RU2563148C2 - System and method for semantic search - Google Patents

Abstract

FIELD: physics, computer engineering.

SUBSTANCE: invention relates to search techniques. A method of facilitating search in corpuses for a computer system comprises preliminary semantic-syntactic analysis of a corpus with determination and storage in memory of values of linguistic parameters of sentences, including comparison of words of the sentence of a certain plurality of possible lexical meaning - each with a corresponding estimate of the probability of inclusion of said lexical meaning in the sentence, and constructing semantic structures of sentences making up the corpus, and subsequent indexation thereof and storage of the indices. Fragments are then searched in pre-indexed corpuses which satisfy a query. Relevance of the results is evaluated and the search results are ranked according to relevance.

EFFECT: high accuracy of information search by allowing the user to obtain only relevant search results owing to the use of semantic queries.

40 cl, 25 dwg

Description

FIELD OF THE INVENTION

The present invention relates to search technologies, in particular, the implementation of this invention relates to the search for accessible electronic content, for example, on the Internet and other electronic resources such as text corpuses, dictionaries, glossaries, encyclopedias, etc.

BACKGROUND

Search technologies are widely known that allow you to generate a search result based on keywords entered by the user as part of a search query.

However, due to the homonymy and homography available in natural languages, a search result based on a keyword search may include a significant amount of irrelevant and unrelated information. For example, if a user searches for texts containing the word "page" in the sense of "page" (court position), he will receive a lot of irrelevant information, where "page" refers to web pages, pages of newspapers, magazines, pages of memory devices, etc. . This is because these values are much more frequent than the "page" in the lexical meaning "page". Similarly, in the Russian language for the keyword "glass" you can get all the texts containing the verb "flow" in all kinds of word forms.

Existing systems allow the use of simple query languages to search for documents that contain or do not contain words or a word specified by the user. However, the user is not able to specify whether these words should be in the same sentence or not. Also, the user cannot formulate his query immediately for a certain set of words belonging to a certain class or possessing some properties or characteristics. As a rule, these systems do not allow to formulate a request in the form of a common question in a natural language. And finally, existing search systems do not allow users to search for sentences with specified syntactic and / or semantic properties, for example, illustrating a given semantic relation (connection), and do not allow creating queries based on grammatical values, semantic and / or semantic positions (relationships), syntactic models, stylistic and / or semantic features. Such types of searches can be especially useful to lexicographers, philologists, linguists, students and teachers of their native or foreign language, as well as to many ordinary users. This invention is a development of the solutions set forth previously in US Patent Applications No. 13 / 173,649 and 13 / 173,369, filed June 30, 2011, as well as No. 12 / 983,220, filed December 31, 2010 and partially uses analysis technology also patented in the United States (Patent No. 8,078,450).

The closest solution is the solution for US Patent Application No. 12/601901, which describes a method of organizing a search on a variety of electronic documents for a computer system, which consists in combining the search result with keywords with the results of semantic search and presenting the user with the result of such a combined search .

However, the known method has disadvantages. The semantic search, as described in this application, makes no distinction between a "word" and the set of its lexical meanings, and it has a limited character - in fact, only metadata is used as semantic information - some external tags, tags, which, automatically or manually, are supplied documents. It does not allow you to search for words in the selected meaning. The triplestore used additionally stores information only about relations between objects in the form of "triples" of the form {subject, relation, object}.

The technical result from the use of the present invention may consist in the following. Unlike the specified prototype, the method of the present invention involves a computer search of not just words, but words in certain lexical meanings, since it includes an exhaustive semantic-syntactic analysis of the corpus of texts with the extraction of the entire range of lexical, morphological, syntactic and semantic parameters of sentences and the construction of their semantic structures . And subsequent indexing of semantic structures provides the ability to search not only "triples", but sentences of any given structure, including the ability to formulate a query in a natural language.

SUMMARY OF THE INVENTION

The present invention is a method of organizing an information search in electronic text cases for a computer system, which consists in the following sequence of actions being performed at least once: semantic-syntactic analysis of the text body is carried out with determination of the values of the linguistic parameters of sentences, their preservation in memory and the construction of semantic sentence structures that make up the texts of the corpus of texts, and their subsequent indexing and storage HAND indices, then, is searched fragments in pre-indexed housings of electronic texts, satisfying the request, shall assess the relevance of the results, ranking the search results according to the degree of relevance to display them to the user.

The specified comprehensive semantic and syntactic analysis of the text corpus is carried out using a wide range of language descriptions, and for each sentence of the text corpus includes: generating all potential relationships between sentence elements and storing them in memory, constructing a graph of generalized components, generating from the graph of generalized components, at least at least one syntax tree, the establishment of non-wood links and the construction of the syntactic structure of the sentence on the basis of the syntax tree and its storing in the memory, the generation of a universal semantic structure of the sentence on the basis of its syntactic structure and its preservation in memory. The indicated semantic-syntactic analysis of the corpus of texts also includes a lexical choice, which, in one of the implementations, consists in choosing the most probable lexical meaning of a word from the set of possible lexical meanings of this word, taking into account a priori and calculated integral estimates. In yet another implementation, the lexical choice may consist in attributing to the word a certain set of possible lexical meanings, each with a corresponding estimate of the probability of occurrence of this lexical meaning in the sentence. Additionally, the semantic-syntactic analysis of the text corpus includes the definition and extraction of the values of the morphological, lexical, syntactic and semantic parameters of the text corpus elements and the choice of the best syntactic structure of the sentence.

Also, the additionally indicated semantic-syntactic analysis of the text corpus can be accompanied by attribution of the text corpus elements by the concepts of the domain model (ontology) and the restoration of referential relations between sentences.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 illustrates the upholstered scheme of the method of deep analysis of the corpus of texts and the construction of indices according to one implementation of the present invention.

Figa contains an example of a fragment of text with referential connections.

Figure 2 illustrates the sequence of structures built in the process of analysis of the proposal according to one or more implementations of the invention.

Figure 3 illustrates an example of a syntax tree obtained by accurate parsing of the English sentence "This boy is smart, he'll succeed in life".

Figure 4 illustrates a diagram of the semantic structure obtained by analyzing the sentence "This boy is smart, he'll succeed in life."

5A-5D illustrate a fragment of a semantic hierarchy, according to one or more implementations of the present invention.

6 is a diagram illustrating language descriptions 610, according to one possible implementation of the invention.

7 is a diagram illustrating morphological descriptions, according to one possible implementation of the invention.

FIG. 8 illustrates syntax descriptions in accordance with one possible implementation of the invention.

Figure 9 illustrates semantic descriptions, according to one possible implementation of the invention.

10 is a diagram illustrating lexical descriptions, according to one or more implementations of the present invention.

11 and 11B illustrate an example user interface of a search engine that allows a user to formulate queries with a choice of lexical values.

11C and 11D illustrate the results of a semantic query.

12A illustrates one example of a semantic query.

12B illustrates another example of a semantic query.

12C illustrates another example of a semantic query.

13 illustrates an example of a semantic query with a “gap”.

Fig. 14 illustrates an exemplary flowchart for executing a semantic query according to one implementation of the present invention.

14A illustrates an example of a semantic query with a selection of ontology objects.

15 illustrates an example hardware diagram.

DESCRIPTION OF PREFERRED EMBODIMENTS

An embodiment of the present invention discloses indexing methods and a system for indexing natural language texts. Also, the methods and system of semantic search are described.

In general terms, the proposed method for creating an index involves performing comprehensive syntactic and semantic analysis of texes in natural language to construct at least one index for each text or text corpus. At the same time, sets of not just words are considered, but sets of lexical meanings of all words found in the text (corpus), and all lexical, syntactic and semantic information about each sentence obtained in the process of syntactic and semantic analysis is stored and indexed. The stored information may also include data obtained at the intermediate stages of analysis, the results of lexical selection, including the results obtained in the process of resolving ambiguities. The index obtained in this way is used to organize semantic search as described below.

Existing systems use simple query languages to search for documents that contain or do not contain words or words specified by the user. However, the user is not able to indicate in which of his meanings the word should be found, whether the indicated words should be in the same sentence or not. Also, the user cannot formulate his query right away for a certain set of words belonging to a certain class or possessing some properties or characteristics. As a rule, these systems do not allow to formulate a request in the form of a common question in a natural language. And, finally, existing search systems do not allow users to search for sentences with specified syntactic and / or semantic properties, for example, illustrating a given semantic relation (connection), and do not allow creating queries based on grammatical values, semantic and / or semantic positions (relationships) , syntactic models, stylistic and / or semantic features.

The problems of existing search engines are fully or partially solved by the methods described below.

The implementation of the invention allows the user to search and find relevant information using a semantic query, which is formulated either in a special language of semantic queries, and also, possibly, in a natural language. The same analyzer is used to analyze a query in natural language, to recognize its syntactic structure and build a semantic structure and, thus, the system “understands” the meaning of the request. The search is implemented in accordance with the syntax and semantic information that is stored in a resource that allows these types of searches. Thus, the user can get only relevant query results.

In addition, since the search query can be formulated or translated in universal, language-independent, semantic terms, the search can be carried out in cases that include documents in different languages. Thus, the user can receive information presented in various resources regardless of the language in which the search request is formulated. Search results can be presented to the user both in the language of the resource, in the original form, as found in the document, and can be translated into the query language using the machine translation system.

U.S. Patent U.S. Patent 8,078,450 describes a method that includes in-depth syntactic and semantic analysis of texts in natural language, based on exhaustive linguistic descriptions. This method can be used at the stage of analysis of the described method of constructing indexes. The method uses a wide range of linguistic descriptions, as universal semantic mechanisms that relate to a specific language, which allows you to reflect all the real complexities of the language without simplification and artificial restrictions, without fear of a combinatorial explosion, uncontrolled growth of complexity. Moreover, these analysis methods are based on the principles of holistic and targeted recognition, i.e. hypotheses about the structure of a part of a proposal are verified as part of a test of the hypothesis about the structure of the whole proposal. This avoids the analysis of a large number of anomalies and variations.

An in-depth analysis includes lexical-morphological, syntactic and semantic analysis of each sentence of the corpus of texts, as a result of which semantic structures are independent of the language (language-independent semantic structures), in which each word of the text is associated with a corresponding semantic class. Figure 1 illustrates the General scheme of the method of depth analysis and construction of indices according to one implementation of the present invention. The corpus of texts 105 is subjected to an exhaustive semantic-syntactic analysis 106 using linguistic descriptions of both the source language and universal semantic descriptions, which allows us to analyze not only the surface syntactic structure, but also the deep, semantic, expressing the meaning of the statement contained in each sentence, and also links between sentences or text fragments. Linguistic descriptions may include lexical descriptions 101, morphological descriptions 102, syntactic descriptions 103, and semantic descriptions 104. Analysis 106 includes parsing implemented as a two-stage algorithm (crude parsing and accurate parsing) using linguistic models and information of various levels to calculate probabilities and generating the most probable (“best”) syntactic structure. Figure 2 illustrates the sequence of structures built in the process of analysis of the proposal according to one or more implementations of the invention.

Then, a language-independent semantic structure 107, which represents the meaning of the original sentence, is constructed. This stage may also include the restoration of referential relations between offers. An example of a referential connection is anaphora - the use of linguistic constructions that can only be interpreted taking into account another, usually the previous, fragment of the text. 1A illustrates a text fragment with referential connections between sentences. These relations at the level of the corresponding semantic structures are established at step 107. In particular, for subsequent indexing, personal, demonstrative pronouns, as well as other objects between which reference relationships are established, are indexed taking into account the connection with their antecedent. Those. for the text fragment shown in Fig. 1A, the identity of the objects "David Cameron", "Prime Minister of the United Kingdom", "he", "his", "Prime Minister" within the given fragment or the whole text will be established. The restoration of referential connections is performed on semantic structures using special rules.

Then, the original sentence, the syntactic structure of the original sentence, and the language-independent semantic structure are indexed 108. The result is a collection of index collections 109. An index can usually be presented in the form of a table, where each value of a text characteristic (for example, a word, expression or phrase, the relation between elements sentences, morphological, lexical, syntactic or semantic property, as well as syntactic and semantic structures) the list of addresses of their entries is compared in the document tions in this document. According to one implementation of the present invention, morphological, syntactic, lexical and semantic characteristics, as well as structures and fragments of structures, can be indexed in the same way as a word is indexed in a document.

In one implementation of the invention, indices may include all or at least one value of morphological, syntactic, lexical, and semantic characteristics (parameters). These values or parameters are generated during the two-stage semantic analysis, which is described in more detail below. Indexes can be used in many tasks of natural language processing, in particular, for organizing semantic search. According to one implementation of the present invention, morphological, syntactic, lexical and semantic descriptions are structured and stored in a database. This set of descriptions may include at least a morphological model of the language, models of syntactic constructions of the language, lexical-semantic models. According to one implementation of the present invention, an integral model is used to describe syntax and semantics for analyzing complex language structures, recognizing the meaning of a sentence and correctly conveying the information contained in it.

Figure 2 illustrates a detailed diagram of a proposal analysis method according to one or more implementations of the invention. Referring to FIGS. 1 and 2, a lexical-morphological structure 222 is determined at the analysis stage 106 of the original sentence 105. Then, parsing is performed as a two-stage algorithm (rough parsing and accurate parsing) using linguistic models and various information levels for calculating probabilities and generating the most probable (“best”) syntactic structure.

Rough parsing is applied to the original sentence and includes, in particular, the generation of all potential lexical meanings of the words that make up the sentence or phrase, all potential possible relationships between them, all potential components. All possible surface syntactic models are applied for each element of the lexical-morphological structure, then all possible components are constructed and generalized so that all possible variants of the syntactic analysis of the sentence are presented. As a result, a graph of generalized components 232 is formed for subsequent accurate parsing. The generalized component graph 232 includes all potential relationships in a sentence. Rough parsing is followed by precise parsing on the graph of the generalized components, as a result of which one or more syntax trees 242 representing the structure of the original sentence are "extracted" from it. The construction of the syntax tree 242 includes the lexical choice for the vertices of the graph and the choice of relations between the vertices of the graph. A lot of a priori and statistical estimates can be used when choosing lexical options and when choosing relationships from a graph. A priori and statistical estimates can also be used both for estimating parts of the graph and for estimating the entire tree. In one implementation, one or more syntax trees are constructed or ordered in descending order of rating. Thus, the best syntax tree can be built first. At this point, non-timber relationships are also being tested and built. If the first syntax tree is not suitable, for example, due to the inability to establish the necessary non-wood links, the second syntax tree, etc., is better considered.

Since the mentioned lexical choice for the vertices of the graph and the choice of relations between the vertices of the graph are made on the basis of a priori and statistical estimates, in one of the implementations of the method not only all options are considered and evaluated, but these options are also stored and indexed at step 108, taking into account their integral estimates. Those. index 109 contains not only highly probable options for parsing a sentence, but also unlikely with an appropriate weight if such a parsing ended successfully. The weights of the parsing options are subsequently used in calculating the relevance score of the search result.

A wide range of lexical, grammatical, syntactic, pragmatic, semantic characteristics is extracted at this stage of the analysis 106 and the construction of semantic structures 107. For example, the system can extract and store lexical information and information about the lexical units belonging to semantic classes, information about grammatical forms and linear order, about syntactic relations and superficial positions, the use of certain forms, aspects, tonality, such as positive and negative tonality, Lubin products, non-timber connections, semantics, etc.

Also, at step 107, an ontological analysis can be carried out with the aim of extracting knowledge about the subject area, extracting ontobjects and ontofacts. The extraction of ontobjects and ontofacts and fixing the relations between them is carried out, for example, using a special type of rules, inference rules, and other means. This information is recorded in ontologies 110. For example, returning to the example presented in Fig. 1A, the ontology stores information that David Cameron is the Prime Minister of Great Britain, that his address is 10 Downing Street, that he spoke at Social Impact Investment Forum, which took place on June 6, 2012 in London.

Information from the ontology is involved in the process of constructing indexes 108. This then allows, in the search process, to find information about the object, even if it is expressed implicitly in the body of texts. For example, the information from the fragment shown in FIG. 1A, if it is included in the ontology, allows us to answer the question on which street the residence of the British Prime Minister is located, or when the coalition government worked in the UK.

Figure 3 illustrates an example of a syntax tree 300 obtained by precise parsing of the English sentence "This boy is smart, he'll succeed in life". The tree contains fairly complete syntactic information, such as lexical meanings, parts of speech, syntactic roles, grammatical meanings, syntactic relations (positions), syntactic models, types of non-woody links, etc. For example, the pronoun “he” is defined as belonging to the noun “boy” as the subject of anaphoric communication 310. “Boy” is defined by the subject 320 of the verb “be.” “Not” by subject 330 of the verb “succeed.” The adjective "smart" appears to be related to the noun "boy" with the relation "control - complement" 340.

Referring to FIG. 2, this two-step parsing approach provides the construction of a better syntax structure 246 of the original sentence selected from one or more syntax structures. Figure 3 illustrates a diagram of the best syntactic structure obtained by parsing the sentence "This boy is smart, he'll succeed in life." The approach of two-stage analysis follows the principle of holistic and purposeful recognition, that is, hypotheses about the structure of a part of a sentence are checked using available linguistic descriptions in the framework of the structure of the whole sentence. With this approach, there is no need to analyze many dead-end parsing options. In most cases, this approach can significantly reduce the amount of computing resources needed to analyze the proposal.

The proposed analysis methods ensure maximum accuracy in understanding the meaning of the proposal. Figure 4 illustrates a diagram of the semantic structure obtained by analyzing the sentence "This boy is smart, he'll succeed in life." This structure contains all syntactic and semantic information, such as semantic classes, semantems (which are not shown in the figure), semantic relations (deep positions), non-wood communications, etc.

The language-independent semantic structure of the sentence is represented in the form of an acyclic graph (a tree supplemented by non-wood connections), where each word of a certain language is replaced by universal (language-independent) semantic entities, called here semantic classes. The semantic class is one of the most important semantic characteristics that can be extracted and used to solve the problems of semantic search, classification, clustering and filtering of documents written in one or several languages. In addition, information in language-independent structures can be used semantems that reflect not only semantic, but also syntactic, grammatical and other language-dependent information.

Semantic classes can be ordered into a semantic hierarchy, where the "child" semantic class and its "descendants" inherit a significant part of the properties of the "parent" and all previous semantic classes ("ancestors"). For example, the semantic class SUBSTANCE (substance) is a daughter class of a fairly wide class ENTITY (entity), and at the same time it is the "parent" for the semantic classes GAS (gas), LIQUID (liquid), METAL (metal), WOOD_MATERIAL (wood as material), etc. Each semantic class in the semantic hierarchy is equipped with a deep (semantic) model. The deep model is a set of deep positions (types of semantic relationships in sentences). Deep positions reflect the semantic roles of child components (structural units of a sentence) in various sentences with objects of a given semantic class as the core of the parent component and possible semantic classes as placeholders. These deep positions express semantic relations between components, for example, "agent" (agent), "addressee" (addressee), "instrument" (instrument), "and" (quantity), etc. The child class inherits and adjusts the deep model of the parent class.

5A-5D illustrate a fragment of a semantic hierarchy, according to one or more implementations of the present invention. The semantic hierarchy is structured in such a way that more general concepts are at the upper levels of the hierarchy. For example, in the case of documents whose types are illustrated in Fig. 5B and Fig. 5C, the semantic classes are PRTNTED_MATTER (print, 502), SCIENTIFIC_AND_LITERARY_WORK (scientific works and literature, 504), TEXT_AS_PART_OF_CREATIVE_WORK (other text classes and others), then TEXT_OBJECTS_AND_DOCUMENTS (text objects and documents, 501), and the class PRINTED_MATTER (print, 502), in turn, is the parent for the semantic class EDITION_AS_TEXT (edition as text, 503), containing the classes PERIODICAL (periodicals) and NONPERIODICAL (non-period) ), where PERIODICAL (periodic Editions) - the parent class for the ISSUE (issue), MAGAZINE (magazine), NEWSPAPER (newspaper), etc. classes. The approach to dividing into classes may vary. This invention is primarily based on the use of concepts that are independent of the language.

6 is a diagram illustrating language descriptions 610, according to one possible implementation of the invention. Language descriptions 610 include morphological descriptions 101, syntactic descriptions 102, lexical descriptions 103, and semantic descriptions 104. Language descriptions 610 are combined into a common concept. 7 is a diagram illustrating morphological descriptions, according to one possible implementation of the invention. FIG. 8 illustrates syntax descriptions in accordance with one possible implementation of the invention. Figure 9 illustrates semantic descriptions, according to one possible implementation of the invention.

Referring to Fig.6 and Fig.9, as part of the semantic descriptions 104, the semantic hierarchy 910 is a characteristic of language descriptions 610, which combines language-independent semantic descriptions 104 and language-dependent lexical descriptions 103, as indicated by a double arrow 623, and morphological descriptions 101 and syntax descriptions 102, as indicated by the double arrow 624. A semantic hierarchy can be created at one time, and then can be populated for each specific language. The semantic class in a particular language includes lexical meanings with appropriate models. The semantic descriptions 104 are language independent. Semantic descriptions 104 may contain descriptions of deep components and may contain a semantic hierarchy, descriptions of deep positions, a system of semantems and pragmatic descriptions.

Referring to FIG. 6, in one possible implementation of the invention, morphological descriptions 101, lexical descriptions 103, syntactic descriptions 102, and semantic descriptions 104 are associated. Several superficial (syntactic) models due to semantems and pragmatic characteristics can have lexical meaning. Syntactic descriptions 102 and semantic descriptions 104 are also related. For example, the diathesis of syntactic descriptions 102 can be considered as an “interface” between language-dependent surface models and language-independent deep models of semantic description 104.

7 illustrates an example of morphological descriptions 101. As shown, the components of morphological descriptions 101 include, but are not limited to descriptions of inflection 710, grammar system (grams) 720, and descriptions of word formation 730. In one possible implementation of the invention, grammar system 720 includes a set of grammatical categories such as “Part of speech”, “Case”, “Rod”, “Number”, “Face”, “Return”, “Time”, “View” and their meanings, hereinafter referred to as grammes. For example, grammes meaning parts of speech may include an adjective, noun, verb, etc .; case grammars may include “Nominative”, “Genitive”, “Dative”, etc .; gender grammes may include "Male", "Female", "Medium", etc. Referring to FIG. 7, inflection descriptions 710 describe how the initial form of a word can vary depending on case, gender, number, time, etc. and include in a broad sense all possible forms of a given word. Word formation descriptions 730 describe which new words can be constructed using a given word. Grammes are units of the grammatical system 720 and, as shown in reference 722 and reference 724, grammes can be used to construct descriptions of inflection 710 and descriptions of derivation 730.

8 illustrates syntactic descriptions 102. Components of syntactic descriptions 102 may include surface models 810, descriptions of surface positions 820, descriptions of reference and structural control 856, descriptions of control and alignment 840, non-wood descriptions 850, and analysis rules 860. Syntactic descriptions 302 are used to construct possible syntactic sentence structures for a given source language, taking into account the word order, non-wood syntactic phenomena (for example, matching, ellipsis, etc.), reference ny control (management), and other phenomena.

9 illustrates semantic descriptions 104 according to one possible implementation of the invention. While surface positions 820 reflect syntactic relations and ways of their implementation in a particular language, deep positions 914 reflect the semantic roles of daughter (dependent) components in deep models 912. Therefore, descriptions of surface positions, and more broadly - surface models, can be specific to each specific language. Descriptions of deep models 920 contain grammatical and semantic restrictions for placeholders for these items. The properties and limitations of the deep positions 914 and their placeholders in the deep models 912 are very similar and often identical for different languages.

The semantem system 930 represents many semantic categories. Semantems can reflect lexical, grammatical properties and attributes, as well as differential properties and stylistic, pragmatic and communicative characteristics. For example, the semantic category "DegreeOfComparison" (degree of comparison) can be used to describe the degrees of comparison expressed by different forms of adjectives, for example, "easy", "easier" and "easiest". So, the semantic category "DegreeOfComparison" can include semantems, for example, "Positive", "ComparativeHigherDegree", "SuperlativeHighestDegree". As another example, the semantic category "RelationToReferencePoint" can be used to describe in which linear order - before or after an object or event a link to it is in the sentence, and its semantems are "Previous", "Subsequent". Another example - the semantic category "EvaluationObjective" can record the presence of an objective assessment, such as "Bad", "Good", etc. Lexical semanthemes can describe specific properties of objects, for example, “being flat” or “being liquid” and are used in restrictions on placeholder placeholders. Classifying differential semantems are used to express differential properties within a single semantic class. For example, in English the “hairdresser” for men is translated as “barber”, and the semantema “RelatedToMen” will be assigned to him in the semantic class “HAIRDRESSER”, while in the same semantic class there is “hairdresser” and “hairstylist”, etc. .

The pragmatic descriptions 940 serve to fix the corresponding topic, style or genre of the text during the analysis of the text, and it is also possible to attribute the corresponding characteristics to objects of the semantic hierarchy. For example, "Economic Policy", "Foreign Policy", "Justice", "Legislation", "Trade", "Finance", etc.

10 is a diagram illustrating lexical descriptions 103, according to one or more implementations of the present invention. Lexical descriptions 103 include a lexical-semantic dictionary 1004, which includes a set of lexical meanings 1012, forming together with their semantic classes a semantic hierarchy, where each lexical meaning can include, but is not limited to its depth model 912, surface model 810, grammatical meaning 1008 and semantic meaning 1010. The lexical meaning can combine various derivatives (for example, words, expressions, phrases) expressing meaning with the help of different parts of speech, different forms of words and, single-root words, and so forth. In turn, the semantic class combines the lexical meanings of related words and phrases in different languages.

Any parameter of the 610 language description - lexical meanings, semantic classes, grammes, semanthemes and much more is extracted during exhaustive analysis of the text, and any parameter can be indexed (a characteristic index is created). Indexing of semantic classes is in demand in many tasks related to the analysis of natural language texts, such as semantic search, classification, clustering, text filtering and many others. Indexing lexical meanings (as opposed to indexing just words) allows you to search not just words or word forms, but lexical meanings, i.e. words in a certain semantic meaning. Syntactic structures and semantic structures can also be indexed and stored for use in semantic search, classification, clustering and filtering of documents.

Returning to FIG. 1, after a universal semantic structure is constructed for each sentence of each text in the body of texts, syntactic and semantic structures are indexed. Lexical values are indexed as a result of lexical choice at each vertex of the semantic structure, each parameter of morphological, syntactic, lexical and semantic descriptions can be indexed in the same way as ordinary words. An index of words in a document usually includes at least one table, where each word (token or word form) that appears in the document is accompanied by a list of numbers or addresses of positions in this document. According to the implementation of this invention, the index is built for all lexical values, all semantic classes, for any values of morphological, syntactic, lexical and semantic parameters. These parameter values are generated during a two-stage syntactic-semantic analysis, and the resulting indices can be used to achieve higher accuracy and relevance of the semantic search in cases of texts in natural languages. For example, a user can formulate his request with the ability to search for sentences with nouns that have the property “being flat” or “being liquid” or sentences containing words (nouns and / or verbs) that indicate a process, for example, production, destruction, movement etc.

In one possible implementation of the method of the invention, a combination of two, three or, generally speaking, N numbers can be used to index various syntactic, semantic or other parameters. For example, in order to index surface or deep positions, combinations of two numbers can be used - word numbers, which in the text are related by the relation corresponding to this position. For example, for the semantic structure of the sentence "This boy is smart, he'll succeed in life" shown in Figure 4, the deep position 'Sphere' (450) correlates the lexical meaning "succeed: TO_SUCCEED" (460) with the lexical meaning "life : LIVE (470). " More specifically, the lexical meaning “life: LIVE” fills the deep “Sphere” of the verb “succeed: TO_SUCCEED”. When the index of lexical values is constructed, in accordance with the method of the present invention, the occurrences of these lexical values are assigned numbers in accordance with their position in the text, for example, N1 and N2. When the index of deep positions is built, each deep position is assigned a list of its occurrences in the document. For example, the 'Sphere' deep position index will, among others, include a pair (N1, N2).

Because not only words are indexed, but their lexical meanings, semantic classes, syntactic and semantic relations, any other elements of syntactic and semantic structures, it becomes possible to search for a context not only by keywords, but also a context containing lexical meanings, meanings belonging to certain semantic classes , a context that includes elements with certain syntactic and / or semantic properties and / or morphological properties or sets (combinations) of such properties. Also, sentences with non-wood syntactic phenomena, for example, ellipsis, composition, etc. can be found. you can search for semantic classes, it becomes possible to search for semantically related words and concepts.

Also, fragments of syntactic and / or semantic structures can be searched. The search result may be sentences or paragraphs, or other fragments, depending on the user's choice of the corresponding option. Because all sentences in each case are analyzed and then stored together with the results of their syntactic and semantic analysis, syntactic and semantic structures can also be presented to the user in graphical form.

The most common search volume in regular search engines is keyword search. But a problem arises if one or more keywords are ambiguous, for example, in English, “bank” can mean 1) a financial institution, a bank; 2) storage, repository; 3) the shore of a river or lake. There are other, less frequent, bank values. In this case, in response to a request for such a keyword, the user receives many unnecessary results, because the user does not have the opportunity to tell the search engine which of the possible values he is interested in. For example, most of the results of a regular search for the keyword "apple" refer to the name of a computer corporation, since it is much more common in Internet resources, and it is almost impossible to find documents mentioning "apple" as a fruit (they will appear on the farthest pages) without adding additional words (for example, "fruit") in a request or exceptions in a special way of a computer vocabulary. In the implementation of the present invention, the user can specify one of the possible lexical meanings for the keyword. For example, FIG. 11A shows a search engine user interface comprising a query window 1101 and a Find button 1102, and furthermore, a popup window 1103 is shown with a list of lexical meanings of the word “center” that are shown to a user to select a suitable value when the cursor is over on word. The user can see that the word "center" has several meanings that are in different semantic classes, and to select the relevant one in the query field, click on the desired one with the mouse button. There are different ways to specify lexical meaning, for example, by specifying a semantic class. Another way is to provide each lexical meaning with an interpretation, like an article in the explanatory dictionary, as shown in Fig. 11A, or an indication of the synonyms closest to the given meaning.

The user can select any possible lexical meaning of the word to search, this value becomes selected and marked, for example, as shown in Fig.11B. If the query includes several words, lexical meanings from the corresponding semantic classes can be indicated for any of them or for all.

Regardless of whether the user specifies the query words, the user can in the found fragment, for example, by clicking on the right mouse button, see in what lexical meaning the word occurs in the found fragment, its semantic class, as well as some other parameters, for example, synonyms , syntactic model, compatibility, examples of phrases with a given word, etc. An example of such a request is shown in FIG. 11C.

Using search keywords in conventional search engines, the user can explicitly or implicitly use Boolean operators such as AND (and), OR (or) and NOT (not) to create their own query. However, he cannot choose in which lexical meaning the given word should be sought. The method of the present invention allows the user to apply the indicated Boolean operators AND (and), OR (or) and NOT (not) to lexical values, i.e. keywords for which the corresponding semantic classes are indicated.

The word order can be significant or not in accordance with the option selected by the user provided to the user in the list of options, for example, 1104. Another option selected by the user allows you to limit the distance between the objects of the request. For example, a user can use the W / n operator to search for documents containing no more than n words between query objects. In one implementation, the user can use it explicitly, in others - choose the distance as an option.

An additional feature inherent in semantic search, in contrast to the usual keyword search, is to search by synonyms. Synonyms are words that have the same or very close meaning. In the semantic hierarchy, where all synonymous lexical meanings, for example, "food", "meal", "alimentary" are in the same semantic class and have the same or similar semantic characteristics - semantems. Then, if the user turns on the option “Search for synonyms” (1104) and wants to find “food”, then its lexical meaning, semantic class is determined first, and as a result, documents can be found where “meal” or “alimentary” or perhaps the other most representative representatives of the semantic class FOOD. In such cases, the search results may be more or less relevant, more or less close to the desired result. A measure of relevance can be introduced, for example, based on an assessment of the "proximity" of the lexical meaning from the query to the found synonym, and, taking into account the context, word order and other factors, it can be extended to a sentence, fragment, etc.

Another way to specify the lexical meanings of the words involved in a query, if the query is represented by a coherent phrase or sentence, is its full syntactic and semantic analysis. Many words in a language have several different meanings and meanings. Some words in the query can also have different meanings, expressed in different lexical meanings. To determine in which lexical meaning the word represented in the query should be searched, a complete syntactic and semantic analysis of the query represented by the sentence or phrase is performed to determine the lexical meaning of the words making up the query. Comprehensive parsing includes crude and accurate parsing. During a crude analysis, all potential meanings for each word are determined, during an accurate analysis, based on linguistic descriptions, language rules, compatibility, context analysis, statistics and other factors, the most relevant lexical meanings are selected. Thus, as a result of lexical choice during accurate analysis, for each word in the user's query, its semantic class is determined. 11D illustrates an example of an interface with a user request in the form of a phrase, where the user can check which lexical meaning of the word will be searched in the documents of the text corpus. Documents found can contain all (or most) of the query words in the selected values. The word order can be significant or not in accordance with the option selected by the user. In one implementation, the word order may be required by default. The search system can also search for documents with an arbitrary word order of the query, as well as indicating the maximum distance between them, but in this case, the relevance score for the found document may be lower. In another implementation, the user is given the choice of search option - to take into account the word order or not.

12A-12C show examples of user queries formulated as questions in a natural language. 12A shows an example of a query that answers the question "What countries were discovered?" (Which countries were open?). The query in natural language is analyzed by the same analyzer as the texts of the corpus, and is translated into a semantic query. The query window 1210 shows the query in natural language, and in terms of semantic classes and relations it looks, for example, as "discover [Object:" COUNTRY "], and means that the semantic search system searches for sentences with the English verb" discover "and an object belonging to the semantic class COUNTRY. Search results are shown in window 1220.

Another example of a semantic query is shown in Fig. 12B, it answers the question "What can be made from milk?" (What can be prepared from milk?). It is also formulated in natural language and translated into the semantic query "CREATION_VERBS [Object: <Noun>, Fabricative: milk]", where "CREATION_VERBS" is a semantic class of verbs that indicate the creation process, such "to make" (to do), "to produce "," to manufacture "," to confect ", etc. Any of these verbs belonging to the semantic class" CREATION_VERBS "can be found in the body texts. "Fabricative" means the deep (semantic) position of the "raw material" from which something is produced; the raw material in this query is "milk". Thus, the result should be a noun (<Noun>) filling in the semantic position of "Object" in the sentence. Search results are shown in window 1220.

Another example of a query to a semantic search system is shown in FIG. 12C. Meaningfully, the query means "What do the people collect?" (What do people collect?). In terms of a semantic query, the query shown in window 1310 looks like this: "collect [$ Subject: people, $ Object_Direct: <Noun>]". The semantic search system searches for sentences with the verb "collect" (collect) and the word "reor1e" (people), which fills the surface position "Subject". The purpose of the search is for nouns (noun or nouns) as the direct object of the verb "collect". Search results are shown in window 1220. In one implementation of the invention, all morphological forms of the words indicated in the request are taken into account, and all morphological forms can be found. Restrictions on morphological forms can be included in the request in the form of restrictions on grammatical values, and can be shown, for example, in angle brackets <>. An index obtained in accordance with these methods is an integral part of the semantic search system.

Various implementations of the present invention also include a semantic search method in which a user can formulate his question in a natural language. The same analyzer, which is used for syntactic-semantic analysis of the corpus of texts, is used for syntactic-semantic analysis of the user's question, its syntactic structure is recognized, a semantic structure independent of the language is built, and in this way the system recognizes the meaning of the sentence. The constructed semantic structure is then translated into the search engine query language. For example, the mentioned questions "What coun tries were discovered?" or "What can be made from milk?" originally formulated by the user in a natural language, analyzed by the analyzer, and translated into a semantic query. If the request is formulated in the form of an interrogative sentence, then as a result of the analysis, the structure of the sentence is constructed, which may be a potential answer to the asked question, in this case it contains "gaps."

The method of the present invention allows the creation of indices of syntactic and semantic structures stored in a tree or graph. The desired structures are described in the search query using one or more of the desired parameters. One or more parameters can be undefined, can be set using variables, can be set as a range of possible values of these variables. In other words, a search query can be represented by a natural language sentence with “gaps”. Lacunae can be covered or include both single words and phrases, phrases, groups of words that form a related component (constituent), embedded (subordinate) sentence, etc. The result of the query and search are options for filling these gaps in the texts available in the marked up body of texts. To be shown to the user, they can be sorted by frequency of occurrence.

13 shows one example of a query using a variable to indicate a gap in a search term. Inquiry

[$ Subject: [... [people]]] think

demonstrates the ability to search for proposals of a certain structure. In this implementation, a variable, for example, [$ Subject: [... [people]]] is used to describe the search sentence, including a noun phrase with the core "people" filling in the syntactic (surface) position "Subject" and the verb "think" in any form .

The method makes it possible to search for a wide range of entities, such as relationships, non-wood relationships, lexical classes, semantic classes, etc. These entities - words and phrases can be found by grammatical characteristics, lexical properties, syntactic and / or semantic properties. Klose (sentences) can be found by any - lexical, syntactic, semantic features.

Using the methods of the present invention, very specific search types can be implemented. For example, you can search by the properties of the nodes of the syntactic or semantic structure - grammatical values, by superclass (a class with all descendants), by semantems (for example, "time" regardless of the form of expression). Another search option is to search for relationships by property. For example, it is possible to search for surface or deep positions with verification (or without verification) of inheritance, anaphoric relationships, or any number of nodes with the specified properties. Additionally, it is possible to extract the values of these characteristics in the results found.

It is significant that since a search query can be translated into a semantic, language-independent form, the search can be performed in texts in different languages, on resources, including corpus of different languages. Thus, the user can receive information presented on all resources, regardless of the language of the request. The search result can be presented to the user in the language of the resource (as presented in the original) or the search result can be translated into the query language using a machine translation system.

In some implementation, not only one or more indices can be used for the search, but also formal models representing knowledge of one or more subject areas. For example, ontologies can be used as models of regions. An ontology may include, among other things, a multitude of concepts and entities related to the subject area and the relationships between them. They are used for domain modeling and inference.

FIG. 14 is a flowchart of a search engine performing a semantic search in a markup, i.e. subjected to exhaustive semantic-syntactic analysis, the corpus of texts. The search query 1410 undergoes an exhaustive semantic-syntactic analysis, similar to how it is done for each sentence of the corpus of texts according to steps 106 and 107, the method presented in FIG. 1. Thus, the semantic-syntactic analysis of a query includes lexical analysis using morphological 101 and lexical 103 descriptions, then parsing is performed using syntactic 102 and semantic 104 descriptions, during which, in particular, all potentially possible lexical meanings of words that form a sentence or collocation and all potential relationships between them. As a result, the most probable syntactic structure is selected, lexical meanings and the corresponding and semantic classes of all elements of the sentence are fixed.

The semantic structure of the query is constructed, which, along with the syntactic structure of the sentence, can be used for formal conclusions using the data stored in the ontologies of 1480. For example, to the question "Who won the Roman fencing Olympiad?" must be found, in particular, the sentence "Petrov became the Olympic champion in fencing in 1960." The unification of semantic structures is carried out with the help of transformational rules, ontology rules and ontology data comparison that "the Roman Olympics were in 1960", "the 1960 Olympics were held in Rome", "Roman Olympics" = "Olympics in Rome", " win the Olympics "=" become an Olympic champion. " And also, "Olympics in Rome" = "1960 Summer Olympics."

An ontology may include a description, for example, of entities, classes (concepts, concepts), attributes, relationships, and ontofacts. For example, entities are instances or objects of classes that represent the basic level of concepts (objects). Also, classes can represent sets, collections, concepts, classes in programming, types of objects, types (types) of things, etc. Examples of classes can be Person (geographic objects), Company (companies), Organization (organizations), Numerical Value (numerical values), etc.

Attributes express aspects, properties, attributes, characteristics, characteristics or parameters that objects or classes may have. Relations are a way of expressing the relations of entities and classes to each other and to each other. Some of the events may cause a change in attributes or relationships. Ontologies are encoded using special ontology languages.

The mentioned ontology objects - entities, classes (concepts), attributes, relationships, ontofacts can be included in the search query. Different methods can be used to specify the words included in the query as ontological objects. For example, in a query, “Dakota” may mean at least a location, person, or organization. In one implementation, such objects may be marked explicitly, for example, "Dakota% person". In another implementation, if the “Search in ontology” option is enabled, the user can see a menu for selecting the type of ontology object, as shown, for example, in FIG. 14A. In yet another implementation, the type of ontological object can be determined automatically. For example, based on an analysis of a request, a pragmatic or contextual analysis of previous requests, a topic of a corpus selected for search, user information, etc. For example, for the query "(apple OR НТС) AND NOT litigation", the system can analyze the subject matter of the words in the query and conclude that it is more likely that for the "apple" one should choose the corresponding lexical value in the semantic class APPLE_COMPUTER_FIRM than in the classes "FRUIT "or" FRUIT_TREE ", and if the" Search in ontology "option is enabled, the system will automatically select the type of object - Company and the user will be able to see information from the ontology related to Apple. In case of ambiguity, the user may be asked one or two questions to clarify the type of object.

Now, taking into account the variants of the desired semantic and syntactic structures and the data obtained from the ontology 1480, the system can compile an exhaustive semantic query to the indices 1460 of the corresponding corpus of texts 1470. Since the meaning (semantics) of the original question is fixed, the lexical and semantic values are searched in the search process 1440. structures of sentences containing the answer to the desired question, taking into account possible options for the wording of the question and various syntactic and lexical options for the answer. Also, the search is performed taking into account the anaphoric and other referential connections between sentences established during the construction of the semantic structures of the sentences being analyzed at step 107 (Fig. 1A), so that the system finds fragments of text where the search object is expressed implicitly, using links, somehow personal, indicative, relative pronouns, reference nouns, null groups, substitutions, etc.

Each piece of text found can be formally rated 1450 according to the degree of relevance to the request. In particular, in one implementation of the invention, the relevance score 1450 takes into account the indexing of lexical values with probability when constructing the index 109, as described in the description of step 108 shown in FIG. 1. Thus, the search results 1490 can be ranked by relevance and shown to the user in the corresponding order.

15 is a possible example of computing means 1500 that can be used to implement the present invention, implemented as described above. Computing means 1500 includes at least one processor 1502 connected to memory 1504. Processor 1502 may be one or more processors, may include one, two, or more processing cores. Memory 1504 can be random access memory (RAM), and also contain any other types and types of memory, in particular, non-volatile memory devices (eg, flash drives) and read-only memory devices, such as hard drives, etc. In addition, it may be considered that the memory 1504 includes hardware for storing information physically located elsewhere in the computing means 1500, for example, a cache memory in a processor 1502, a memory used as virtual and stored on an external or internal read only memory 1510.

Computing means 1500 also typically has a number of inputs and outputs for transmitting information to the outside and receiving information from the outside. To interact with a user, computing means 1500 may comprise one or more input devices (e.g., a keyboard, mouse, scanner, etc.) and a display device 1508 (e.g., a liquid crystal display). Computing means 1500 may also have one or more read-only memory devices 1510, for example, an optical disc drive (CD, DVD or another), a hard disk drive, a tape drive. In addition, computing means 1500 may have an interface with one or more networks 1512 that provide connectivity to other networks and computing devices. In particular, it can be a local area network (LAN), a wireless Wi-Fi network connected to the Internet or not. Computing means 1500 are intended to include suitable analog and / or digital interfaces between processor 1502 and each of components 1504, 1506, 1508, 1510, and 1512.

Computing tool 1500 is running an operating system 1514 and executes various applications, components, programs, objects, modules, etc., indicated collectively by the number 1516.

In general, programs executed to implement the methods of this invention may be part of the operating system or may be a stand-alone application, component, program, dynamic library, module, script, or a combination thereof.

The present description sets forth the main inventive concept of the authors, which cannot be limited to those hardware devices that were mentioned earlier. It should be noted that hardware devices are primarily designed to solve a narrow problem. Over time and with the development of technological progress, such a task becomes more complicated or evolves. New tools are emerging that are able to fulfill new requirements. In this sense, these hardware devices should be considered from the point of view of the class of technical problems they solve, and not purely technical implementation on a certain elemental base.

Claims (40)

1. The method of organizing a search in electronic text cases for a computer system, which consists in the fact that at least once the following sequence of actions is performed:
- preliminarily carry out a semantic-syntactic analysis of the corpus of texts with determination of the values of the linguistic parameters of sentences, including matching at least one word of a sentence with a certain set of possible lexical meanings - each with a corresponding assessment of the probability of occurrence of this lexical meaning in a sentence, their storage in memory and construction semantic structures of sentences that make up the texts of the corpus of texts, and their subsequent indexing and preservation of indices,
- they search for fragments in pre-indexed cases of electronic texts that satisfy the request,
- evaluate the relevance of the results,
- rank the search results in accordance with the degree of relevance. 2. The method according to p. 1, where the semantic-syntactic analysis of the corpus of texts for each sentence of the corpus of texts includes:
- the generation of all potential relationships between the elements of the proposal and their preservation in memory,
- construction of a graph of generalized components,
- generation of generalized components from the graph of at least one syntax tree,
- the establishment of non-wood relationships and the construction on the basis of the syntax tree of the syntactic structure of the sentence and its storage in memory,
- the generation of a universal semantic structure of a sentence based on its syntactic structure and its preservation in memory. 3. The method according to claim 1, where the semantic-syntactic analysis of the corpus of texts further includes selecting the best syntactic structure of the sentence. 4. The method according to claim 3, where the choice of the best syntactic structure of the sentence includes lexical selection, which is made taking into account the most probable lexical meaning of at least one word from the set of possible lexical meanings of this word based on a priori and calculated integral estimates. 5. The method according to claim 1, where the semantic-syntactic analysis of the corpus of texts includes determining and extracting the values of the morphological, lexical, syntactic and semantic parameters of the elements of the corpus of texts. 6. The method according to p. 1, where the semantic-syntactic analysis of the corpus of texts further includes attributing the elements of the corpus of texts with concepts of the domain model (ontology). 7. The method according to p. 1, where the semantic-syntactic analysis of the corpus of texts further includes restoring the referential relations between sentences. 8. The method according to p. 1, where the indexing is to build an index of lexical values and write it to the system’s memory. 9. The method according to p. 1, where the indexing is to build an index of the values of at least one linguistic parameter and write it to the system’s memory. 10. The method according to p. 1, where the indexing is to build an index of syntactic structures and write it to the system’s memory. 11. The method according to p. 1, where the indexing is to build an index of semantic structures and write it to the system’s memory. 12. The method according to claim 1, where the elements of the query are lexical values. 13. The method according to p. 12, where the query is a combination of lexical values using the Boolean operators "AND", "OR", "NOT". 14. The method according to p. 12 or 13, where the lexical values of the query elements are selected by the user by selecting the appropriate semantic class from the list of possible semantic classes. 15. The method according to p. 12 or 13, where the lexical values of the query elements are selected by the user by selecting the appropriate interpretation from the list of possible interpretations. 16. The method according to p. 12, where the lexical values of the elements of the query are determined automatically taking into account probability estimates. 17. The method according to p. 12, where the lexical values of the elements of the query are determined automatically based on pragmatic and contextual analysis. 18. The method according to p. 12, where to determine the lexical values of the elements of the query produce semantic-syntactic analysis of the request, including at least parsing, fixing the lexical values of the elements of the request, recognition of the semantic structure of the request. 19. The method of claim 1, wherein the query element is a syntax structure. 20. The method according to claim 1, where the query element is a semantic structure. 21. The method according to claim 1, where the request is formulated in the form of a sentence in a natural language. 22. The method according to p. 21, where the request is formulated as an interrogative sentence in a natural language. 23. The method according to p. 21, where the request in the form of a sentence in a natural language contains a variable (lacuna). 24. The method according to claim 1, where the request element is attributed to the concept of a domain model (ontology). 25. The method according to p. 24, where the attribution of the query element by the concept of the domain model (ontology) is performed automatically. 26. The method according to claim 1, where the search is performed taking into account the restored referential relations between sentences. 27. The method according to claim 1, where the search also includes the search for lexical values that are close in value to the query elements. 28. The method according to claim 1, where the search consists in finding identical to the requested or similar syntactic structures. 29. The method according to claim 1, where the search consists in finding identical semantic structures to the requested or similar. 30. The method according to claim 1, where the relevance of each found fragment is evaluated taking into account the “measure of similarity” of each element of the request with the corresponding element in the found fragment. 31. The method according to p. 1, where the relevance of each found fragment is assessed taking into account the probability of the desired lexical meaning in this fragment, which was determined during the lexical choice in the process of preliminary analysis of the text corpus. 32. The method according to claim 1, where the search results contain information about the lexical values of the query elements in the text of the fragments found. 33. A system for organizing searches in electronic text buildings, including:
- one or more processors;
- one or more memory devices;
- software instructions for a computing device recorded in one or more memory devices that, when executed on one or more processors, control the system for:
- preliminary semantic-syntactic analysis of the corpus of texts with determination of the values of the linguistic parameters of sentences, including matching at least one word of a sentence with a certain set of possible lexical meanings - each with a corresponding assessment of the probability of occurrence of this lexical meaning in a sentence, their storage in memory and construction semantic structures of sentences that make up the texts of the corpus of texts, and their subsequent indexing and preservation of indices,
- search for fragments in pre-indexed cases of electronic texts that satisfy the request,
- assessment of the relevance of the results,
- ranking of search results in accordance with the degree of relevance. 34. The system of claim 33, wherein the semantic-syntactic analysis of the corpus of texts for each sentence of the corpus of texts includes:
- the generation of all potential relationships between the elements of the proposal and their preservation in memory,
- construction of a graph of generalized components,
- generation of generalized components from the graph of at least one syntax tree,
- the establishment of non-wood relationships and the construction on the basis of the syntax tree of the syntactic structure of the sentence and its storage in memory,
- the generation of a universal semantic structure of a sentence based on its syntactic structure and its preservation in memory. 35. The system of claim 33, wherein the semantic-syntactic analysis of the corpus of texts further includes selecting the best syntactic structure of the sentence. 36. The system of claim 33, wherein indexing is to build an index of lexical and / or syntactic and / or semantic values and write it to the system’s memory. 37. The system of claim 33, wherein the query elements are lexical meanings and / or syntactic structure and / or semantic structure. 38. The system of claim 33, wherein the request is formulated as a sentence in a natural language. 39. The system of claim 33, wherein the request element is attributed to the concept of a domain model (ontology). 40. The system according to claim 33, where the relevance of each found fragment is evaluated taking into account the “similarity measure” and / or taking into account the probability of the desired lexical meaning of each query element with the corresponding element in the found fragment.

Images