Fruits "," books worth reading "," sharp glasses "," skillful carpenters "and "Certificates with good grades", while other subject combinations, such as those at other content assignments to the the words mentioned can result formally, for example in the sense of "morally good Fruits "or" tasty people "etc., can be avoided. Details of the complex formations KK used for this, in which in particular averages according to q = d of the terms Kxyu, Kxyu1, Kq and K211 are to be formed, are explained using further examples below. In the present case it proves to be particularly advantageous that a conditional one saves data  Probability p (K211 / Ka) in the formation of all contents F application finds.

The formulation of the properties H (K) of the things to be used K a Fp takes place in accordance with the, also specified in more detail by specifications Vr Circumstances of the communication to be conducted regarding which the partners interact by definition, where applicable also as a user act and according to the invention human as well as machine users are taken into account. So in an apprenticeship, which one natural language communication with partner Pa Electronic calculators do the things K and complexes F = KK of them defined symbol contents according to the tasks to be solved by computers used as known from command languages clearly defined for this purpose are. For human users, in training for partners Pa, too Things K and their links q in addition to defined symbols if necessary verbal descriptions also to be made in the language L concerned used, which are so detailed that sufficiently clear Differentiations of different things K and complexes F formed from them and facts Wk result as well as a definition provided interaction between the Communication partners Pa is possible in a sufficiently clear manner. At The wording of the items K a Fp a Wp should be noted that their Links according to the instructions q a Fp again pragmatic content F and Data Wk result with that of experts in the languages L and thus treated facts Wk recognized content F sufficiently agree, whereby, as exemplified, the things K also using other Terms A including things K such as generic terms, subsets, Complements etc. are formulated and according to the by Probabilities of higher order p (K / K ') etc. given references Data Wp by definition also include complexes Wk = KK of such complexes KK contain the properties H (KK).

In a configuration adapted to the communication file to be performed data Wp with a limited repertoire R (K) of terms K used, which can be described by facts Wk is adapted, which are addressed in the communication. In one here adapted training is a repertoire R (Vt) of symbols Vt takes into account those things K of this repertoire R (K) with sufficiently high Probabilities p (K / Vt) g ps are assigned, as well as other symbols Vt that is sufficient in utterances Tp, based on a standard length Tps high probabilities p (Vt / Tps) are to be expected. In line with the Communication conditions become subject areas, for example takes into account which the order and delivery of certain goods and Assortment of services, the production of certain goods, transport offers for  Goods and people, making appointments, operating certain machines including computers or credit bureaus.

According to the semantic assignment provided according to the invention Messages TNF for both incoming and outgoing texts T are the Complex formation for this through all according to the method Mp both as listeners Hr as well as author At machine working communication partner Pa, which for example as a conversation partner, translator, interpreter, stenographer or Corrector are used to make the data Wp for this The data Ws and Wst mentioned must be made available lock in.

Compliance with the arrangements Pr of the method Mp Symbols Vtxy a Gx a N in a grammatical content N in one according to one Linking instruction qxyu a Fpxyu a Vtxy of a symbol Vtxy Linking one thing Kxyu a Fpxyu with other things Kxyu 'more Symbols Vtxy 'a Gx' a N is generally followed by following Conditions Bc reached, which are based on method M created grammatical messages TN is formulated: Kxyu thing is on a way indicated by hint qxyu with one or more things Kxyu ' a Vtxy ', which, with Vtxy' a Gx 'a Gw' and Gx 'a Gw', one or belong to several classes K1 = Gw '= (Gx') of units Gx ', which also are specified by reference qxyu, such a unit Gx 'of the unit Gx in the same branch As of the content N closest in one also by reference qxyu is assigned to the given ranking direction. Condition Bc closes the Possibility that instead of a thing assigned to the symbol Vtxy 'Kxyu' a subject complex Kxyu "for linking with Kxyu is also to be provided, which is characterized by Linking Kxyu 'with other things Kxyu' '' more in content N with Vtxy 'a Gx', especially lower-level linked symbols Vtxy '' 'a Gx' '' was modified, where appropriate, as mentioned above, the languages L adapted sequences are to be considered.

Adapted to the conditions of the languages L and the available ones Devices become certain in TNF messages with sufficient frequency occurring complexes VtVt of symbols Vt with assigned complexes KK of items K assigned to these symbols Vt also used as idioms which as additional, each associated with a further grammatical symbol Gr Symbols Vt1 = VtVt with associated semantic references Fp1 and these assigned terms K1 = KK and instructions for use q1 a Fp1 Find use. This will increase slightly at the expense of one Storage effort a reduction in processing effort in education of the TNF news reached.

In a further training, as far as the properties of the languages L this allow assigned to the symbols Vt usage instructions q which  Links of the relevant symbol Vt1y a G1 assigned to a branch As with symbols Vt2y a G2 of units G2 of this branch As specify which ones are arranged in a certain direction, for example higher. Here, for example, in the case of G1 k G2, an indication q1yu a Fp1yu a Vt1y determines a G1 the link between a symbol Vt1y and a symbol Vt2y a G2. The associated waiver of an additional assignment of notes q2yu the symbols Vt2y a G2 for their connection with the words Vt1y a G1 der lower-ranking units G1 has an advantageous data saving in the description of the symbols Vt in the data Wp.

The formation of semantic messages TNF by assigning semantic content F for texts T with assigned grammatical messages TN using of the Bayesian approach Ba described above is used in the following example 3) first grammatically clearer, correspondingly clear for the case identified and unique messages assigned to TN, texts T a TN explained, which is initially formulated here abstractly and below by real texts T to be fleshed out. First consider 2 units G1 and G2, which in a content N1 in a predetermined by the valences Um of these units, Arrangement Pr1 according to G1 = Gk1 = Ar (G2) according to a positioning P1 = G2: G1 are present, with one word Vt11 a G1 and Vt21 a G2 is taken into account, so that content N1 is assigned a set S11 a N1 = Vt21 Vt11. Accordingly, in addition to text T1, message T1N1, content N1 and sentence S1 also apply here than with evidence, i.e. p = 1, available evidence D. Content is taken into account F, which are formed from things K a Fp a Vt, by the evidence D, so here by the complexes S of the words Vt a N1, a priori with sufficiently high Probabilities p (K / Vt) g ps above threshold ps to be determined are displayed. Here Vt11 are two things K111 a Fp111 a Vt11 and K112 a Fp112 a Vt11 assigned. Symbol Vt21 shows a semantic hint Fp211 a Vt21 on, the one thing K211 a Fp211 and a usage note q211 a Fp211 comprises a linkage of K211 according to (Klyu a Ar (G2)) q211 K211 with q211 = d requests. The evidence D = S11 = Vt21Vt11 are therefore a priori two contents F111 = K111 d K211 and F112 = K112 d K211 Probabilities p (F111) = p (K111) × p (K211 / K111; Pr1) and p (F112) = assign p (K112) × p (K211 / K112; Pr1), where Pr1 is the above Shows arrangement condition. For the values p (D / F) still required in Ba applies here, with D = Vt21Vt11, p (D / Fu) = p (Vt21 / K211) × p (Vt11 / K11u) with u = 1 or 2.

According to the present data Wp, which includes a prerequisite situation Wk1, with t kk 1, p (Kxyu) = t ³ and, with the intended consideration of condition Bc1, p (Kxyu) = t ³ continue to apply for the a priori probabilities to be used in sufficient approximation. K211 / K111) = t, p (K211 / K112) = t ³ and p (Vt21 / K211) = 1, and, as explained below, p (Vt11 / K111) = ½ and p (Vt11 / K112) = ¹ _{/ 2nd} According Ba follows p (F111 / D) ue 1 and p (F112 / D) ue t ² wherein Ne ue ^t4 / 2. For the two competing hypotheses Ks1 = F111 and Ks2 = F112, p (Ks1 / D gg p (Ks2 / D) follows with respect to the available evidence D.

Example 3) also serves to explain the non-linear processing Bb provided in one of the mentioned embodiments of the method Mp according to the invention of the probabilities p (F / D) determined according to approach Ba. Process part Bb provides for competition hypotheses Ksx a (Ksx) of a disjoint class K1 = (Ksx) probabilities p (Ksx) ge pr above a reference value pr to be determined, possibly also up to p = 1, and values p (Ksx) k pr, possibly also to p = 0, taking into account compliance with the probability sum Ca for these hypotheses Ksx. In the preferred embodiment mentioned, reference values pr are determined such that they are a certain factor bk 1 below the probability p (Ks1) of the hypothesis Ks1 with the highest probability p (Ksx). If a value b = t is selected accordingly in the present example 3), then p (Ks1 / D) = 1 and p (Ks2 / D) = 0. This results in the probability p (K111 d K211) for this on the basis of the available evidence D. that K111 and K211 exist together, increased from a value p = t ⁴ to a value p = 1 expressed by facts Wk2 compared to a situation Wk1 to be assumed a priori according to the data Wp.

According to an embodiment of the invention mentioned above, facts Wk by a combination of several grammatical symbols Vt a G Evidence D = TN also gradually changed into further evidence Wk, whereby an order starting with the head G1 = Gk1 must be observed, that with a Trailer G2 = An (G1) and other trailers of G2 as well as if necessary is continued by Gk1. In the example given, this initially results Issues Wk2 = Wk2 ', where due to the evidence D = Vt11 Probabilities p (K111 / D) = ½ and p (K112 / D) = ½ apply, which apply to the subsequent consideration of the trailer G2 = An (G1) in facts Wk2 = Wk3 'can be modified, for which here again the above Result in probability values p.

In the general case of a grammatically ambiguous text T, the evidence D of a disjoint class (TN, TN) of sequences of grammatical Messages TN is to be assigned, is provided in the approach Ba corresponding further a priori probabilities of the form p (T / N) and correspondingly p (T / N, N) to consider. In an advantageous embodiment, according to the Principle of maximum entropy for these probabilities of equal distribution Assume f 64848 00070 552 001000280000000200012000285916473700040 0002010015858 00004 64729 if none are sufficiently different, for example there are indications resulting from the grammatical procedure M.

The use according to the invention of, in the messages TNF at the same time as text symbols Gt, grammatical symbols Gr and semantic symbols Gp to be used, symbols Vt, each of which has a disjoint class K1 = (Gp)  alternatively, applicable semantic information is assigned to Fp, which is both different technical terms K as well as different usage instructions q for this purpose, on the one hand, allows the typical L natural language semantic ambiguity of individual text sections T in sufficient to the Languages L to reflect adapted dimensions, while on the other hand the intended Taking quantitative properties H (Vt) including H (K) into account Symbols Vt and technical terms K in connection with the quantitative Processing methods Ba and Bb at the same time an effective and safe definition of the transmitted by such a text section T in a message TNF makes semantic content possible, the characteristic of languages L. Dependency of this content on the context is considered as it through further text sections T of this message TN and through the relevant described by a pragmatic prerequisite Wk1 Situation is given. This assignment has several semantic each Notes Fp to a symbol Vt advantageously result in the fact that in a language L a large variety of semantic content F and pragmatic facts Wk represented by a comparatively small number of symbols Vt and can be transmitted, provided that the storage and Processing capacity of available devices to adaptable approaches the combination of which are available as they are by the method according to the invention Mp are shown. It allows the intended use of words Vt, which are both grammatical symbols Gr and semantic symbols Gp act, such combinations both due to grammatical, syntactic Relationships taking into account properties H (Gr) as well as additional due to semantic and pragmatic properties H (Gp) including H (K) generate, resulting in increased security when assigning semantic TNF messages especially in the case of disturbed or non-standard or incorrectly written texts result. Method Mp does not rule out, in addition for the mentioned data H (Vt), further available data, for example, about statistical properties H (Gt) of text symbols Gt such as characters Z or To take into account character complexes ZZ, such as those mentioned in the at the current state, Markov probabilities used for the Sequence of such complexes ZZ including lexical words available.

To take account of the dependency of the spelling or pronunciation of the words Vt to be used in languages L on the respective text context, which saves data, these words Vt are also noted in the data Wp using the above-mentioned operators o , which are also used in method M and which are used in texts T cause certain changes and usually disappear in them. From this, o = / g / requires the capitalization of the first word text Ts a Vt a Su a N belonging to a text T a Su of sentences Su a N, as well as the maintenance of a distance to a preceding text section T. Operator o = / _ / requires the Compliance with a spacing from a preceding text T, o = / = /, which is also guaranteed by a line change, the writing together as a prefix or suffix with an adjacent text T while removing any spacing characters that may be assigned to it, capitalization of this text T possibly being taken from the prefix becomes. Other operators o adapted to the respective languages L see, for example, the extension of the lexical word / a / to / on / before vowels etc. in English. Operator functions are also assigned, for example, to certain punctuation marks, which, for example, require the deletion of adjacent punctuation marks, without T in the text disappear. A further data saving in the storage of objective words Vt a G in the data Wp is also achieved by their notation as complexes Vt = vtvt of the partial words vt mentioned, which are assigned to subunits g = (vt) arranged in corresponding complexes G = gg.

For a further explanation of the method Mp according to the invention, the words Vt of example 3) above are substantiated by texts Ts a Vt from the German written language L, which are represented by their standard texts Tsa, also noted as Tsaxy = Vtxy, using the operators mentioned o be noted as a sequence of block letters and introduced punctuation marks Z. In accordance with the uniqueness assumed for this purpose, the words Vt and word sequences VtVt thus noted are obtained as clear evidence D. The data Wp from example 3) are initially assigned the objective words Vt11 = vt111vt121vt131 = / g plums./ with vt111 = / g /, vt121 = / _plums / and vt131 = /./ and Vt21 = / _reife / Set S11 = vt111Vt21vt121vt131 = / Ripe plums./ can be combined. Here G2 = An (G1) applies, this arrangement Pr being determined in more detail by a position specification P = g11: G2: g12, in which G2 by definition also includes other units G which are possibly dependent on G2.

The things K a Fp a Vt assigned to the symbols Vt of Example 3) are here by the following quantities and supersets that are not explicitly stated A described descriptions characterized: K111 = "fruits of the type A11; K112 = "People A12 with poor performance A21 a A2 in any Disciplines A2 "; K211 =" concerning the degree of ripeness A3 of fruits A11 Property A31 ". The one to be made for content F111 of S11 Averaging F111 = K111 d K211 is shortened accordingly as Quantity formation A11 d A31 of the content "according to A31 ripe fruits of the type A11" circumscribed. For F112 = K112 and K211, a maturity follows according to "according to A3 Minor provider A12 of the type A21 "circumscribed content, according to the data Wp, as stated above, an extremely small one compared to p (F111) Probability p (F112) occurs. The in the data Wp by p (Vt11 / K11u) = ½ Approximate probabilities express that K111 is also indicated by an above Word Vt12 not mentioned with the sub-word vt122 = / plum / and K112 can also be indicated by another subword vt123 = / bottles /.  

In example 3) there is an additional note Fp212 = for the symbol Vt21 (K212; q212) with q211 = d and K212 in the sense of "no more viable and therefore ripe for being replaced by other people " If p (K212 / K112) = t is taken into account, F113 = follows for the further possible content K112 d K212 approximately p (F113 / D) = p (F111 / D). Furthermore, the possibility takes into account that the sentence S11, for example by a previous statement or current situation, a situation deviating from the general data Wp Wk1 must be assumed that p (K112) = (1-pa) is highly likely, With pa kk 1, it can be expected that said "under-contractors of type A21" will be addressed even p (F113 / D) g p (F111 / D) follows. Note FP212 is below neglected.

Further uses of the pragmatic information Fp can be seen in an extension of Example 3) by an objective word Vt31 = / _un = / with Vt31 a G3, in which the operator o = / = / above is a compilation of Ts311 = Vt31 as a prefix one word Ar (G3) assigned to a carrier such as Vt21 a G2. Together with Vt11 and Vt21, a set S21 = vt111Vt31Vt21vt121vt131 = / unripe plums / with S21 and N2 is possible, where N2 is determined by G2 = An (G1) and G3 = An (G2). According to Fp311 a Vt31, with q311 = f, the negation K21u f K311 of the content K2yu a Vt2y a G2 of the carrier G2 = Ar (G3) treated above is to be formed. As explained above, the relationship A31 s A32 = A3 applies to the thing K211 = A31 a A3 characterized by the generic term A3 = "relating to the degree of plant ripeness", where A32 stands for the negation of A31, ie for "immature", for which with sufficient approximation p (A31 / A3) + p (A32 / A3) = 1 apply. Furthermore, p (K211 / K111) = p (A3 / K111) × p (A31 / A3) = t with p (A3 / K111) = a, p (A31 / A3) = b and a × b = t and p (K311 / K211, K111) = p (A3 / K111) × (1 - p (A31 / A3)) = a × (1-b). Since p (A3 / K112) ue p (K211 / K112) = t ³ , according to Ba and Bb a single pragmatic content F211 also follows for S21 with sufficient probability p (F211 / D)), which with "fruits A11 without ripeness A31 "can be described.

The data Wp of example 3) can be adapted to the usage of language. by another thing K213a Fp213 a Gp21 a Vt21, also with q213 = d, add, after which the word / _reif / also "a high degree of development of Perfection A22 in any discipline can display A2 " Taking K213 into account, evidence S21 can be used alongside F211 Assign content F212, for example, to the sense of "underdeveloped non-experts" because the Probabilities required for this are p (K112), p (A2 / K112) and (1 - p (A22 / A2)) have sufficiently high values so that p (F212 / S21) uep (F211 / S21) follows.

To further explain the function of the usage instructions q, example 3) is expanded by a further unit G4 with the words Vt41 = / _ and large / and Vt42 = / _ and green /. According to the grammatical communication method M used, grammatical symbols Vt are used here, which are composed of several natural words in accordance with the transformations Q on which the methods M and Mp are based, whereas, also in accordance with the Q, the above-mentioned word Vt31 is a part represented by a natural word. For the two semantic symbols Gp4u a Vt4u, with u = 1 or 2 and v = 1 or 2, 2 pragmatic instructions Fp4uv are taken into account, each of which contains the same thing K4u1 = K4u2, but each contains 2 different linking instructions q4ul and 14u2. For this, q411 = q421 = G4 d (Ar (G4)) d Ar (Ar (G4))) with G1 a Ar (Ar (G4))) and G2 a Ar (G4), where the units Gx indicated by the Symbolize sentence Sa N displayed Fpxyu a Gpxy a Vtxy a Gx a N symbolize. Furthermore, with this symbolism q412 = q422 = (G4 d Ar (Ar (G4))) s (Ar (G4) d Ar (Ar (G4))). The assigned items K41v and Gp41 and Vt41 are described by K41v = "a certain extent A71 of a property spatial dimension A7" and K42v = "a certain expression A81 of a property color A8". The sentences S31 = / ripe and large plums./ and S32 = / ripe and green plums./ with S31 a N3 and S32 a N3 are considered, whereby the grammatical content N3 by expanding N1 with the additional tag G4 = An (G2 ) is produced. For example, based on Theorem 31, q411 indicates that an average of Vt41 must be averaged with the average of Vt21 and Vt11, while q412 requires the union of the average of Vt41 and Vt11 with the average of Vt21 and Vt11. Accordingly, two semantic contents F3uv are to be taken into account for each of the two sentences S3u if the items K112 and K121 considered above are neglected here. According to approach Ba, the following 4 probability values must first be taken into account with u = 1 or 2, a priori:
p (F3u1) = p (K111) × p (K211 / K111) × p (K4u1 / K211, K111) and
p (F3u2) = p (K111) × p (K211 / K111) + (p (K4u1 / K111) -pu),
where pu = p (K211 / K111) × p (K4u1 / K211, K111) indicates the probability of the average of K211 and K4u1 under the conditions listed.

In addition to the above data, Wp applies
p (K411 / K211, K111) = 0.8,
p (K421 / K211, K111) = t ² and p (K4u1 / K111) = t.

This follows p (F311) = 0.8t ⁴ , p (F312) = 1.2t ⁴ , p (F321) = t ⁶ and p (F322) ue 2t ⁴ . Under the simplifying assumption, which often does not correspond to the usage of the language, that for the probabilities p (S / F) to be taken into account in Ba, the same values p (S3u / F3uv) = a, for example with a = 1, follow in sentence S32 for content F321, which speaks of "plums that are both ripe and green," there is a very low probability p (F321 / S32) = 0.5t ² , while for F322, which speaks "ripe plums and additionally green plums "are addressed, p (F322 / S32) ue 1. In the case of sentence S31, which according to method M, also according to the intuitive judgment of experts, is classified as grammatically identical to sentence S32, the contents F311 = "plums that are ripe and large" and F312 = "ripe" Plums and large plums "have approximately the same probabilities p (F311 / S31) = 0.4 and p (F312 / S31) = 0.6.

However, since content F321 is due to a priori probabilities to be considered is practically inexpressible by sentence S32 is the use of language corresponding to p (S32 / F321) kk 1, specifically also p (Vt42 / Fp421; Fp211) kk 1 put. According to expert judgment, sentence S31 does not make content F312, but instead F311 expressed, according to which a use of language has developed that by p (S31 / F311) gg p (S31 / F312). Is used for the data Wp correspondingly, for example, p (S31 / F311) = 0.8 and p (S31 / F312) = 0.2, Ba follows the expert judgment better corresponding values p (F311 / S31) ue 0.7 and p (F312 / S31) ue 0.3.

In order to achieve a data-saving formulation of the data Wp, the Invention before, semantic references such as the Fp4yu a Vt4y of the words Vt4y a unit G4, as far as the language L in question permits, largely Identical links, so q4yu1 = d and q4yu2 = s. Because of the combinatorial nature of a repertoire R (Vtxy) according to Wp with the symbols Vtxy exploding possibilities of sentence formations S = VtVt are also the above, for the probability determination according to Ba required probabilities p (S / F) as products of terms p (Vtxy / Fpxyu) formed in the data Wp approximately linear with the number of words Vt grow. For example, in example 3), values are simplified p (Vt1y / Fp1yu) = 1, p (Vt2y / Fp2yu) = 1, p (Vt4y / Fp4yu1) = 1 and p (Vt4y / Fp4yu2) = 0.1, whereby, as above, q4yu1 = d and q4yu2 = s applies, for the follow S31 = vt111Vt21Vt41vt121vt131 and S32 = vt111Vt21Vt42vt121vt131 from this assignment probabilities p (S31 / F311) = 1, p (S31 / F312) = 0.1, p (S32 / F321) = 1 and p (S32 / F322) = 0.1. With the above calculated values p (F) result in S31 probabilities p (F311 / S31) = 0.87 and p (F312 / S31) = 0.13, after which the content F formed with q = d is displayed as applies. In contrast, probabilities follow for S32 assuming t = 0.1 p (F321 / S32) = 0.05 and p (F322 / S32) = 0.95, after which, also in accordance with an expert judgment, which is expressed as content F with q = s.

According to one of the configurations mentioned, the invention also provides for usage instructions q the linking parameters such as q412 a Fp412 a Vt41 and q422 a Fp422 a Vt42 from Example 3) to be supplemented by parameters qr, which a numerical reduction in the connection of things K41w a F41 or K42w a F42 with the things K of the words Vt2y a G2 and Vt1y a G1 of this example achieved probabilities p (KK) of the respective one Effect complex KK. For example, according to language usage (qr a Fp412) = 0.1 set, it follows, now again assuming p (D / F) = 1, according to intuition p (F311 / S31) = 0.87 and p (F312 / S31) = 0.13. With this qr value At the same time, the relationship p (F321 / S32) kk p (F322 / S32) is preserved in sentence S32,  as it corresponds to the expert opinion. For example, at a value t = 0.1 probabilities p (F321 / F32) = 0.05 and p (F322 / S32) ue 1.

To further explain the possibility given by method Mp, in accordance with intuitive human language use, by combining ambiguous text sections T with quantitatively determinable certainty and uniqueness, to express a multitude of different semantic contents F and pragmatic facts Wk in a context-appropriate manner, example 3) is illustrated by further symbols Vt expanded. The words Vt14 = vt111vt124vt131 = / g researchers./ and Vt15 = vt111vt125vt131 = / g velocities./ with the things K141 a Fp141 a Vt44) = "people A12 who carry out research activity A51 are considered, again with Vtxy a Gx "and (K151 a Fp151 a Vt15) =" concerning the timing A61 of any processes A6 ". Furthermore, Vt22 = / _large / with the thing already assigned to Vt41 K221 = K411 = "extent A71 of a spatial extent A7". In adaptation to the use of this language L, Vt22 and Vt41 are assigned yet another thing (K222 a Fp222) = (K412 a Fp412), which is described as "a certain extent A91 relating to any property A9". For all logic parameters q21z = q41z = d applies. The further sentences S12 = Vt22Vt14 = / large researchers./ and S13 = Vt22Vt15 = / large speeds./ are to be added to the grammatical content N1 considered above. According to the semantic hints Fp considered here, S12 can be assigned a content F122 = K141K222 in the sense of "highly qualified researchers A12A51", on the other hand, because of the possible meanings K221 = "spatially large" of Vt22, a content F121 also seems disjoint not impossible in the sense of "tall researchers". However, p (F122 / S12) vs. p (F121 / S12) proves if the values p (Fxyu) × p (S12 / Fxyu) to be taken into account according to Ba are compared. Specifically, with the introduced generic terms A, taking into account data Wp to be provided accordingly, the probabilities p (F122) = p (A12A51A91) = p (A12) × p (A51 / A12) × p are sufficiently high for the terms determining the content F122 (A91 / A51), where here A51 = A9. Sufficiently high values p (S12 / F122) = p (Vt14 / A12A51) xp (Vt22 / A91) also apply to the assignment of the selected words Vt to this content. The alternative content F121, on the other hand, is to be presented as a subject complex F121 = K141A7K221 = A12A51A7A71, because only because of the terms mentioned and the related references can a sufficiently high probability p (F121) = p (A12) × p (A51 / A12) × p ( A7 / A12) × p (A71 / A7). The word choice S12 is supported by content F122 = "great researcher" by sufficiently probable terms according to p (S12 / F122) = p (Vt14 / A12A51) xp (Vt22 / A91). In contrast, for F121 = "tall researcher" there is an insufficiently probable value p (S12 / F121) = p (Vt14 / A12A51A7) xp (Vt22 / A71) for the first product term, since Vt22 is often used to express spatial size, but not the word Vt14 = / g researcher./, if "people A12" should not only address their "research activity A51" but also their "spatial extension A7".

Also in sentence S13 = / high speeds. / Is not of the "expansion aspect A7 "of case K221 a Vt22 because there is a basis for this Contents F132 = K151A7K221 with K151 = A6A61 in the one to be created for this Probability p (F132) = p (A6) × p (A61 / A6) × p (A7 / A6) × p (A71 / A7) dem Term p (A7 / A6) k ps and thus p (F 132) k ps too high p values get. S 13 clearly addresses content F131 = K151K222 = A6A61A91 with A61 = A9, which literally means "high Velocities "means with the probability that p (F131) = p (A6) x p (A61 / A6) × p (A91 / A9) assume sufficiently high p-values for all terms are.

The variety of methods in Mp by combining ambiguous symbols Coverable language content that can be transformed into Mp by combination Let semantically ambiguous symbols be clearly covered, will continue clarified by supplementing example 3) by a further unit G5, whose Possibilities of grammatical arrangements Pr among other things by a Rank G2 = Ar (G5) and position P = Ar (G5): G5 can be described. Consider the words Vt51 = /, large / and Vt52 = /, green /, which semantic hints Fp5uv with things K511 = K411 and K521 = K421 as well same usage instructions q511 = q521 = d for averaging d the Thing KSuv with one thing Kxyu a Vtxy a G1 of a word Vtxy of the closest higher-ranking unit Gx g G5 are assigned, whereby by definition Kxyu can be modified by things K a Vt a G further trailer G = An (Gx). A correspondingly formed sentence S41 = / Ripe, large plums. / Is hereby clearly assigned a content F411 = F311, to represent it S41 Because of the ambiguity of sentence S31 above, it is preferable from which also, according to the use of the German language L, the above low probability p (S31 / F311) required for approach Ba follows. The one with Vt52 producible set S42 = / Ripe, green plums./ comes with the named Data a content F421 = F321 in the sense of "green but still ripe" clearly assignable, while the same content F321 because of it, according to Ba too considering low factual probability p (F321) = p (F421) is practically not expressible by the above sentence S32.

The advantageous result of the introduction of the subunits g a G Data saving when formulating the data Wp is taken into account another unit G6 = g11g61g12g13 in Example 3) clearly, which under Use of the sub-units g11, g12 and g13 used for G1 and one Sub-unit g61 with sub-words vt61y a g61 like vt611 = / _die / and vt612 = / _diese / etc. is formulated. When using another unit G7 with words like Vt71 = / _ tires /, Vt71 = / _large / etc. with G6 = Gk6 and G7 = An (G6) if a position P = g61: G7: g12 is observed, blocks like S51 = / Die  ripe plums / etc. possible. That unit G6 grammatically increases from unit G1 is different, it follows, for example, that arrangements Pr with G2 = An (G6) and G7 = An (G1) for incorrect texts T ne Tf as T = / Die mature Plums./ or T = / Ripe Plums./ would lead.

Coping with ambiguities common in language according to the invention is also eliminated a further extension of example 3) by a unit G8 with verb forms the 3rd person plural like Vt81 = / _schmecken / and G9 with adverbs like Vt91 = / _gut / and Vt92 = / _sauer / can be seen. According to the data Wp apply to the Arrangements Pr in addition to other options G8 = An (G1) and G9 = An (G8) corresponding position information P, after which sentences like S61 = / plums taste./, S71 = / plums taste good./ and S72 = / taste plums acidic / become possible. The transformations Qs become Vt81 disjoint semantic clues Fp811 with one thing K811 and Fp812 with one thing Thing K812 = K811 and K911, both with a linking parameter q811 = q812 = d assigned to the averaging with one thing K a Ar (G8). Further applies for Vt91 a note Fp911 with thing K911 and for Vt92 a note Fp921 with Thing K921, both with q911 = q921 = d for averaging with one Subject complex KK with things K a Ar (G9) and K a Ar (Ar (G9)). Here, again simplifying, author opinions are displayed, according to which a thing linked according to q811 "has any effect on the taste buds", at K911 "a subject complex linked according to q911 a pleasant effect comes ", and with K912" has an effect that is perceived as sour " K111 a Vt11 are understood as "fruits" as above. The According to language usage, Vt81 becomes another thing K812 in addition to K811 = K811 d K911 in the sense of "pleasant on the taste buds" assigned.

With the corresponding data Wp, a semantic content F611 = for sentence S61 K111 d K812 = K111 d K811 d K911 in the sense of "plums taste good" compared to the more pragmatic, but less semantic probable content F612 = K111 d K811 in the sense of "taste plums somehow "preferred since the difference of the pragmatic to be used in Ba Values according to p (K812 / K111) k p (K811 / K111) by the counteracting one in Ba Difference in usage according to p (Vt81 / K812) vs. p (Vt81 / K811) is overcompensated. Despite this usage, word Vt81 shows in sentences S71 and S72 not K812, but K811, since in S71 the formation of a content F711 = K111 d K811 d K911 = F611 by adding the thing K911 a Vt91 to Contents F611 a S61 because of p (F71) = p (F611) do not add information j, so Condition would violate Bbr. In sentence S72 the addition of Thing K921 a Vt92 to F611 because of p (K921 / K911) ue 0 for one with K812 too educational pragmatic content F722 = K111 d K812 d K921 because of p (K921 / K812) kk 1 a correspondingly negligible pragmatic probability p (F722) ue 0. As from the other text sample / chefs taste it Difference. / Follows, a text section can taste = / _ / also as transitive  Verbform Vt are used which, according to method M, other than that symbol Vt81 a G8 is to be assigned and accordingly further, in Example 3) indicates not considered semantic hints Fp. That that in Example 1) as a word (Vt31 a G3) = / _not / introduced symbol not also the Unit G9 of example 3) can be assigned, also follows from further texts T. Sections like /. . . taste very good /, in which when transforming Q Replacement of / _good / by_ / not / invariance conditions I would be violated.

The meaning of the use of semantic hints Fp a Vt, which assign the words Vt not only disjunctly different technical terms K a Fp, but also disjunctively different usage instructions qa Fp, for a language-appropriate applicability of the method Mp can be seen by extending Example 3) by another subword vt126 = / _flags / a corresponding word Vt16 a G1 and by a word (Vt23 a G2) = / _blue / and by the words (Vt32 a G3) = / _weiß = / and (Vt33 a G3) = / _grey = / make clear. For this, 2 sets S22 = / gray-blue plums./ and S23 = / white-blue flags./ are considered, the contents of F221a S22 roughly by "plums A11, which are uniformly colored in a color B3 mixed from blue B1 and gray B2" and F231a S23 "Flags B4, which are colored blue B1 and additionally (in other places) colored white B5", are described, whereby in addition to the A11 introduced above, the terms Bx characterize the corresponding technical terms Kxyu a Fpxyu a Vtxy according to B1 = K231, B2 = K331, B4 = K161 and B5 = K321. Adapted to the usage of language, the usage instructions q must be taken into account, which are noted as q321 = q331 = d and, for simplification, q322 = q332 = s (d). Here d indicates an averaging G1 d G2 d G3, in which the units G stand for the respective items K a Vt a G, while s (d) more complexly requires a union of averages according to (G1 d G2) s (G1 d G3) , the linking instructions, as generally defined, each refer to the units G arranged closest to content N. This can be used, among other things, to formulate the contents (F221 a S22) = A11B3 with B3 = B1B2 for sentence 22 and F231 = (B4 d B1) s (B4 d B5) for S23, for whose probabilities p (F ) sufficiently high values p (F221) = p (B1) × p (B2 / B1) × p (A11 / B3) and, in the assumed case of mutually exclusive content (B4 d B1) and (B4 d B5), p ( F231) = p (B4) × p (B5 / B4) + p (B4) × p (B1 / B4). On the other hand, a content F232 of the sense "... Of color B6 mixed from blue B1 and white B5" that is alternative to F231 a S23 is ruled out because of an insufficient reference p (B6B1) k ps as well as because of p (Vt32Vt23B6) k ps, because the use of language for such a color mixture B6 would prefer a word complex Vt3yVt23 = / _hellblau / with p (Vt3yVt23 / B6) g ps. For a content F222 a S22 that also seems possible with q332 = s (d) analogous to F231, there is also a low probability p (F222 / S22) kk p (F221 / S22) due to the nature and language.

Compared to a training of the Invention in which word complexes such as Vt33Vt23 = / _graulaue / and  Vt32Vt23 = / _white-blue / as idioms, i.e. as additional words Vtxy a Gx of an additional one Unit Gx with relevant terms to be described Kxyu a Fpxyu a Vtxy, where Gx = G2 can be selected in Example 3) has the here exemplified formation of semantic content using the Transformations Qs sufficient symbols Vt the advantage that the Data Wp of the language L to be provided is a comparative one require little data and thus storage volume, but the Processing costs increased due to the additional complex formation. Accordingly, the invention provides for the use of the memory and idioms Adapt processing capacity of the available devices. Independently whether G3 is retained as a separate unit becomes complex Vt32Vt23 = / _white blue / advantageous then as a separate idiom symbol Vt2y a G2 in the data Wp is taken into account if this also causes K in the communication to be expressed, which is also represented by the symbol Vt2x = / _bayrische / is to be covered, this idiom formation also with the above Transformation condition Qsa a Q1 is in line.

Further details of the method by Mp using quantitative methods achievable linguistic treatment of certain lexical words such as the above-considered word / and / including symbols Vt, their semantic Notes Fp not only with regard to assignable semantic content K, but also also ambiguous regarding their links to complexes KK and content F. are based on an extension of Example 3) to include the words associated with unit G4 in addition to symbols Vt41 and Vt42 Vt43 = / _or green / and Vt44 = / _or blue / shown. The following are considered 2 others the one identified by G1 = Gk1 = Ar (G2) and G2 = Ar (G4) grammatical content N3 assigned sentences S33 = vt111Vt21Vt43vt112vt113 = / Ripe or green plums ./ and S34 = vt111Vt21Vt44vt112vt113 = / Ripe or blue plums. / The two words Vt4y, with y = 3 or 4, are each 2 disjoint semantic clues associated with Fp4y1 and Fp4y2, whose Instructions for use q simplified by q431 = q432 = o and q432 = q442 = s are given, but the latter link is a probability p (F) associated with the resulting content F significantly reducing factor qr kk 1 becomes. This note q4y2, like q412 = q422 for Vt41 and Vt42, requires that unification according to Fb = noted here in the above simplified spelling (G1 d G2) s (G1 d G4) of the two averages D1 and D2 of the corresponding ones Things Kxyu a Vtxy a Gx, for example in the sense of "or also" or "both as well". On the other hand, q4y1 requires the formation of an exclusive Fa content Alternatives according to Fa = (G1 d G2) o (G1 d G4) of these averages D1 and D2 in the sense of "either or". When using the Ba approach, please note the respective probabilities p (Fa) = p (D1) + p (D2) - 2p (D1 d D2) and p (Fb) = qr x (p (D1) + p (D2) - p (D1 d D2)). Based on sentences S33 and S34 is here of an average D1 in the sense of "ripe plums" and 2 averages D2 = D21 in the sense of "blue..." and D22 in the sense of going out "green plums". To simplify the explanation  p (D1) = p (D21) = p (D22) assumed, and further that the quantities D1 and D21 completely exclude each other, with p (D1 d D21) = 0, and that the sets D1 and D22 overlap completely, with p (D1 d D22) = p (D1) = p (D22). For set S33 This results in a possible content F33a in the sense of "either or" Probability p (F33a) = p (D1) and for F33b in the sense of "both as also "a value p (F33b) = qr × p (D1) lower by the introduced factor qr Theorem 34 follows for the corresponding alternative statement p (F34a) = 0, so that with p (F34b) = qr × p (D1), here only one, despite the reducing factor qr exaggerated by "ripe or just as well blue plums" describing, content F34b can be addressed.

The use of language observed in sentence S33, according to which the word / _or / preferably mutually exclusive content F is addressed in a further development of the invention takes into account that instead of the above introduced correction factor qr for the probabilities p (F) different Probabilities p (Vt / Fp) for the to an indication Fp and thus also a link q of content K to take into account the choice of words become. In addition to the considerations above, these references are made up of links from Subject terms K resultant probabilities p (F) according to the in Approach Ba to take into account product terms p (F) × p (Vt / F) also for that the assignments of contents F to word complexes made according to Mp VtVt and their texts T reflect the use of language.

According to the possibility also covered by the invention, that too Unusual or incorrectly written texts T are to be edited, which, as well as from here, if necessary, by translating or typing generating texts T, content F only with a low probability p (F / T) display, the corresponding messages TNF also from components Em can be constructed, the content F with comparatively low, below Probabilities ps are assigned to threshold values ps Process Mp in one embodiment, such texts T and corresponding sentences S. the communication partners Pa as well as benefits if necessary to display corresponding markings defined by specifications Vr. In a Further training will also include ambiguous texts T, for their most likely content F probabilities p (F / T) apply that are around distinguish correspondingly small values ps using known ones Methods marked accordingly.

The invention provides to use data Wp in which the data to be used Components Em are listed according to certain criteria, so that they with appropriately adapted search strategies based on available data enough effort can be found. Such criteria also apply the addresses ad. Addresses Ad are preferably used, which their components Em with sufficiently high probabilities p (Em / Ad) g pr Show above values pr, whereby components Em as addresses Ad  serve for further components Em, and one component Em also is listed several times after several addresses Ad, while on the other hand not it is excluded that an address A ambiguously means several Components can be addressed. To those listed in Wp Components Em include the symbols Vt to be used according to method Mp with their standard texts Tsa and assigned sizes Fp, K and q as well as the Symbols Gr and G. The address Ad for a word Vt is used in detail in addition to its standard text Tsa also a complex Kp with text properties H (T) Tolerances dH, which uses the texts T5 a Vt of this word Vt, in a Training also includes a sufficiently high proportion of this. In one too the grammatical procedure M specified training will be such Text properties H (T) also as complexes Kp further, above as characters Z introduced text symbols Z = H (T), as shown for example in Written languages by letters or parts thereof and in oral Texts T are given by sounds or combinations of sounds, with a Symbol Vt is also noted as a complex Vt = Kp1 = ZZ of such signs Z and such a complex Kp1 as well as parts Kp2 a Kp1 thereof as addresses Ad be used. In a further training, the above are also mentioned Subwords vt a Vt as well as character complexes assigned to them ZZ a vt as Addresses Ad for such words Vt as well as subunits g and units G used. Especially in the case of method Mp in the role of author At delivering facts Wk2 through texts T are also Technical terms K as addresses Ad for the symbols Vt used.

Method Mp includes the application of known measures for inclusion, Storage and output of the constituent of the communication texts T. Speech signals and those to be transmitted in the role of author At or Listener Hr determined contents F and facts Wk in correspondingly coded Shape with a.

Method Mp is described in the following, taking into account the above Details of process parts Mp1 to Mp4 to be carried out: The at the assignment in the role of the listener Hr semantic messages TNF for incoming texts T including the Determination and transfer of the data through the communication partner Pa Submitted facts Wk is part of a procedural part Mp1 carried out. The transmission to be performed in the role of the author At predefined facts Wk to communication partner Pa through semantic Texts T to be output associated with TNF are also used of process part M1, as part of a process part Mp2. The at the Communication, if necessary, appropriate correction incoming texts T in correspondingly improved texts T to be passed on Language L is used within the framework of one that uses parts Mp1 and Mp2, Process part Mp3 carried out. Process part Mp4 includes the translation  and interpreting using process parts Mp1, Mp2 and possibly also Mp3 transmission of incoming texts T one Source language L in texts T 'to be output including a target language L' the transcription from oral to written that occurs when typing as well as the transmission of written in oral texts T.

Process part Mp1 a Mp for determining an incoming through one Communication partner Pa wrote, text T transmitted facts Wk2 through a handset Hr includes several of the following measures: the incoming Speech signals are recorded and saved. Text features H (T) that as Addresses displaying ad symbols Gs of languages L are identified. The Language L of text T is identified. The addresses Ad become disjoint Repertoires Ra (Gs) of symbols Gs to be used as components (Em) with taking conditions taking into account the positions P of the addresses Ad Ve assigned. The text T is a correspondingly disjoint repertoire Rb (NN) of complexes ordered according to these positions P Kpx = NN each assigned one or more grammatical contents N. The Content N is assigned in accordance with method M. At Continuous texts T are used for the formation of content N each Text sections T are taken into account, the minimum length of which is given by the fact that at least 2 repertoires Ra (Gs) are indicated by their addresses Ad, whose symbols Gs are suitable as head symbols Vtk a Grk. The formation of the Contents N a Kpx takes place in the order of the incoming text sections T. Ein or possibly disjoint several complexes Kpx, whose messages N sufficiently high probabilities p (N / T) become one semantic processing. For this purpose, a regulation Vr fixed repertoire Rc (Wk1v) of one or more at partner Pa pragmatic facts to be assumed Wk1v are taken into account. Permutations of semantic references Fp a Vt a N of symbols Vt each of a content N become a disjoint repertoire Rd (Fu) of one or more semantic content Fu, through which the facts Wk1v in a correspondingly disjoint repertoire Rd (Wk2w) of facts Wk2w be modified. Components Em a Ra are taken into account which correspond to the forming complexes Kp = TNF with sufficiently high probabilities p (Em) g ps can be assigned above threshold values ps. The assignment semantic content F to the grammatical content N of a complex Kpx and the change to be made as a result of presupposed circumstances Wk1v in Facts Wk2w takes place in the Kpx given by this complex Sequence. The resulting repertoire Rd (Wk2w) becomes one, or if necessary also disjointly selected several facts Wk2w by the semantic contents F of a complex Kpx were generated, its grammatical messages TN with the highest, predetermined threshold values ps not below, maximum values pr, however, also not exceeding Probabilities p (F / TN; Wk1v) are assigned. The resulting  Issues Wk2 are in accordance with the requirements of Vr and the criteria given thereby checked for their usability. You will be one of Vr specified further use. The facts Wk2 are in a clearly coded form that can be used by users to such users passed on. Mechanical devices are used as users by Commands corresponding to the facts Wk2 are operated and in operation be set. Electron computers are assigned as users, which are controlled by the Issues Wk2 given, commands of the intended Programming languages are put into operation.

The one to be carried out in process part Mp1 according to specifications Vr Checking the usability of incoming texts T and if necessary assignable facts Wk2 includes a check on their Usability as well as the triggering of any other necessary Actions with a. These actions are performed using the procedural part Mp2 by outputting this action content Communication texts T or also clearly reproducing these contents Signals executed. Such actions include the announcement of a any incomprehensibility of a text T to its author At, for example also the responsible text section T. It will Errors identified and made known to partner Pa, if that is clear from his Messages F conveyed probabilities p (K) a Wk2 from as objective correct, especially given values given by p (K) = 1 or p (K) = 0 differentiate. Insincerities are noted and, if necessary, in Actions implemented if the listener Hr is known Wp based on the data available, that deviations of the probabilities p (K) in transmitted facts Wk2 from the p (K) values as objectively correctly declared facts Wk dem Author of the message F and WW2 are known. By texts T of a partner Pa transmitted facts WW2, which contain requests or questions, or which is about sufficiently high probability values p (K) of differentiate between the specified setpoints p (K), in, accordingly communicating communication texts T or factual data K expressed actions implemented.

Process part Mp2 a Mp for transmitting a as a complex Kp of Terms K given facts Wk to a communication partner Pa by a text T to be output by an author At comprises several of the following measures: It is the one to be assumed by partner Pa Situation Wk1 deviating share of the situation Wk determined and by a disjoint repertoire R1 (Kpx) of complexes Kpx of by the data Wp given to the symbols Vt assigned terms K a Fp a Vt and others, for example formed as generic terms K thereof Subject K is shown. Considering the procedural part Mp1 of Mp the symbols Vt a Kpx each become a complex Kpx, a disjoint Repertoire R2 (Kpy) forming, complex Kpy of grammatical messages  TN merged to form semantic messages TNF with semantic Contents F are added. Of these, a text T = T1 of a complex Kp1 a R2 issued by TNF messages to convey the facts Wk, the contents F according to process part Mp1 with sufficiently high Probabilities p (F / TN; Wk1) and sufficient unambiguity Wk2 result, which reproduce the facts Wk with sufficient accuracy.

Process Mp provides, depending on available data and on specifications Vr through the communication tasks to be solved according to the invention Use of parts of one or both process parts Mp1 and Mp2 perform, with details of the invention given above accordingly are taken into account.

An improvement to be made according to Vr specifications by one Communication partner Pa drafted incoming text T = T1 with respect to grammatical content N and semantic content F to be transmitted by it is carried out according to a process part Mp3 a Mp in that the text A transmitted situation Wk = Wk2 is assigned to T1 in accordance with method part Mp1 is, according to the procedural part Mp2 a communicating this fact Wk2 Text T2 is assigned, by means of which the parts T3 of Text T1 to be replaced. In addition to correcting spelling errors, Punctuation and grammar, such as those used in inflections and positioning natural words are noticeable and are often found in authors At, the one Not adequately proficient in language L, process part Mp3 also includes content-preserving stylistic corrections, such as those made by Avoid repetition of words, ambiguous or misleading Words or overly nested or unnecessarily short natural ones Sentences are possible. Looks according to the intended application Process part Mp3 with corresponding specifications Vr also the clarification given texts T regarding their messages F and WW2 as they pass through On the basis of modified, for example less explicitly presupposed Initial facts Wk1 and correspondingly more concrete to express Messages F is possible according to the invention.

Process part Mp4 for the transmission of an incoming text (T) of a Source language L in a related to the matter transmitted Wk2 sufficiently equivalent text T 'one defined by specifications Vr Target language (L '), which by definition is written translations of written or oral texts, interpreting, stenotyping Oral to written transmissions and speech generators including conversions in writing from written to oral several of the following measures: Using process part Mp1 from Mp the incoming text T becomes a complex Kp1 = TNF, TNF one or several semantic messages assigned to TNF, by their semantic Contents F a fact Wk11 to be presupposed by the author At of this text T.  is modified in the facts Wk2. In the event of ambiguity, the Text T also a disjoint repertoire R (Wk2) of such facts Wk2 with a corresponding repertoire R (Kp1x) of complexes Kp1x of such messages Assigned to TNF. Using the process part Mp2 of Mp, that of Situation (Wk11) deviating part of the respective situation Wk2 as disjoint repertoire R1 (Kpx) of complexes Kpx from data Wp 'der Target language L 'assigned subject K' reproduced. Symbols Vt 'a Wp' a L ', to which these technical terms K' a Fp 'a Vt' via their semantic references Fp ' are assigned to a disjoint repertoire R2 (Kpy) of complexes Kpy each supplemented by one or more semantic messages T'N'F '. As the transmission 1 'of the text T, a text T' becomes one of these Message complexes Kpy = T'N'F ', T'N'F' one by the specifications Vr specified further use supplied, the semantic content F 'according Process part Mp1 with a sufficiently high probability p (F '/ T'N'; Wk11 ') and uniqueness to be presupposed by the listener Hr of this text T ' Change the situation (Wk11 ') into a situation (Wk2') that is sufficient Accuracy corresponds to the situation Wk2. In the case of an ambiguous one Text T accordingly becomes a plurality of texts T 'to be used disjointly or one Text T 'output, according to which several issues Wk2' disjointly are assignable. Texts T 'to be used as a transmission are transmitted by corresponding markings for the listener Hr and possibly also for the user characterized.

In a preferred embodiment, process part Mp4 using the grammatical procedure M performed. Here the grammatical Messages TN assigned to the respective complex Kp1 as symbols Vt objective words Vt completely as complexes Kp11 of vocabulary Vqq assigned symbol complexes Vq = VtVt, each one disjoint class K1 = (Vq ') of, these with probabilities p (Vq' / Vq) assigned complexes Vq '= Vt'Vt' of objective words Vt 'of the target language Show L '. To determine the vocabulary Vqq in the data Wst the data Wp the words Vt a Kp1 are used as addresses Ad. The words Vt one Complex Kp1 are completely represented as complexes Vq a Vqq. The words Vt 'a Vq' a Vqq of the vocabulary Vqq assigned to a complex Kp1 according to method M, taking into account the arrangements Pr to a corresponding disjoint repertoire R (Kp2) of complexes Kp2 of one or more each grammatical messages T'N 'put together. The news TN and T'N ' become semantic messages using method part Mp1 TNF and a correspondingly disjoint repertoire R (Kp2) of complexes Kp2 supplemented by semantic messages T'N'F '. There will be a text T 'a Kp2 one of the Complex Kp2 as a transmission T 'of the text T and the specified one fed further use, the semantic content F 'one at the listener Hr of this text T 'to be presumed facts Wk11' with sufficiently high Probability p (F '/ T'N'; Wk11 ') and uniqueness in the above Modify facts (Wk2 '), which with sufficient accuracy the by text  T transmitted facts Wk2 reproduces. The accuracy of match is determined by tolerances dp for the things K and K 'in Wk2 and Wk2' future probabilities p (K) and p (K ') are determined. If there is ambiguity of the text T are correspondingly disjoint texts T 'or several Issues Wk2 'issued.

According to the P.. , Process part M3 of the grammatical communication method M described also provides for process part Mp4 to carry out transmissions from a source language L into a target language L 'via one or more intermediate languages L ", L"", etc. For example, in such a training the transmission of Texts T of the oral German language L in texts T 'of the English written language L' via texts T "of the German written language L"; the same applies to the languages L. of other language communities. Furthermore, as in M3 a M, transmissions from any language L, L 'etc. of a repertoire R (L) into any other language L' a R (L) also via a language L "of this repertoire R (L) called central language L" a R (L) In further training, process part Mp4 also provides for the use of process part Mp3, according to which an incoming incorrect text T = T1 of an output language L according to Mp3 is next improved into a text T2 from L, which is then transferred into a text T 'from L'. In the transmissions from oral to written and vice versa to be carried out in accordance with method part Mp4, vocabulary Vqq is used according to the invention, in which a word Vq '= Vt' of the source language L is generally assigned to a word Vq '= Vt' of the target language L ', the different from which Text-dependent spelling or pronunciation in the data Wst a Wp is advantageously represented with the help of the operators o mentioned.

In a further embodiment of method part Mp4, also according to existing specifications Vr, also takes into account that the Communication partners Pa from different language communities L and L ' possibly also differing from each other, as noted Wk1 'ne Wk1 Issues Wk1 or Wk1 'are to be assumed, whereby accordingly also different probabilities p (Vt '/ K1) ne p (Vt / K1) for the Using symbols Vt and Vt 'indicating the same thing K1 are to be applied. Accordingly, in this training the texts T 'der Target language L 'deviating from the initial facts Wk1 from L. Issues Wk1 'on the basis that in both languages L and L' as allow sufficiently identical issues Wk2 and Wk2 'to be achieved.

In one embodiment of the invention, process part Mp3 for improvement incoming texts also carried out as a special case of procedural part Mp4, the transmission from a source language L into a target language L ', which is equal to the source language L '= L.  

As can be seen from the description, process part Mp4 a Mp includes the Application of further, possibly also all, in method Mp described process parts and measures. Some of this Solving communication tasks are based on an example 4) for Transmission of texts T from the oral German language L ins Written English L 'explained. There is a text T1, the sound sequence here again is approximated by a string of letters T1 = / ainerwaissesget /. Under Applying part M2 of the grammatical method M to this text T1 a disjoint repertoire Ra (Kpx) of complexes Kpx of one or several grammatical messages TNxy with assigned sentences Sxy a Tnxy assigned, some of which here by texts Tx "= Sx" or Tx "= Sxy "Sxy" from the German written language L "are displayed: T1" = / Eine knows, it goes./, T2 "= / A proof: It goes./, T3" = / One: "Weißes geht!" /, T4 "= One: "Does Geet know?" These texts Tx "are also by application of the Process part M3a M from M for grammatical transmission of texts T is generated from the oral German language L into the written L ", whereby / Geet / in T4 "indicates a proper name, provided that a prerequisite Situation Wk11 a corresponding technical term K with sufficient Probability p allows. According to procedural part Mp1 a Mp is a matter of fact Wk12 presuppose, after which, simplified, "a discussion between Business people about the sales opportunities of different colored porcelain types according to Mp4, it follows that from the variety of possible translations the Texts Tx "a translation such as T3 '= / One of them:" White china sells well. "/ is preferable, taking into account the context Wk12 already acts as a restriction of the content N of the given text T1.

Example 4) is expanded by a further oral text T2 = / disertonklingthel /, whose possible messages F according to M3 a M can be found in the grammatically clear, semantically but ambiguously assignable text T5 "= / This tone sounds bright ./. According to process part Mp4 Here a fact to be assumed Wk12 is primarily taken into account when selecting the word complex Vq '= Vt'Vt' of the vocabulary Vqq1 indicated by Vq1 = / g the sound /. So, according to Wk12, with sufficiently high probabilities p, it is roughly "out of sound burned vessels, which you knock on to infer the quality and integrity of the sound ", the translation follows, for example, 15 '= / The earthenware sounds clear./.

The consideration provided according to the invention is different Initial issues Wk11 and Wk11 'with communication partners Pa different languages L, which also according to their origin Use different knowledge, is in example 4) based on the further, here German text T6 = / Tomorrow we will visit Hohenschwangau./ illustrated, the Bavarian listener Hr announces a "castle tour", foreign tourists Hr but often only with sufficient security if he is in T5 '= / Tomorrow we are going to visit Hohenschwangau castle./  is translated. Details of the symbols Vt, G, used in Example 4) Vqq, their arrangements Pr as well as references Fp and things K follow from the above information provided using examples.

On the advantages that result from the application of the Process parts Mp1 to Mp4 result in the above-mentioned configurations each in connection with the descriptions of this measures to be used.

In a further embodiment, the invention provides for the method Mp for semantic communication in natural languages L to be carried out with a computer-aided device 1 , which comprises the following units, which are supplied by one or more power supply systems and each consist of one or more subunits: one according to the method Mp programmed central processing unit 11 for handling the TNF and associated semantic content F a TNF and pragmatic facts Wk for the creation of semantic messages TNF as well as further data arising here as well as operations provided for the process organization. A storage unit 12 for storing the data Wp to be used for this in the languages L concerned, the facts Wk to be taken into account, the intermediate and final results that occur and incoming and output data D including texts T and specifications Vr. A unit 13 for receiving incoming and outgoing data D including texts T, any markings assigned to them, as well as messages TN as well as TNF and facts Wk in the encodings and media provided for this purpose, the data exchange being communication partners Pa and, optionally also machine, Includes users. An operating unit 14 for defining and monitoring operating states, for monitoring intermediate and final results, for entering specifications Vr as well as data Wp including their addition and modification in the storage unit 12 . These units 11 to 14 are connected by data buses 15 which have bandwidths adapted to the tasks.

In various configurations, 12 RAM memories (with "read only memory") are used for the memory unit. RAM memory modules are used to store the symbols Gr, Vt assigned to the data Wp, semantic information Fp, technical terms including their properties H (Gr) etc. and addresses Ad. Memories 12 are used which allow the data Wp to be supplemented and corrected. Computing units 11 and memories 12 are used, which are constructed from customer-specific integrated circuits adapted to the data Wp and the method Mp. A device 1 is used which consists of a central unit 111 , which comprises a central computer 11 and a memory 12 , and a number of user units 112 arranged separately, preferably also mobile, each of which has an operating unit 14 and a receiving and Output unit 13 include, wherein the user units 112 are linked to the central unit 111 by, possibly also wireless, working data buses 15 . This design has the advantage that the effort associated with the central unit 111 benefits a larger number of users on the comparatively low-effort units 112 , for whose connections with the central unit 111 data buses 15 with comparatively low bandwidths are sufficient, as they are when receiving and outputting of texts T as well as during operation.

In a further embodiment, the central processing unit 11 and the storage unit 12 are constructed using parallel computers, in which the symbols Vt a Gr and stuff K a Fp a Vt are assigned computing modules 1111 which are linked in parallel in accordance with the references given in the data Wp are.

Patent / laid-open documents

DE 30 42 408 C2
DE 31 42 540 C2
DE 35 03 233 C2
DE 36 15 972 A1
DE 36 16 011 A1
DE 30 32 664 C2
DE 31 51 106 A1
DE 32 11 313 C2
DE 37 33 391 C2
DE 35 14 286 A1
DE 37 10 507 A1
DE 37 11 348 A1
DE 41 35 261 C1
DE 42 32 482 A1
DE 43 11 211 A1.

bibliography

// 1 // John Lyons, The Language, Beck, Munich (1992)
// 2 // Hermann Helbig, Artificial Intelligence and Automatic Knowledge Processing, Technology, Berlin (1991)
// 3 // John Lyons, Introduction to Modern Linguistics, Beck, Munich (1971)
// 4 // Dieter E. Zimmer, Payment may he made - Die Zeit, No. 15 (April 5, 1996)
// 5 // Helmar Frank, Cybernetic Foundations of Pedagogy, Agis, Baden-Baden (1996)
// 6 // Terry Winograd, Fernando Flores. On "Understanding Computers and Cognition": A New Foundation for Design, Artificial Intelligence, Vol. 31, p. 250 (1987)
// 7 // Edward O. Wilson, Sociobiology, Cambridge, Mass. and London (1975)
// 8 // Steven Pinker, The Language Instinct, Morrow, New York (1994)
// 9 // New Scientist, p. 53 (Oct. 22, 1987)
// 10 // Marcus Spies, Out of the Head in Computer Speech Recognition in Electronic Dictation Systems, Spectrum of Science, Dossier 4/97, p. 47 (1997)
// 11 // Wolfgang Wahlster, Verbmobil: recognition, analysis, transfer, generation and synthesis of spontaneous speech, spectrum of science, dossier 4/97, p. 52 (1997)
// 12 // Hadumod Bußmann, Lexicon of Linguistics, Stuttgart (1990)
// 13 // Robert Jourdain, Music, the Brain, and Ecstasy, New York, p. 279 (1997).

Claims (30)

1. A method for communication relating to semantic content and communicating pragmatic facts in one as well as several natural languages, characterized in that

• 1.1.1. The communication between communication partners (Pa) is carried out in accordance with the existing requirements (Vr) according to a procedure (Mp), according to which pragmatic facts (W) formed as complexes (Kp = KK) of technical terms (K) are used by authors (At) Speech signals of the media to be formed for the media intended for the respective natural languages (L) or to be evaluated by their recipients (Hr), texts (T) are transmitted, which are assigned to grammatical by assigning grammatical content (N) taking into account the probabilities (p (TN)) Messages (TN) and semantic content (F), taking into account the probabilities (p (TNF)), can be supplemented to semantic messages (TNF), and the facts to be transmitted (Wk = Wk2) by changing the requirements of the communication partners (Pa) , pragmatic facts (Wk = Wk1) are generated according to the semantic content (F), whereby
• 2. 1.2. the semantic messages (TNF) as complexes (Kp = EmEm), taking into account their probabilities (p (Em)), which may also be due to further components (Em), are combined by components (Em) described by data (Wp), which Include technical terms (K) and symbols (Gs) adapted to natural languages (L), which, in each case to a disjoint class (K1 = (Vt)) of text sections (T = Ts), are determined in more detail by tolerances (dH). Text features (H (Ts)) superlated, text symbols (Vt = (Ts)) and, from this disjoint classes (K1 = Gr) superlated, grammatical symbols (Gr = (Vt)), each comprising a repertoire (R (Pr) ) of, more precisely determined by probabilities (p), the possibilities of their arrangements, determined by rank orders (Ra) and the positions (P) of the text sections (Ts) of the text symbols (Vt) assigned to these symbols (Gr) in texts (T) ( Pr) pretend
• 3. 1.3. the text symbols (Vt) each give a disjoint class (K1 = (Fp)) of semantic references (Fp) that take into account the possibilities of their arrangements (Pr), each of which contains a technical term (K) and a repertoire (R (K1)) of disjoint classes (K1 = (q)) of usage instructions (q) for their assignment to semantic messages (TNF) including their linkage with technical terms (K) and further text symbols (Vt) in the relevant semantic message (TNF) as well as with technical terms ( K) include presupposed pragmatic facts (WW1) including their modification, and whereby
• 4. 1.4. pragmatic facts (Wk2) to be transmitted by texts (T) taking into account existing data (D) including specifications (Vr) are each generated by one or more grammatical symbols (Gr) displayed thereby and a permutation of the text symbols (Gt) assigned to them taking into account the arrangements (Pr), they can be combined to form a grammatical content (N), which may be arranged in a position (P), possibly also discontinuously, at a position (P) assigned by a permutation of these symbols (Gt) and given by these arrangements (Pr) , Text sections (Ts) merged text (T) is supplemented to a grammatical message (TN), and that the technical terms (K) a permutation of these text symbols (Gt) assigned semantic references (Fp) according to a permutation of these references (Fp) assigned instructions for use (q) taking into account the probability given by a pragmatic fact (Wk1) en (p (K) 1) as well as other probabilities (p (Em)) present in the data (Wp) as a complex (Kp = KK) of these technical terms (K) with probabilities (p (K) 2) combined, semantic content (F) is formed, through which the grammatical message (TN) is supplemented to a semantic message (TNF) and the pragmatic situation (Wk1) is changed to the pragmatic situation (Wk2) to be transmitted, one of which is changed by the Requirements (Vr) given further use.

2. The method according to claim 1, characterized in that for a text (T) from a disjoint repertoire (R (Wk1x)) of assumed facts (Wk1x) Situation (Wk11) is selected with which the data (Wp) is observed semantic content (F) results with a given grammatical message (TN) the highest probability not to fall below a threshold value (ps) (p (TNF)) can be assigned. 3. The method according to claim 1 and 2, characterized in that for communication a repertoire (R (Wk1x)) of facts (Wk1x) with technical terms (K) and Probabilities (p (K) 1) is assumed, which the possibilities of Differences of opinion as well as knowledge between the communication partners (Pa) considered. 4. The method according to one or more of claims 1 to 3, characterized in that from a given pragmatic situation (Wk1) a text (T) assigned disjoint repertoire (R (Wk2x) of pragmatic issues (Wk2x) a fact (Wk21) is used as a fact to be transmitted, the based on this text (T) technical terms (K) with probabilities (p (K) 2) are to be assigned, which differ from the facts (Wk1) or data (Wp) given probabilities (p (K) 1) differ by amounts (p) which lie between the minimum values (p1) and maximum values (p2) to be determined. 5. The method according to claim 1, characterized in that the communication under Using text symbols (Vt) indicated by texts (Ts), which are adapted to a natural language (L) in that the texts (Ts) the statistically evaluated results of transformations (Qs) on texts (T) of these Language (L) is sufficient, after which an observable invariance (I) is observed  with regard to its semantic content (F1) as correctly assessed text (T1 = Tf1) Shortened by the shortest possible, one or more, discontinuously arranged text sections (T) comprising a text part (T11) with respect to the modified semantic content (F = F2) also as correctly assessed text (T2 = TV) is changed, this text part (T11 = Ts) being assigned to a symbol (Vt) a, a technical term (K) and a usage note (q) for this comprehensive, semantic note (Fp) is assigned by which the semantic Content (F2) is changed to the original content (F1), and being this Transformation Qs as a necessary condition that the relevant symbol (Vt) with subordinate rank (Ra) by a further text part (Ts') of the text (T2) displayed further symbol (Vt ') is assigned, as well as a sufficient condition this means that none is indicated by another text part (Ts ") of the text (T1) there is another symbol (Vt "), which this symbol (Vt) with a parent Rank (Ra) is assigned. 6. The method according to claim 1 and 5, characterized in that the communication using, each to a disjoint class (K1) of symbols (Vt) superated, grammatical symbols (Gr = Grx) is carried out, which the Results of the transformations (Qs) that are applied to texts (T1 ') are sufficient, alternatively each as a text section (T11) a text section (Tsxyz a Vtxy) different symbols (Vtxy a Grx) of a grammatical symbol (Grx). 7. The method according to one or more of claims 1 to 6, characterized in that the complex formation to be carried out taking into account the probability (p (TNF)) of semantic messages (TNF) from components (Em), which according to, as evidence existing data (D) with addresses (Ad) as well as specifications (Vr) with sufficient high probabilities (p (Em)) in the form of disjoint repertoires (R (Em)) are taking into account the data (Wp) available Probabilities (p (Em)), according to the Bayesian approach (Ba), after which the disjunctions occurring in the superation stages to be run through Complexes (Kp = EmEm) as each assigned to a disjoint class (K1 = (Ks)) Competitive hypotheses (Ks) according to their data-related (D) Probabilities (p (Ks)) are taken into account, whereby in the hypotheses (Ks) Components (Em) already taken into account also as addresses (Ad) for further, if necessary, components (Em) to be used disjointly can also be used. 8. The method according to one or more of claims 1 to 7, characterized in that those involved in the formation of semantic messages (TNF), also as Interim results, competing hypotheses (Ks) taking into account Probabilities (p (Ks)) are taken into account, which using non-linear, depending on the probability (p) Processing operations (Bb) are determined, after which with probabilities (p (Em)) components to be used, disjoint repertoires (R (Em)) (Em) and from components (Em) with resulting probabilities (p (Ks)) assembled competition hypotheses (Ks), their probabilities (p) each below the limit values (ps) to be determined, taking into account the limits for each disjoint repertoire (R) of components (Em) or hypotheses (Ks) applicable sums of probability (Ca) are neglected, with threshold values  (ps) so adapted to the communication files to be performed and available devices that permissible processing times and error rates are not exceeded and at least one competitive hypothesis (Ks) from a repertoire (R) is used. 9. The method according to one or more of claims 1 to 8, characterized in that complex formation of semantic messages (TNF) using as disjoint repertoire (R (Em)) of components to be used (Em) Terms (K) using non-linear processing steps (Bbr) is carried out, after which components (Em = Em1), the one message (TNF) with a probability that surpasses a nearby threshold (ps3) (p (Em1)) can be assigned to others belonging to the repertoire (R (Em)) Components (Em) neglected with probabilities (p (Em)) below (ps3) become. 10. The method according to one or more of claims 1 to 9, characterized in that the communication is carried out with terms (K) for their assignment to Texts (T) and semantic messages (TNF) assignment probabilities (p (K)) to text symbols (Vt) and other terms (K) which the Possibility of using subjective as well as faulty ones obtained from objective ones Data (p (K)) of deviating values by the communication partner (Pa) with lock in. 11. The method according to one or more of claims 1 to 10, characterized in that that communication is assigned using text symbols (Vt) semantic information (Fp) is carried out, each as a set of Terms (A) formed technical term (K) and a repertoire (R) of disjoint classes (K1) of instructions for use (q) for it, taking into account the instructions (Pr) of the grammatical symbols (Gr) to which the symbols (Vt) belong Include links with other technical terms (K) to complexes (Kp = KK), which with (q = d) averaging and (q = s) the union of two Terms (K = K1) and (K2), with (q = m) a distance according to (K3 m K4) trailing term K4 from another term K3, with (q = f) according to (K5 f K6) the replacement of a term (K5) by a negation (K6 = nK5) of this term (K5), with (q = c) according to (K7 c K8) the replacement of a term (K7) by one in or several properties (H (K)) of the term (K8), and with (q = o) according to (K14 = K12 o K13) provide for the formation of a disjunctive technical term (K14), which is either the terms (A) of the technical term (K12) or the technical term (K13) includes. 12. The method according to one or more of claims 1 to 11, characterized in that the links of semantic hints (Fp) associated with terms (K) with other subject terms (K) assigned subject terms (K) according to Instructions for use (q) are given, which depend on the type of link Modifications of the complexes (Kp) of technical terms (K) resulting probabilities (p (Kp)) to the respective languages (L) Specify adjusted factors (qr).   13. The method according to one or more of claims 1 to 12, characterized in that the communication using described by the data (Wp) Subject terms (K) is carried out, each leading to a semantic note (Fp) of a symbol (Vt) assigned technical terms (K = K5) each a generic term (K11) contain the as a union (K11 = K5 s K6) of the technical term (K5) with a Technical term (K6) is formed, which represents a negation (K6 = nK5) of the term (K5), where generic terms (K11) are used, which the terms (K5) each with sufficiently high probabilities not less than a reference value (pr) (p (K5 / K11)) are assigned, and one according to the circumstances of the languages (L) according to (K5 f6 K6) negation (K6) of a symbol (Vt) assigned term (K5) by the distance (K6 = K11 m K5) of the term (K5) the generic term (K11) is made. 14. The method according to one or more of claims 1 to 13, characterized in that communication using grammatical messages (TN) is carried out, each consisting of a text (T) and this according to a grammatical communication method (M) assigned grammatical content (N) consist, which is formed as a complex (Kp) from components (Em), the text symbols (Vt) and grammatical symbols superposed to disjoint classes (Kl) (Gr = G) include which transformations determined in addition to the transformations (Qs) (Q) on texts (T) are sufficient, the results of which are determined by observations Invariances (I) on the texts (T) are detectable. 15. The method according to one or more of claims 1 to 14, characterized in that the formation of the semantic content (F) from the semantic information (Fp) of the Symbols (Vt) assigned subject terms (K1) taking into account, by which Data (Wp) described, other terms (K2) is carried out, which the Not understood (K1) with sufficiently high, predetermined threshold values (ps) undershot probabilities (p (K2 / K1)) are assigned, being orthogonal Technical terms (K) are used, after which further technical terms (K3) are defined become that according to (p (K2 / K1; K3) = p (K2 / K1)) the dependence of a technical term (K2) is not influenced by a term (K1) by the terms (K3). 16. The method according to one or more of claims 1 to 15, characterized in that communication using a repertoire (R (K)) of terms (K) is carried out, their mutual references by conditional probabilities (p (K)) are described, the order (O (n)) of a given one Value adapted to the performance of available devices (n = n1) exceeds. 17. The method according to one or more of claims 1 to 16, characterized in that that the structure of semantic messages to be assigned to incoming texts (T) (TNF) from components (Em) in compliance with the instructions for use (q) in one Order according to falling, based on the available evidence (D) and the data (Wp) including the components already used as addresses (ad) (Em) conditional probabilities (p (Em / D; Wp)) of these components (Em) is carried out.   18. The method according to one or more of claims 1 to 17, characterized in that complexes (Kp = VtVt = Vt1) of put together in arrangements (Pr) Text symbols (Vt) with these complexes according to their semantic references (Fp) (Vt1) assigned complexes (Kp = KK = K1) of terms (K), which in semantic messages (TNF) occur with sufficient frequency as idioms (Vt1) with a semantic hint (Fp1) assigned to it and this assigned technical term (K1) can be used. 19. The method according to one or more of claims 1 to 18, characterized in that specifications (Vr) are taken into account, according to which texts (T) that do not have semantic Messages (TNF) can be assigned where the probabilities (p (Em)) all components (Em) and (p (F)) of the semantic content (F) therefor specified threshold values (ps) exceed, the communication partners (Pa) such if necessary, users of the method (Mp) by appropriate markings are displayed. 20. The method according to one or more of claims 1 to 19, characterized in that to be carried out by a recipient (Hr) of an incoming text (T) Determination of a fact (Wk2) transmitted by its author (At) according to a process part (Mp1) of process (Mp) is carried out, after which the at Data (D) including the incoming text (T) as well as the specifications (Vr) are recorded and saved by in it identifying addresses (ad) displayed repertoires (ra (em)) of for the Complex formation of semantic messages (TNF) to be considered Components (Em) with removal conditions (Ve) can be determined, thereby given, grammatical symbols (Gr) assigned text symbols (Vt) under consideration of the probabilities (p (N / T)) for a correspondingly disjoint repertoire (Rb) of Complexes (Kpx) of one or more grammatical messages (TN) each grammatical content (N) are combined, after which the technical terms continue (K a Fp a Vt) of permutations of the symbols (Vt a N a Kpx) of each one Complex (Kpx) assigned semantic clues (Fp a Vt) according to their Instructions for use (q a Fp) considering a disjoint repertoire (Rc) one or several circumstances (Wk1v) required by the partner (Pa) for one disjoint repertoire (Rd) of complexes (Kpz) of complexes (Kpx a Rb) assigned semantic messages (TNF) with semantic content (F) be put together, and after which a situation (Wk11 a Wk1v), in which for the semantic contents (F a TNF a Kp1) of a complex (Kp1 a Kpz) the highest, a predetermined maximum value (pr) not exceeding probabilities (p (F / TN; Wk11)) result from this semantic content (F a Kp1) under consideration the order given by this complex (Kp1) in the order given by the text (T) transmitted facts (Wk2) is modified, one by specifications (Vr) specified further use is supplied. 21. The method according to one or more of claims 1 to 20, characterized in that in the role of an author (At) by a text (T) to be carried out as a complex (Kp) of terms (K) Facts (Wk) to a communication partner (Pa) according to a procedural part (Mp2) is carried out by method (Mp), after which that of one at the partner (Pa)  fact to be assumed (Wk1) deviating part of the facts (Wk) by a disjoint repertoire (R1 (Kpx)) of complexes (Kpx) from the data (Wp) associated terms (K) is shown and text symbols (Vt) to which these Subject terms (K a Fp a Vt) via their semantic references (Fp) are assigned to a disjoint repertoire (R2 (Kpy)) of each complex (Kpx) corresponding complexes (Kpy) each of one or more grammatical Messages (TN) are merged by assignment under consideration the notes (Fp) formed semantic content (F) for semantic messages (TNF) are added, of which a text (T1 a Kp1) of a complex (Kp1 a R2) the partner (Pa) is output, the one to be determined in accordance with method part (Mp1) Content (F) with sufficiently high probabilities (p (F / TN; Wk1) and sufficient clarity the assumed facts Wk1 in a fact Modify Wk2, which with sufficient accuracy with the facts Wk matches. 22. The method according to one or more of claims 1 to 21, characterized in that an improvement to be made in communication according to specifications (Vr) one written by a communication partner (Pa) in one language (L) incoming text (T = T1) regarding its transmission of grammatical content (N) and semantic content (F) according to a method part (Mp3) of method (Mp) is carried out, according to which the text (T1) according to method part (Mp1) is a transmitted Issue (Wk = Wk2) is assigned to the one according to the procedural part (Mp2) Facts (Wk2) transmitting text (T2) is assigned, and the thereof different parts of text (T1) are replaced by the corresponding parts of text (T2) become. 23. The method according to one or more of claims 1 to 22, characterized in that that the transmission of an incoming text (T) of a source language (L) in a with regard to the facts thus transmitted (WW2) are sufficiently equivalent Text to be output (T ') of a target language (L') defined by specifications (Vr) is carried out according to a process part (Mp4) of process (Mp), after which the text (T) according to process part (Mp1) of (Mp) a complex (Kp1) one or several semantic messages (TNF) is assigned by their semantic Contents (F) a fact to be assumed at the author (At) of this text (T) (Wk11) is modified in the facts (Wk2), and according to the procedural part (Mp2) of (Mp) using the data (Wp ') of the target language (L') of the facts (Wk11) deviating part as a disjoint repertoire (R1 (Kpx)) of complexes (Kpx) from the data (Wp ') associated terms (K') is represented, and symbols (Vt ') of the language (L'), to which these technical terms (K 'a Fp' a Vt ') about their semantic Notes (Fp ') are assigned to a disjoint repertoire (R2 (Kpy)) of Complexes (Kpy) of one or more semantic messages (T'N'F ') can be added, of which a text (T1 'a Kpy) as a transfer of the text (T) one further use specified by the specifications (Vr) is supplied Contents (F ') according to the process part (Mp1) with a sufficiently high probability p (F '/ T'N'; Wk11 ') and uniqueness one with the listener (Hr) of this text (1') Change the facts to be assumed (Wk11 ') into a fact (Wk2') which agrees with the facts (Wk2) with sufficient accuracy.   24. The method according to one or more of claims 1 to 23, characterized in that the transmission to be carried out in accordance with process part (Mp4) of an incoming Text (T) of a source language (L) into a text (T ') of a target language (L') below Application of the grammatical communication procedure (M) is carried out, after which the text (T) is a complex (Kp1) of one or more grammatical messages (TN) is assigned, whose text symbols (Vt) as complexes (Kp11) of, vocabulary (Vqq) associated, symbol complexes (Vq = VtVt) are shown, which each a disjoint class (K1 = Vq ') of, these with probabilities (p (Vq' / Vq)) assigned complexes (Vq '= Vt'Vt') of text symbols (Vt ') of the target language (L') Show which, taking into account the instructions (Pr) for a corresponding disjoint repertoire (R (Kp2)) of complexes (Kp2) of one or more each grammatical messages (T'N ') are combined, the messages (TN) and (T'N ') on semantic messages (TNF) and a corresponding repertoire (R (Kp2)) are supplemented by complexes (Kp2) semantic messages (T'N'F ') and a text (T 'a Kp2) of a complex (Kp2) as a transfer (T') of the text (T) is output, the semantic content (F ') of which the listener (Hr) of this text (T ') to be assumed (Wk11') with a sufficiently high probability (p (F '/ T'N'; Wk11 ')) and uniqueness in a situation (Wk2'), which with sufficient accuracy the facts conveyed by text (T) (WW2) reproduces. 25. Device for computer-aided implementation of the method according to claims 1 to 24, characterized in that a device ( 1 ) is used, which is composed of, each composed of one or more subunits, supplied by one or more power supply units, which one central processing unit ( 11 ) for processing the semantic messages (TNF), associated texts (T), semantic content (F), pragmatic facts (Wk) as well as other data (D) and the operations involved in the process organization, a storage unit ( 12 ) for storing the data (Wp) to be used for this purpose in the languages concerned (L) including facts to be taken into account (Wk), intermediate and final results as well as incoming and output data (D) including texts (T) and specifications (Vr ), a unit ( 13 ) for the exchange of incoming and outgoing data (D ) including texts (T) and, if necessary also to trigger predetermined operations of certain facts (Wk) with communication partners (Pa) and, including machine devices, users in the media and codes provided for this purpose, as well as an operating unit ( 14 ) for defining and monitoring Operating states and specifications (Vr), these units ( 11 ) to ( 14 ) being connected by data buses ( 15 ) with bandwidths adapted to the respective data rates. 26. Device according to claim 25, characterized in that RAM memory modules are used in the memory unit ( 12 ) for the data (Wp) including the symbols (Gs) and the semantic information (Fp) associated therewith. 27. The device according to claim 25, characterized in that a device ( 1 ) is used, the central processing unit ( 11 ) and memory unit ( 12 ) using the method (Mp) including the data (Wp) adapted customized integrated circuits are. 28. Device according to one or more of claims 25 to 27, characterized in that a central unit ( 111 ) comprising the central computer ( 11 ) and the memory unit ( 12 ) comprises a plurality of central units ( 111 ), optionally also spatially separated therefrom and separate from one another, each for different communication partners ( Pa) and users are assigned available user units ( 112 ), each of which comprises an operating unit ( 14 ) and a receiving and output unit ( 13 ) and are connected to the central unit ( 111 ) by data buses ( 15 ), which may also be implemented wirelessly The bandwidths of which are adapted to the data rates that occur when reading and outputting texts (T) including the data assigned to them (D) as well as during operation and use including the transmission of specifications (Vr). 29. Device according to one or more of claims 25 to 28, characterized in that the central processing unit ( 11 ) and storage unit ( 12 ) are constructed using parallel computers in which the symbols (Gs) to be used in semantic messages (TNF) as well as technical terms (K) arithmetic modules ( 1111 ) are assigned, which are linked in parallel according to their references specified in the data (Wp). 30. The method and device according to one or more of claims 1 to 29, characterized in that the method (Mp) is carried out using data (Wp) which is adapted to the language use of human communication partners (Pa) appearing as authors (At) Repertoire R (Vt) of symbols (Vt) as well as a limited repertoire (R (K)) of subject terms (K), which is adapted to the issues to be covered by the communication, including the triggering of predetermined operations, such that these issues by a According to the method (Mp) device ( 1 ) functioning as a receiver (Hr) as semantic content (F) to be assigned as incoming texts (T), which can be generated by linking these technical terms (K) and thereby certain facts (Wk2) which can be represented in the form of , clearly coded signals defined by specifications (Vr) of a further, triggering of predetermined operations in the case of machine-assigned users en devices including use.