JPH0743727B2 - Pattern matching processor - Google Patents

Description

TECHNICAL FIELD The present invention relates to a pattern matching processing device for performing pattern matching processing using a concept tree, and more specifically, data conversion of a concept tree represented in a tree structure. The present invention relates to a pattern matching processing device that uses data converted to and for pattern matching.

Examples of fields of use of the present invention include an expert system that performs pattern matching processing, a database system, a program reuse system (CASE tool), and a search system.

<Prior art> Currently, expert systems have been put to practical use in various fields, but the process is to perform pattern matching of the input problem with the condition part of the production rule, which is internal knowledge, and to find the conclusion part of the successful rule. It repeats the mechanism of execution.

This method requires an exact pattern match, so
Only the ones that succeed in a perfect match are processed, preventing flexible reasoning. This also applies to database systems and the like.

Therefore, the pattern matching method using the concept tree is drawing attention. There are various methods for this pattern matching method, but basically the concept of two elements that are the subject of pattern matching (the side that performs pattern matching is called the target element and the side that is pattern matched is called the base element). The success / failure of pattern matching is determined by checking whether or not the keys on the tree are on the same concept tree. If there is more than one base element, the base element with the smaller distance in the concept tree is set as the appropriate pattern match for the target element.

<Problems to be Solved by the Invention> In the pattern matching method using the concept tree described above, the range of pattern matching is expanded to the same concept tree, and the number of base elements for which pattern matching succeeds greatly increases.

That is, there are many combinations of base elements and target elements for which pattern matching succeeds, and many of them compete with each other. For this reason, the ones with a short distance on the concept tree will be set as an appropriate set, but the number of sets for which pattern matching succeeds will be huge, and the success / failure of pattern matching will be determined, and the concept tree It took an enormous amount of time to calculate the distance.

In view of this problem, an object of the present invention is to convert a concept tree into data suitable for pattern matching in advance, and to use the converted data to determine the success / failure of pattern matching and calculate the distance. The purpose is to provide a pattern matching processing device.

<Means for Solving the Problem> Means for inputting a concept tree consisting of concept words, and the input concept tree from the terminal concept word to the root word of the concept tree for each terminal concept word of the concept tree A conversion means for converting an intermediate concept word existing in the above and the concept distance between the terminal concept word and the word of the root of the concept tree into data, and when performing pattern matching of the terminal concept word, the data by the conversion means. While obtaining the lowest common concept by performing matching from the higher order of the intermediate concept word of each terminal concept word with reference to, the number of matches of the intermediate concept word and the concept of each terminal concept word and the root word of the concept tree Pattern matching means for obtaining the distance between each terminal concept word and the lowest common concept word and the matching distance between each terminal concept word based on the tree distance.

<Operation> According to the present invention, when the concept tree is input, the concept tree is input for each terminal concept word of the concept tree, and the intermediate concept word existing from the terminal concept word to the root word of the concept tree. , The concept tree distance between the terminal conceptual word and the root word of the conceptual word is converted into data describing.

Then, when performing the pattern matching of the terminal concept words, the pattern matching is performed based on the converted data.

That is, the lowest common concept is obtained by sequentially collating the higher intermediate concept words of each terminal concept word, and the lowest common concept is obtained by the number of matching intermediate concept words and the concept tree distance of each terminal concept word. The distance to the concept and the matching distance between each terminal concept word are obtained.

<Example> Hereinafter, the present invention will be described in detail with reference to the drawings.

FIG. 1 is a block diagram of a pattern matching processing device according to the present invention.

In FIG. 1, 1 is input means for inputting a concept tree or a problem, 2 is output means for outputting the result of pattern matching to a file or another system, and 3 is pattern matching means for performing pattern matching using the concept tree. Reference numeral 4 is a display unit for displaying the pattern matching result on the display for the user, 5 is a conversion unit for converting the concept tree into a form suitable for each unit, 6 is a knowledge management unit for managing the concept tree, and 7 is the above. It is a control means for controlling each of the means described above.

FIG. 2 is a conceptual diagram of functional blocks according to the present invention.

In FIG. 2, reference numeral 8 is a module for taking in a problem by dialogue with the user. Reference numeral 9 is an editing module for acquiring a concept tree from a user, and 11 is a conversion module for converting the acquired concept tree into a data expression for a knowledge management module or a data expression for a pattern matching module. Reference numeral 10 is a knowledge management module that manages a concept tree, and 12 is a pattern matching module that performs pattern matching on the input problem using the concept tree.

The editing module 9 will be described.

Figure 3 is a part of the concept tree created based on the functional terms used in VE (Value Engineering). The concept tree is an editing module 9
Input is performed in the format desired by the user, for example, in the tree structure as shown in the figure, using the graphic interface included in. In the concept tree, the root of the tree is “change” and the end is a word such as “split” or “transform”. The last word is used as a key for pattern matching.

The conversion module 11 converts the input concept tree into the following data structure (hereinafter referred to as edit data).

[(Root, [change]), (change, [shape change, state change, content change]), (shape change, [divide, transform]),
(State change, [transformation, variation]), (variation [hold, transition]), (hold, [temporary storage, store, indicate]), (content change, [content conversion, deletion]), (content conversion, [ Calculation,
Change])].

This is a data representation managed by the knowledge management module 10. This data structure expresses a concept tree as a pair of a parent and a child list. The parent is represented by one element, and the child is represented by a list of elements. That is, it is expressed in the format of (<parent>, [<child 1>, <child 2> ...]). <> Represents one word. Explaining the data structure concretely, the root of the concept tree is only "change", and the children of "change" are "shape change", "state change", and "content change". Although there is only one root here, multiple roots can be specified.

The transformed concept tree is then transformed into the data structure of Table 1 when used in the pattern matching module 12.

1st element 2nd element 3rd element (temporary storage, 4, [change, state change, change, hold]) (store, 4, [change, state change, change, hold]) Table 1 The above data structure is a concept “Temporary memory” and “store” on trees
The converted data of the element (hereinafter referred to as compiled data) is expressed. In each data, the first element (temporary storage, storage) is the concept name at the end of the concept tree, which is the target of actual pattern matching, and the second element (4,
4) expresses the distance between the end concept name and the root of the concept tree by a natural number, and is called a concept tree distance. That is, this value is the same as the number of elements (the number of elements) of the third element described later. The third element (change, state change, ...) Is a list of concepts contained between the root and end concepts,
It is called an intermediate concept list.

On the contrary, the conversion module 11 also converts the compile data into the edit data, or converts the edit data into data for the edit module. In this way, the conversion module 11 converts data between the modules.

The knowledge management module 10 manages the edited data in which the concept tree is converted as described above. Management contents include output to a file and reading from a file.

The dialogue / reasoning module 8 creates a question. To create a problem, a pair of elements to be compared (hereinafter referred to as an element pair) is created from two sets to be compared. Next, for each element of the pair, use the knowledge of the element and the key on the concept tree (managed by the dialogue / reasoning module) to find the key (word, end of the concept tree) registered in the concept tree. It will be taken out and the pair (hereinafter referred to as a registered word pair) will be created.

For example, a set A of office components (a storage, a meeting room, a book, a manager's seat, OA equipment, a section, etc.) and a set B of components of a microprocessor (register, ALU,
From the bus interface, program counter, instruction decoder, control timing generator, IR, etc.), create element pairs such as (storage, register) and (storage, ALU), and store registered word pairs in storage or registers. It is created from knowledge about keys on the conceptual tree.

For example, in the case of an element pair (storage, register), a registered word pair called (store, temporary storage) is created.

The inference module 8 performs such processing and sends the created set of registered word pairs to the pattern matching module 12.

In the pattern matching module, the success / failure of the matching of each pair is checked, the concept name for which the matching is successful (hereinafter referred to as the lowest concept name) is acquired, the distance between the terminal concept word and the lowest common concept (the lowest common concept). And how far each registered word is on the concept tree) and the matching distance on the concept tree (how far the two keys of the registered word pair are on the concept tree). Do and reasoning module 8
These data will be returned to. Inference module 8
If not, these processes will be performed by the interaction with the user.

In the pattern matching module 12, the set of registered word pairs sent is compared using the above-mentioned compiled data. The processing algorithm is shown in FIG.

This algorithm is applied for each registered word pair. This algorithm is a pair of elements (storage, register)
An example of pattern matching of (stored, temporary storage) which is a registered word pair of will be described.

First, in step 16, the compilation data of each registered word pair is selected. For example, (store, temporary memory)
For the pair, the compile data in Table 1 is selected from all the compile data of the concept tree.

Next, the comparison of the two compile data selected in step 17 is collated from the root of the concept tree. In the example, “change” of “store” and “change” of “temporary storage” are collated. In this case, "restore""statechange"
And “temporary memory” “state change” are collated and succeed. And if the processing in step 19 fails in matching,
The process is repeated until there is no intermediate concept to match (the process of step 21 becomes no). Step 20 if verification fails
Then, the lowest concept name is acquired and the matching distance on the concept tree is calculated. Obtaining the lowest concept name is a step
The concept name has been successfully collated in the process of step 19 before the process of 19 fails. The distance between the acquired lowest common concept and each terminal concept word is the concept tree distance, which is the second element of the compiled data, minus the number of times the processing in step 19 was successful. The matching distance on the concept tree is the sum of the distances between these lowest common concepts and each terminal concept word. When the concept of collation disappears, the same process as step 20 is performed by step 22.

Then, in step 23 and step 24, a flag indicating that the matching is successful is added.

In this example, the collation does not fail and the concept to be collated disappears, so that the process of step 22 is performed and the lowest concept name “hold” is obtained. Further, the distance between the lowest common concept word “hold” and the end concept word “temporary storage” is “1” which is obtained by subtracting the collation count “3” from “4” which is the second element of the compiled data. Similarly, the distance between the lowest common concept word “hold” and the end concept word “store” is also “1”, and the matching distance on the concept tree is the sum “2”.

When the processing for each registered word pair is completed, the data is returned to the dialogue / inference module 8 and the result of the pattern matching is displayed to the user or used as the data for the inference module. .

In this way, pattern matching success / failure processing can be performed at high speed by examining the concept tree from the root word. That is, the failure of the pattern matching on the concept tree only needs to examine the first one of the intermediate concepts. Further, the matching distance can be easily calculated only by managing the number of successful matchings.

<Effects of the Invention> According to the present invention, the processing speed can be increased by previously converting the concept tree used for pattern matching into data suitable for pattern matching. The larger the concept tree, the greater the effect.

Further, it is possible to immediately obtain the distance between each terminal concept word and the lowest common concept and the matching distance between each terminal concept word based on the number of matches of the intermediate concept words and the concept tree distance of each terminal concept word. It is possible. The cost of this calculation is much smaller than that of calculating the matching distance from the end of the concept tree, and when multiple matches are found, it is useful when the one with the closest distance on the concept tree is determined as an appropriate set. Is.

Furthermore, when the pattern matching is successful, the lowest-ranked concept that succeeds in matching can be acquired at the same time.

In addition, when the user edits (inputs) the concept tree, the concept tree can be edited in a form that is easy to edit.

[Brief description of drawings]

FIG. 1 is a block diagram of the device of the present invention, FIG. 2 is a conceptual diagram of functional blocks, FIG. 3 is an example of a conceptual tree, and FIG. 4 is a pattern matching algorithm. 1 ... Input means, 2 ... Output means, 3 ... Pattern matching means, 4 ... Display means, 5 ... Conversion means, 6 ... Knowledge management section, 7 ... Control means, 8 ... Dialogue / reasoning Module, 9 ... Editing module, 10 ... Conversion module, 11
…… Knowledge management module, 12 …… Pattern match module.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (56) References JP-A-63-282837 (JP, A) JP-A-62-194533 (JP, A)

Claims (1)

[Claims] 1. A means for inputting a concept tree made up of concept words, and an intermediate part of the input concept tree existing from the terminal concept word to the root word of the concept tree for each terminal concept word of the concept tree. A conversion unit that converts the concept word, the terminal concept word, and the concept distance between the root word of the concept tree into data describing the concept word, and when performing pattern matching of the terminal concept word, refer to the data by the conversion unit. While obtaining the lowest common concept by performing matching from the higher order of the intermediate concept words of each terminal concept word, the number of matches of the intermediate concept word and the concept tree distance between each terminal concept word and the root word of the concept tree. A pattern matching processing device comprising: a pattern matching means for obtaining a distance between each terminal conceptual word and the lowest common conceptual word and a matching distance between each terminal conceptual word based on the above.