JPH01502784A - Theory-based system configuration - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed description of the invention] Theory-based system configuration The present invention is a theory-based system. It pertains to configuration, especially logic programming systems (logicpr). gramming system) or knowledge based on logical programming. It is about the display and management system, which consists of multiple modules, data Provability interface with base or theory 1ninterface).

Theory used in this specification % metatheory oryχ assimilation, and probability Basic terms such as availability were first introduced in New York in 1982. Published by Academic Press, L, C1ark and S. -A, Logic Programming edited by Ternlund, 153~ Page 182 Nori, A.

E3owen and R, A, Kowalski's rAmalgamat ing L anguage and Metalanguage in Logic Pro Logic programming in Ho5ton sponsored by grammingJ and IEEE A. Bowe, 1985 Symposium on Mining, pages 48-53. n and T, rA Meta-1level Extens by Weinberg It follows the usage of ion of Prolog J. The theory used here ” refers to logical databases, modules, and worlds (later The terminology for these terms is in Proc, Fir, edited by L. st Int, Conf, on ExpertDatabase System H. Kauffmann and A. Grumbach on pages 61-73 of s. rRepresenting and ManipulatingKnowle dge within 'Worlds'J) that it includes equipment concepts such as You should understand.

This specification also provides basic knowledge of logic programming as possessed by those skilled in the art. I'm assuming. In particular, assume some knowledge of the logical programming language Prolog. (The basic text is the 1984 Springer-Verl Written by W, F, C1ocksin and C, S, Mellish published by ag see rProgramming in PrologJ). knowledge page knowledge-based management systems system) is also assumed (Proc, First cited above). Int, Conf.

on Expert Database Systems) 0 theory base ・Traditional approach to theory-based systems The theory name, the theory, and its interpreter and assimilator. We have used a fixed built-in scheme that defines the relationship between

Therefore, in the paper by B(+wen and Weinberg) mentioned above, the theory Interpreter, assimilation, and attributes are a single system with built-in links to the theory. included in the data theory.

The main drawback of these traditional approaches is that the theory and its associated entities y) Fixed incorporation between (name, interpreter, assimilation unit, attributes, etc.) The problem is that the link is inflexible and implicit. Ru.

According to the present invention, the theory includes a plurality of named theories and is provable. system that supports the theory within the system. Further theory is used to express the association between (“system” theory) A combinator-based programming system characterized by system (knowledge base management system) is provided. The cooperation relationship is preferably that It has an interpreter meta-theory assimilation part, a meta-theory, and an attribute meta-theory about Contains the relationships between theories.

Preferably, this relationship is established within systems theory as a first-order logic clause or can be expressed as any suitable fragment of it, and systems theory A descriptive model (that provides an explicit description of querying and updating) It has a descriptive mode%. This is a good theory (or in any many-to-many relationship between the name Relationships can be defined. In the same way, the inheritance form of the theoretical quantity (1nher itance-type) and other relationships declaratively. ) can be expressed as

For the ability to interrogate the system as a theory, specific entities (such as interpreters) ) can be selected based on the information contained in the question. This ability again Any level of interpretation introduced when dealing with questions An interpreter that can easily support the system (i.e. system theory), and then this interpreter Interpretation by other interpreters found by interrogating other systems be able to. In practice, this iteration, of course, consists of a built-in interpreter It ends with Rita. Finally, Bgwen and Weinb who quoted the system above system by expressing it as a theory of the kind stated in the - sentence of erg. We can obtain an elegant way to treat the latest information descriptively. child This is done by changing any relationships in the system, including changing named theories. requires changes in system theory (if these changes are to be reflected in the system). In fact, multiple verge runs of a system can be The generality of this approach References to these multiple verges within other theories (such as the current system) You can even say that. Therefore, many problems can be solved with this one device. Become.

Description of the preferred embodiment A theory-based system that embodies the present invention will be explained by way of example with reference to the drawings. I decided to do it. The figure is a diagram showing typical relationships between various types of theory.

Prolog To understand the present invention, a general understanding of Prolog is desirable. Pr A very brief summary of olog is as follows. This summary is of course accurate. Far from being. If you want accuracy, use C1ocksin quoted above and You should refer to standard texts on Prolog, such as Mellish's book. Must be.

Prolog is a descriptive language created primarily for symbolic, non-numeric calculations. A Prolog program consists of a set of fact and rule statements. ing. For example, most of the programs correspond to the Fortran program structure. None.

The final element of Prolog is a constant. Of course, Prolog also has variables. However, these are (literally) “ultimately” instantiated as constants. iate) (“replaced by”). Constants usually start with a lowercase letter. (for example, john), and variables are written with the first letter capitalized (for example, john). baX or Father).

The standard factual statement of Prolog is parent (j　ohn　 .

5usan). This means that john is the parent of 5usan. It can be taken to mean something like that. questions or goals in the same form as factual statements , the question symbol before? Add. Therefore, 'i' parent (john.

Child) is a question. The Prolog system understands the question and variable (in this case Child, whose first letter is capitalized) (which can be identified as variables) and find the constant that makes the question true. Trying to find the "answer" to the stated goal. (In this case, the variable Child is exemplified by the constant 5usan. ) A fact statement can take any number of arguments (ar ) and any of them can be variables. . If the fact statement is a question, the system Try to find a constant for (Goals or questions are composed of sentences For example, a variable can satisfy two or more question statements together. There must be. ) More general relationships are expressed as rules, which are the same as fact statements. Consists of statements that have a general form. To give an example of a relationship, grandfather (X, Y) knee fatber (X, Z).

father (Z, Y) - father (X, Z).

mother (Z, Y).

(Here the comma indicates a conjunction, so for example father (X wZ ) and father (Z, Y) must both be true, but semico Ron represents a disjunction, so it is important to note that Ron is a grandfather. ) can be found on both the paternal and maternal side. ) The system satisfies the goal or question. Rules can be used to draw logical conclusions when trying to add. exchange In other words, if we find that the left-hand part of the rule satisfies the question, then the right-hand part Parts can be assembled into one or more sud-queries. Ru.

A fundamental feature of Prolog is that rules can be recursive.

A simple example is ancestor (X, Y) knee parent (X, Y) .

ancestor (X, Z) knee parent (X, Y) .

ancestor (Y, Z) The second rule is defined in terms of ancestral relations. This kind of recursive relationship has the same must always be accompanied by another non-recursive rule for the same relation. here , the first rule defines another necessary ancestry relationship. (Return the danger of infinite retreat) There are other restrictions that must be observed if they must be avoided. ) theory Graphical illustration of the relationship between theories The diagram shows a hypothetical system that includes a variety of different theories and the relationships between them. This figure shows a fragment of the chemical system. (This is a discussion An example of a system cast for the target architecture. It is important to understand that it is not the architecture itself. )related The main formats are shown in this example. It gives a taste of the wind. There are four different types of relationships between theoretical quantities. Interpretation The true relationship between the two It is a linkage of theoretical quantities. These are I, A3. At, and It is indicated by a different type of line, denoted as T.

The first important point to note about the example is the unifying effect of the architecture. S All data and metadata in the stem are contained in the theory, which is a collection of theories. is influenced by the conceptual and structural roles of various data pieces.

Three theories from 10 to 12 - Employment Theory 10, Parts Theory 11, and Supplier Theory Theory 12- are both equivalent to a conventional database of data on key facts. employment Theory of Use 10 is primarily a collection of facts about various people. This is the employment Not only the details, but also various simple relationships between them (such as spousal relationships and parent-child relationships). Contains eel).

Parts Theory 11 is primarily a collection of facts regarding factory parts and components. Supplier Theory 12 is primarily a collection of facts regarding suppliers of various raw materials for factories. Ru.

Employment theory 1O is interpreter theory 13, assimilation theory 14, and attribute theory 15. It is equipped with Employment theories can include rules other than statements of fact. Therefore, the interpreter 13 has a complete theorem prover (full th eorem prover) must be installed. Assimilation Department Theory 14 Ensure that new data added to the employment theory does not violate specified constraints. It is a special assimilation department that recognizes One form of constraint concerns the nature of the data being added. It is something that For example, it is permitted in its theory to It must not be inconsistent with group data such as de-data. The allowed data groups are , defined by Attribute Theory 15, which contains all the metadata about employment theory. It will be done. Therefore, to perform this check, the assimilation part refers to the attribute theory (results ) is necessary. This is an example of theory-theory linkage. parts theory 11 and supplier theory 12 are both factlike in nature. Therefore, each of them takes this into account, and instead of performing a complete theorem proof, An interpreter can be provided that performs fact searching. Therefore these two The theories have a common interpreter 16. in the same way they are common This means that no rules are added to any theory. This is something to check.

In addition to the three factual theories 10-12, there is genealogy theory 19, which is mainly to define complex genealogical relationships such as ancestry and cousinship. It consists of a set of rules, many of which are therefore recursive. This principle The theory is equipped with a special interpreter 19, which interprets recursive rules. , and if you don't do it that way, you'll have problems that come with such a rule. It is designed to limit the explosion of possible deductions. genealogy theory itself is one theory/another theory chain that refers to simple human relationships included in employment theory. , is a different example.

Returning to employment theory 10, its assimilation theory 14 is also another form of the “meaning” of data. constraints on the formula, such as whether the new employee relationship being added is Consider whether it is consistent with knowledge about relationships held in diagram theory. Assimilation Department The theory will have an assimilation section that requires the results to be verified by genealogy theory. Another example of theory-to-theory linkage.

Some theories have certain other theories declared as interpreters in the system. Note that there are some theories that do, but there are other theories that do not. the latter The theory is that the underlying interpreter of the system is interpreted by the printer. This is a Prolog or Prolog-like installation. Interpreter, depth-first left-to-right Perform a backtrack theorem proof (although another form of theorem proof is fine). given theory There is, in fact, always a chain of interpreter theories, but this does not necessarily apply to embedded It must be terminated by an interpreter (not shown). embedded interpreter The data may or may not be preceded by one or more specific interpreter theories. Ru. In the same way, each theory (meta-theory interpreter, assimilation part, and attribute theory) theory) has an assimilation part meta-theory and an attribute meta-theory. Therefore, this Correspondingly, there are chains of assimilation part theories, attribute theories, and these chains are simply specified There is also a built-in assimilation section that adds or removes items, and a trivial that is also a wall theory. It is concluded by the built-in attribute theory.

In addition to the theories mentioned so far, the data in the normal database society system is There are theories that include information that is not processed. One of these theories is the motion recording theory ( log theory) It was 20, and this was the development of the system over the entire period. and dynamic information about usage. The other is question interface theory 2. It is 1. This is a standard database query language that can be used to query any other theory in the system. can be converted into a form appropriate for the question.

Of these latter theories, the system theory 22 is the most popular.

This theory details all the theories in the system, including itself, and the relationships between them. (preferably, each of these relationships should be defined by an attribute theory and an assimilation part theory). , and hold as a set of quaternary relations associated with interpreter theory).

There are four types of relationships that appear in systems theory. i.e. theory vs. theory name; and three corresponding to the above relationships (theoretical name vs. interpreter name, theoretical name vs. assimilation part name, and theory name vs. attribute theory name). System theory is a drawing Unable to express properly. This is because it exists as a clear theory. but it also contains all other theories and indeed itself.

The theory-theory linkage described above is not a relationship and is not elaborated in systems theory. It should be noted that These are simply proofs from one theory to another. It is only realized as a clear invocation (invocation).

Each relationship in this grid can be a many-to-many relationship. For example, a given theory is It is possible to have more than one interpreter. (In such cases, this theory to determine which interpreter is appropriate each time is called - perhaps Based on the goal to be proven - theory needs to be done within the system. )these All relationships are dynamic. Successive barge runs of systems theory examine existing relationships. It can be updated and new relationships can be added. The relationship is explained by systems theory. It does not necessarily have to be designated as a real thing. The relationship can be similarly specified by arbitrary sloppiness rules. can be done. The only fixed relationship is that systems theory interpreters are built-in It means that it is an interpreter.

For each theory, systems theory describes the relationship between that theory and other theories with which it is related. maintains relevant statements. In other words, all relationships between theoretical quantities are based on system theory. By the logic, it is expressed explicitly in the form of a logical statement of -a. relationship status Since it is convenient for statements to define various individual forms of collaboration and relationships, The form of the relationship of a given theory to other theories is automatically determined. In other words, theory and the interpreter theory, the assimilation part theory, and the attribution theory. There is a person in charge. However, the differences between the various types of relationships depend on the structure of the system itself. A relationship of the form has a specific location (first, second, etc.) in the relationship statement. It is only defined to the extent that it exists. To show the relationship with interpreter theory where to show the relationship with assimilation theory and where to show the relationship with attribute theory? The choice of which location to use is arbitrary and is largely a matter of the interpreter. But of course the choice, once made, must be the same for all theories. Lana (ゝ.

The method of maintaining the relationship between various theoretical quantities in the current system is based on the relationship between theoretical quantities and the theoretical change the number over time. For example, do you want to examine the effects of various changes in exchange rates? Maybe. For this purpose, a financial theory that defines the manner in which discounts are determined. (not shown). A series of questions can be directed to this financial theory, These are a series of individual discounters, each considering one possible exchange rate. mount theory (not shown) may be constructed. Systems theory These theories operate to continuously manage the scout theory, but of course not consistent with each other. (A simple example is when updating a theory. The system includes old theories and new (updated) theories. ) Therefore, the system corresponds to possible different exchange rates or consecutive time periods under consideration, It exists in quite a number of separate versions, and the appropriate one for your particular question. -You have to choose a scene.

Detailed description of this system The above description is relatively brief and also covers the specific possibilities of the system's abrikushi rewards. This shows some combinations. A more precise and abstract description of the system is as follows. It is.

The system is modularized into separate theories for storing, retrieving, and updating knowledge. It can be considered a supporting architecture. Each theory in the system is In relation to the other three theories that take on a corresponding role in the methodological stance of ing. All three are metatheories with respect to their theories, and this corresponds to These meta-theories are object theories. First, the interpreter meta-theory However, this defines a provability relation for the target theory and treats the question as a theorem proof. We will provide you with satisfactory support. The interpreter controls and makes available the inferences within the subject theory. Explicit, domain-specific (domain-specific) case object theory clauses Support characterization. Second, there is the assimilation department meta-theory, which is the underlying knowledge. Extending the logical update primitives provided by the base management system to provide syntactic , forcing semantic constraints on the updating of the object theory.

Third, there is an attribute meta-theory, in which the interpreter and assimilation part reason and Expressing properties of object theory clauses, such as formal constraints, used to control updates .

By explicitly expressing this theory-specific meta-knowledge, several kinds of syninterblitter and assimilation parts are created. can be shared by different subject theories.

Since the interpreter and assimilation part are theories themselves, they are also meta-theoretical. It may also be related to theory. However, for efficiency reasons, in practice only one or Only two levels of interpretation are used. Recursion is a special built-in interpreter theory and concludes with the assimilation theory.

In this architecture, the entire knowledge pace is driven by a particular theory, system theory. This is only distinguished by its role. this system A theory describes the name of the theory, the relationship between the theory itself and the 70 meta-theories. this In defining relationships, the full power of Horn clause logic can be exercised. Ta For example, a named object theory can have many interpreters, and a specific Interpreters can be used with many object theories. Theory and 70 meta-theories The ability to express many-to-many relationships between can make a judgment as to whether it is appropriate or not.

This is a fixed combination of static amalgation. edassociation). The system is itself a theory. By including special axioms, for example, you can search for a named theory and can be easily extended to define a variety of ways to select a meta-theory. .

All operations related to theory, including updates, are descriptive, so there are no side effects. . If a named theory is updated, a new theory will always be generated. Both systems Since it is part of the theory, new systems will also arise as a result. Therefore, all Access to all virgins of theory is potentially possible. This system The update model we are using is fully descriptive. What does this mean for theory? Updating (adding to or subtracting from a theory) is also meant to generate a new theory. Taste. Both the new verge collection and the previous verge collection have no references to them. You can access it as long as it exists. This is a very small amount of time and space This can be achieved at a overhead cost.

Whenever you ask a question during a designated theory, you must retrieve that theory from the system. Must be. This is a predicate interface that is standard for all theories. This is done by asking questions to the system using This interface It operates on an underlying knowledge base management system (KBMS) and uses a theoretical name and a set of goals. Take the rule: Kbms-prove (System, TheoryName , Goals). Systems theory is the current version of the named theory and its implementation. You will be asked to find a name for Tarpurita. 2 for the named theory If you have more than one interpreter, perform inference within the system to extract from them. Make a choice. The KBMS then allows the interpreter to resolve the goal. I will try to prove that. (If the interpreter is required for this interpreter If so, recursion is performed until the built-in interpreter is reached. ) individual principles A theory-specific interpreter provides syntactic and semantic information appropriate for managing reasoning. Constraints can be used to tailor proof techniques to specific areas of various theories. , the efficiency of proving goals increases.

The main use of attribute theory is to separate specific theory-dependent information from the interpreter and assimilation part. The goal is to be able to do so. This makes the synterblita and assimilation part a particular theory. can be made relatively unrelated and therefore applicable to even more theories. Become.

All assimilation is done via standard interfaces, such as the standard question interface. and proceed.

Kbms update (System, TheoryName* Update es.

New System).

In a similar way to interpretation, each assimilation unit can be understood and transformed into an expression of the theory being updated. Must be able to do it. Updates are specified as a list of axils - axils Ultimately, everything is a primitive descriptive addition/deletion operation with respect to the theory being updated. is converted to The simplest form of the assimilation section simply performs additions and deletions like this. Ru. The more elaborate assimilation part is a (though prepared to ignore the inevitable contradictions) or a completely updated theory. Check the consistency of the form. In either case, in the unlikely event that Preparation to backtrack and restore the theory to its original state in the event of an illegal update attempt is being done. K, as in the case of interpreter theory, is consistency checking a theory? The assimilation part theory can be applied to two or more theories by separating it from the assimilation part theory and putting it into the assimilation part theory. The possibility of being able to do so becomes even greater.

In fact, the system contains a description of all the relationships of theoretical quantities and provides the required rationale. The question was first directed to systems theory to find a theory, but no theory can be arrived at. It includes all kinds of theories. Therefore updates to the theory are to the system. Since the system itself is a theory, it is also an update to the new system (or or systems theory) occurs every time an update occurs. Previous barge monk of the system is available as long as it is referenced.

The system itself also has its own assimilation department. Typical work of the system assimilation department When a new theory is being added to the system, all of its assigned meta-theories The best thing to do is to check that it belongs to the system.

Interpreter theory backtracks through a series of theories with inheritance relationships means can be incorporated. One form of inheritance is simply theory selection, and goal is proved by members of a set of theories that have an inheritance relationship. The power and elaboration of inheritance The KM shape can be easily imported into the theory interpreter. For example, When a tarprita fails in its attempt to satisfy the goal of a named theory, the previous Rather than backtracking to the goal, instead the system has current theory inherited from it. It asks you to find the name of a theory that satisfies the current goal of its ancestor theory. try to make it happen Because of this elaboration, each individual goal is Bugor can be tried within each theory.

The present invention provides flexible query (proof) for logic programming systems (KBMS). It is used in realizing the possibility) and assimilation (update) interface. Therefore, the invention is implemented using the Prolog program fragment that implements this facility. can be expressed. - Such a program with statements of order logic thus forming detailed embodiments of the invention. (The reality of theory using Prolog) It's actually quite simple, so I won't explain it here. ) The following explanation describes theoretical names in terms of interpreter names, assimilation part names, and attribute names. This is done in the form of a specific predicate theory (system theory).

For system theory, change the theory name to TheoryTheons *Interface. -preta theory name I n t N a m e, assimilation theory name A s s N 5 positions associated with a m e and attribute theory name AttrName It is a predicate theory with . I n t N a m e is the provability of the theory Name the theory that defines the relationship of forces. Similarly, A s s N a m e is logical. Name the theory that defines the relationship between the ability to update the theory. AT t r N a m e is the name of the theory that holds descriptive information about (the attributes of) the theory. IntN The actual names represented by ame, AssName, and AttrName The theory is found when the TheoryName is given. We must find out by questioning systems theory in exactly the same way as Must be. Any level of interpretation can be introduced into this membrane property, and any relationship can be defined. Its expressive ability is used to represent the system Controlled by the ability of logic (predicate calculation) It is only limited.

A formal definition of the meaning of "proof" and "update" according to first-order logic is as follows. Ru. Clauses that define the conditions under which a particular coordination applies are expressed within systems theory. be done. Systems theory is used in the first stage of the query and update process, and is used in the third statement. It is specified by the word (prove) and the fourth predicate (update).

prOVe (System, system, Goal s) Knee 1nt erpret Theory (System, Goals).

probe (System, prologInt, 1nterpre t (TbeoryName, Theory *AttrName, G oals) Knee 1nterpret Theory (Tbeory, G oals).

prove (System, TheoryName t Goals) : -prove (System, theory (TheoryName, Theory.

IntName, AssName, AttrName).

prove (System, Intname, 1nterpret (TheoryName *Theory e AttrName, Goa ls) -update (System, TheoryName + C1uses, NewSys tNewTheory) knee prove (System, system, theory ryName, Theory + IntName, AssName, AttrName)) *prove (System, AssNa me, assimilate (System *Theory, At trName, C1auses, NewTheory) *upda tesystem (Systems TheoryName, Theo ry　NewTheory　＊　NewSystem).

The meanings of these predicates, variables, and functions are as follows.

Here, the system theory System is a command relation (i.e., which name means which theory, Define the name of the interpreter, assimilation unit, and attribute. In the example shown here , these clauses limit the system's interpreter to Prolog K.

update (System, TheoryName, C1ause s, NewSystem yNewTheory), the clause C1auses is Added to the theory named TheoryName in System. system including NewTheory and NewTbeory. True if a new barge system NcwSystem is created. NewSyst em is a function such that TheoryName now names the new theory (name). On the other hand, it is made from 5ysten.

IntName, AssName* and AttrName are respectively represents the interpreter, assimilator, and attributes (see above) of the theory.

1nterpret (TheoryName, Theory + Att rName, Goals) is the name TheoryNam In a theory with attribute theories named e and AttrName It is true if it can be proven.

assimi 1ate (System, TheoryName, T heory, AttrName.

C1auses, NcwTheory) is AttrName and system management. When a theory is given, clause CI ause5 can be added to the theory, which is new. is true if it yields a new theory NewTbeory.

1nterpretTheory (TheoryName, Goals ) is provable if the goal GOa l S is provable using Prolog's proof procedure. It is true.

updatesystem (System, TheoryName, Theory, New Theory.

NewSystem) is a system theory NewSystem is a theory 01dThe In place of ory, associate NewTheory with theory name TheoryName. It is true if it differs from the system theory System by attaching it.

■ international search report 1m@+vllaaalmu1111Jl16Pu*, pc7/Cmi==miεε /CCCC5 International Search Report GEIεεDOOO5 SA　20193

Claims (5)

[Claims] 1. Contains multiple named theories and supports provability computer-based logic programming system (knowledge base management) system), said system uses another theory (the "system" theory) to Association between theories within a stem and their names elationship). 2. That affiliation includes the theory's relationship to its interpreter meta-theory. The system of claim 1, characterized in that: 3. The partnership is the assimilated part of the theory (meta-theory). any of the preceding claims characterized in that it includes a relationship to surrounding system. 4. A destination characterized in that the affiliation includes a relationship of the theory to its attribute meta-theory. Any claimed system. 5. The relationship is expressed within systems theory by a clause of first-order logic or an appropriate fragment thereof. system theory explicitly describes queries and updates within the system. Having a declarative model that provides a description A system as claimed in any preceding claim, characterized in that: