JP3893600B1 - Base database generation method, base list generation method and apparatus, and computer program - Google Patents

Abstract

A method for generating semantic phase conceptual data for realizing semantic structure analysis processing of a sentence using a computer is obtained.
A method for generating a base database of words by a computer having storage means for storing a word and one or more word sentences explaining the meaning of the word in association with each other, and reading the word from the storage means A step, a step of reading out a word sentence stored in association with the read word from the storage means, a step of dividing the read word sentence by a morphological analysis process, and determining whether the divided word corresponds to a base A step of associating a word corresponding to the base in the determination step as a base for the read word, and a step of recursively performing the above steps until the word corresponds to the base when the divided word does not correspond to the base. According to the base database generation method.
[Selection] Figure 6

Description

  SUMMARY OF THE INVENTION An object of the present invention is to provide a method and apparatus for generating semantic phase concept data used for semantic structure analysis processing of a sentence by a computer, and a base list used for generating semantic phase concept data, and a computer program.

  In recent years, with the development of computer technology, research on computer processing for sentence analysis has been actively conducted. High-precision sentence analysis processing by a computer is expected to be applied to a system that can search for similar sentences. If the analysis of the sentence structure is not based on word level matching processing but the analysis of the semantic concept of the sentence itself can be realized by a computer, it will be a hit only by including the search keyword as in the search result by the conventional search system. This makes it possible to eliminate only the fact that the search text is close to the meaning. By applying the semantic concept analysis process using a computer to a search system in this way, it is possible to obtain a search result that cannot be obtained by a conventional word keyword search. In addition, the process of automatically summarizing a document composed of a large number of sentences (for example, a paper) with a predetermined number of characters using a semantic concept search process using a computer can be performed by simply specifying the number of characters. Application to a system that can be performed while maintaining the similarity is also expected. If a system that performs sentence analysis processing by a computer as described above can be realized, a great contribution to industrial development can be expected. For this reason, a highly accurate analysis technique and a computer program for realizing it are necessary.

  There have been various inventions related to sentence analysis technology using a computer. Among these sentence analysis techniques, a well-known technique is a word thesaurus dictionary. The word thesaurus dictionary defines synonyms and synonyms of words, and highly accurate analysis results can be obtained from data (word groups) included in the word thesaurus dictionary. Therefore, it is necessary to generate a highly accurate word thesaurus dictionary. An example of a method for generating this word thesaurus dictionary will be shown. It is known to analyze a sentence contained in a huge amount of document data (for example, hundreds of thousands of newspaper articles) and analyze a relation (such as the degree of coincidence) between words contained in this sentence. (For example, refer to Patent Document 1). The invention described in Patent Document 1 is characterized in that a statistical method is used in a word thesaurus dictionary data generation method. For each word contained in document data such as newspaper articles, the frequency of occurrence in the document data is measured, and the probability (co-occurrence probability) that each word appears in the same document is calculated. Then, the semantic distance between words is calculated by vector calculation. Data obtained by digitizing the structure based on the distance between the words is used as a word thesaurus dictionary. When the word thesaurus dictionary generated in this way is used, it is possible to specify a document that is statistically close by a word included in the document to be analyzed. Therefore, it becomes possible for a computer to determine another word having a structure similar to a word included in a certain document and a document constituted by the word.

JP 2001-331515 A

  The data (word thesaurus dictionary) generated using the above invention naturally has a deviation (error). That is, since the appearance frequency of words appearing in the document data used for generation is affected by the attribute of the document, a deviation occurs. If such a document attribute is a newspaper article, it is difficult to exclude it because it relates to the content of the article itself. Therefore, when the above invention is used, it is necessary to normalize the data to be generated in order to correct the deviation. However, it is impossible to eliminate the deviation at all even if normalization is performed, and the contents and relations of the document data (for example, newspaper articles) used to obtain the appearance frequency of the word are reflected in the quality of the information to be statistically processed as it is. Will be. In addition, since the meaning of each word itself is not considered at all, statistically accidentally related semantically different words are recognized as “related words” and thereby a word thesaurus. A dictionary will be generated.

  When sentence analysis processing is performed using the word thesaurus dictionary generated using the statistical method as described above, sentence analysis is performed according to the attribute / preference of the document data used to generate dictionary data (word thesaurus in the above example). The accuracy of the result will be reduced. It is difficult to improve the analysis accuracy by the sentence analysis process performed using the above method.

  Therefore, the present invention eliminates the shortcomings of sentence analysis means using statistical methods, and generates a base database that is semantic phase concept data capable of realizing sentence analysis processing with excellent analysis accuracy, and a base database. It is an object of the present invention to provide a method of generating a base list to be used, an apparatus thereof, and a program.

  The present invention is a method for generating a base database of a word by a computer including a storage unit that stores a word and one or more word sentences that explain the meaning of the word in association with each other. A step of reading a word from the storage means, a step of reading a word sentence stored in association with the read word from the storage means, a step of dividing the read word sentence by a morphological analysis process, and the divided word Determining whether or not the word corresponds to the base, associating the word determined to be the base in the determination step as a base for the read word, and if the divided word does not correspond to the base, Recursively performing the above steps until the word falls into the base. And features.

  In addition, a first storage means for storing the word and one or more word explanations explaining the meaning of the word in association with each other, a second storage means for storing the word and the appearance frequency of the word in association with each other, and a third storage for storing the base list A method of generating a base list for use in generating a base database of words by a computer having storage means, wherein the computer reads a word from the first storage means, and reads from the first storage means Reading a reading of the sentence stored in association with the word, a word reading step for dividing the read word sentence into words, a step of registering the divided word in a second storage means, A step of reading out the sentence corresponding to the divided word from the first storage means, and a word dividing process on the read out sentence. , Recursively performing word registration processing and word interpretation data reading processing, and extracting from the words stored in the second storage means as a base, and registering them in the base list of the third storage means And a step of performing.

  According to the present invention, it is possible to clarify the “base” constituting the meaning of each word so that the semantic concept constituting each word can be handled quantitatively. This makes it possible to use a computer for sentence analysis processing using the base, and to perform semantic concept analysis processing that could not be obtained by conventional sentence analysis processing.

  Hereinafter, embodiments of a base database generation method according to the present invention will be described with reference to the drawings. FIG. 1 is a functional block diagram showing a configuration of a base database generation apparatus according to the present invention. The base database generation device 1 is configured by a computer equipped with a base database generation program 2 that executes the base database generation method according to the present invention. Each functional block shown in FIG. 1 is implemented when the base database generation program according to the present invention and the hardware of the computer cooperate to execute processing.

  In FIG. 1, the base database generation device 1 includes a dictionary data storage unit 10, a base database storage unit 30, a dictionary data reading unit 11, a morpheme analysis unit 12, a notation fluctuation correction unit 13, a search information storage unit 14, a base determination unit 15, a base A registration unit 16, an end condition determination unit 17, and a base list 18 are provided.

  The dictionary data storage unit 10 stores dictionary data that stores word data (headword data) that is “headword” and sentence data (word data) that is “word interpretation” in association with each other. An lexical sentence is composed of one or more sentences that explain the semantic interpretation of a word that is a headword. That is, the word interpretation data is composed of sentence data groups that explain the semantic interpretation of the entry word data. The dictionary data storage unit 10 is constructed on an external storage device (such as a hard disk device) included in a computer that executes a program according to the present invention, and stores dictionary data in advance. Dictionary data can read headword data, and by specifying headword data, word sentence data stored in association with the headword data can be read. As long as it can be stored in the hardware means included in the storage method, the storage method, format, and data structure are not limited.

  An example of dictionary data stored in the dictionary data storage unit 10 is shown in FIG. In FIG. 2, dictionary data 100 includes headword data 101, reading data 111 indicating how to read (pronounced) the headword data 101, and sentence interpretation data 102 that is a sentence explaining the meaning of the headword data 101. A completed flag 112. In FIG. 2, each headword data 101 stores a read flag 112 in association with it. This read flag indicates that flag data (for example, “*”) is inserted after the headword generation program according to the present invention reads the headword data 101, and that the program has already processed the headword data 101. Used to show. Since the word interpretation data 102 is stored in the dictionary data storage unit 10 using the entry word data 101 as an index, by specifying one entry word data 101, all of the word interpretation data 102 stored in association with it is specified. It is comprised so that it can read.

  The storage means for mounting the dictionary data storage unit 10 may be an external storage device such as a CD-ROM in addition to the above-described hard disk device. The present invention can obtain the same effect as long as it is an apparatus capable of performing the predetermined reading process on the dictionary data 100 according to the instruction of the computer program according to the present invention.

  The base database storage unit 30 is a storage device that stores a base database generated by the program according to the present invention. The base database stored in the base database storage unit 30 has a table structure in which word data corresponding to the entry word data 101 is used as an index and word data 18a included in the base list 18 is used as a field.

  FIG. 3 shows an example of the base database stored in the storage unit 30. In FIG. 3, the base database 400 has a table structure with a word 401 as an index and a base 402 as a field. The value “1” is stored in the base 402 related to the semantic structure of the word 401, and “0” is stored in the base 402 that is not related. The base database 400 may have any structure as long as it has a structure capable of registering associations between words and bases, and is not limited to the example shown in FIG.

  The dictionary data reading unit 11 performs a predetermined inquiry process on the dictionary data 100. The predetermined inquiry processing includes reading processing of the headword data 101, processing of reading the word sentence data 102 stored in association with the headword data 101, and inserting a value into the read flag 112 of the read headword data 101. Processing for adding the headword data 101 to the searched data in the search information storage unit 14. Also, a process of adding 1 to the search hierarchy information when the interpretation data 102 is read.

  The morpheme analysis unit 12 performs a morpheme analysis process on the word sentence data 102 read from the dictionary data 100 and divides it into word data.

  Here, the morphological analysis process will be described. The morpheme analysis process is a process of dividing a document into minimum units of words called morphemes and determining the part of speech of the divided morphemes. The morpheme analysis unit 12 performs segmentation processing, reading assignment processing, and part-of-speech addition processing on the interpretation data 102, and information on the utilization type, the utilization form, and the basic form regarding the utilization form and the utilization form. The process of giving is performed. The morpheme analysis unit 12 can generate word data that can be used for subsequent processing from the word interpretation data 102. The morphological analysis process can be carried out using software that is already known. As data that can be used in the morphological analysis unit 12, dictionary data including all words included in a predetermined language system as data is prepared, and a word included in a document to be analyzed matches a word in the dictionary data. If there is a match, the word contained in the document is divided into software and the statistical probability method based on a large corpus is used to arithmetically divide it into morphemes to determine the part of speech. There is software that breaks up words.

  The notation fluctuation correction unit 13 performs a notation fluctuation correction process on the word data generated by dividing the word sentence data 102 by the morphological analysis unit 12. The notation fluctuation means that the notation is different even if the pronunciation and meaning are the same. For example, “interface” and “interface” can also be described as “interface”. In order to correct the notation fluctuation, the notation fluctuation correcting unit 13 performs a process of matching the notation of the word data divided from the word interpretation data 102 with the entry word data 101. For example, if the headword data 101 is represented as “interface”, when the divided word data is “interface”, the headword data 101 is converted into “interface”.

  The search information storage unit 14 stores an unsearched list generated by associating the divided word data with the search hierarchy information. A specific example of the unexplored list is as shown in FIG. In FIG. 4, an unexplored list 700 has a table structure with unexplored data 701 and associated search hierarchy information 702. The exploration hierarchy information is information used to count the number of times of reading the word sentence data 102 recursively when the word sentence data 102 is read using the headword data 101 that has undergone the base determination process.

  The search information storage unit 14 also stores searched data (not shown). The searched data is a file in which the headword data 101 and the unsearched data 701 that have been subjected to the base determination process in the base determination unit 15 are accumulated at any time. This searched data is used for avoiding duplicated base judgment processing for the same data. Therefore, the storage format and the like of the unsearched list and the searched data are not limited to the above as long as the program according to the present invention, which will be described later, can handle predetermined processing.

The base determination unit 15 determines whether or not the headword data 101 read by the dictionary data reading unit 11 corresponds to a base by an inquiry process using the base list 18. The entry word data 101 determined to be included in the base list 18 based on the result of this processing is passed to the base registration unit 16. In addition, the base determination unit 15 performs a process of adding the entry word data 101 that has been subjected to the above inquiry process to the searched data stored in the search information storage unit 14.
The base determination unit 15 reads the unsearched data 701 stored in the unsearched list 700 and deletes the unsearched data 701 from the unsearched list 700, and then the unsearched data 701 is stored in the search information storage unit 14. To check whether it is included in the searched data stored in. If the unexplored data 701 is not included in the searched data as a result of the confirmation process, the unsearched data 701 is added to the searched data and the inquiry process is performed. If it is determined that the unexplored data 701 is included in the base list 18 as a result of the inquiry process, the unsearched data 701 is passed to the base registration unit 16. If the unsearched data 701 is included in the searched data in the confirmation process, a process of newly reading the unsearched data 701 from the unsearched list 700 is performed.

  The base list 18 is a file stored in advance in a memory included in the computer, and word data corresponding to the base is stored in a list format. The base is a word that represents a basic concept that constitutes the semantic concept of the word, and is determined in advance by a human. The base list 18 is configured to include one or more word data as a base, and when the inquiry processing is performed using the word data with the base determination unit 15, the result (present / not present) can be determined. If there is, the storage method and the structure of information are not limited to this, and the effects of the present invention can be obtained in the same manner even if other structures are used. An example of the base list 18 is shown in FIG. In FIG. 5, the base list 18 has a file structure for storing one or more word data 18a of a word as a base.

  The base registration unit 16 performs processing for storing the headword data 101 or the unsearched data 701 determined to be the base by the base determination unit 15 in the base database 400. The structure of the base database is as described above. The base registration unit 16 adds the entry word data 101 or the unsearched data 701 to the index (word 401) of the base database, and the same base 401 as the base processed in the base determination process in the base 401 related to the index. 1 is added to the field corresponding to, and 0 is added to the field of the base 402 that is not applicable.

  The end condition determination unit 17 performs an inquiry process as to whether or not the unsearched data 701 exists in the unsearched list 700. If unexplored data 701 is not extracted as a result of this processing (that is, if there is no unexplored data 701), the searched data is cleared and the value of the search hierarchy information is set to zero. Further, the end condition determination unit 17 performs a process for inquiring whether or not there is the entry word data 101 in which the read flag 112 of the dictionary data 100 is blank. As a result of this processing, if there is no entry word data 101 whose read flag 112 is blank, the operation of the program according to the present invention is terminated.

  A base database generation method performed by the base database generation apparatus 1 having the above functions will be described. FIG. 6 is a flowchart showing the flow of processing of the computer program according to the present invention for executing the base database generation method.

  First, step 201 is a headword reading process. In step 201, the dictionary data reading unit 11 reads the headword data 101 at the head position of the storage area of the dictionary data 100 stored in the dictionary data storage unit 10 and adds “*” to the read flag 112. Reading of the headword data 101 is performed in the order stored in the dictionary data 100.

  Step 202 is a basic judgment process. In step 202, the base determination unit 15 performs a process of inquiring whether or not the headword data 101 read out above matches the word data included in the base list 18. If the headword data 101 is extracted from the base list 18 as a result of the inquiry process, it corresponds to the base, and the headword data 101 is transferred to the base registration unit 16. If the entry word data 101 is not extracted as a result of the inquiry process, the entry word data 101 is transferred to the dictionary data reading unit 11 because it does not correspond to the basis. The base determination unit 15 adds the entry word data 101 subjected to the inquiry process to the searched data.

  Step 204 is a base registration process. In step 204, the base registration unit 16 adds the entry word data 101 passed from the base determination unit 15 to the index (word 401) of the base database 400 stored in the base database storage unit 30 and uses it for the base determination process. Then, “1” is added to the base 402 which is the same word data as the word data in the base list 18 and “0” is added to the other bases 402.

  Step 213 is an all headword end determination process. In step 213, the end condition determination unit 17 performs an inquiry process as to whether or not there is a blank in the read flag 112 stored in association with the entry word data 101. As a result of this processing, if there is a blank, the entry word data 101 has not been subjected to base determination processing, so the processing of this program is recursed to step 201. As a result of the above processing, if there is no blank in the read flag 112, the program is terminated.

  Next, processing when the headword data 101 does not correspond to the base in the above step 202 will be described.

  Step 203 is a sentence reading process. In step 203, the dictionary data reading unit 101 performs a process of reading the word sentence data 102 stored in association with the headword data 11 read previously.

  Step 205 is a morphological analysis process. In step 205, the morpheme analysis unit 12 performs morpheme analysis processing on the word sentence data 102 read in step 203. The word sentence data 102 is divided into word data by this morphological analysis process.

  Step 206 is a notation fluctuation correction process. In step 206, the notation fluctuation correction unit 14 performs a correction process for matching the notation of the word data generated by the morpheme analysis unit 13 with the notation of the entry word data 101.

  An example of the result obtained by the above steps 205 and 206 is shown in FIG. FIG. 7 shows an example of a result obtained by performing the above processing on the word interpretation data 106 shown in FIG. As shown in FIG. 7, the analysis result data 600 includes basic form, reading, part-of-speech classification, part-of-speech subclassification, presence / absence of utilization type, type name, presence / absence of utilization form for each word data (morpheme) divided from the interpretation sentence 106 It is constructed by associating its model name. Among the morphemes included in the analysis result data 600, non-independent words (for example, particles and auxiliary verbs) that are morphemes that are not directly related to the semantic concept of words, formal nouns (for example, “ko”), auxiliary words (for example, “is” ”“ Yes ”“ No ”), ie“ electronic computer ”,“ different ”,“ equipment ”,“ device ”,“ between ”,“ connect ”,“ do ”,“ communication ”,“ In step 206, the correction of notation is performed on “control”, “possible”, “device”, and “software”. As a result, “electronic computer”, “different”, “equipment”, “device”, “interval” are performed. ”,“ Connect ”,“ Yes ”,“ Communication ”,“ Control ”,“ Available ”,“ Apparatus ”,“ Software ”as unexplored data 701 in the unexplored list 700 of the exploration information storage unit 14. Remember.

  As a result of performing the base determination process on the word data divided from the word sentence data 102 in the above process, when it is determined that the word data is not further the base, the word data is used again for the word sentence data. Processing to read 102 is performed. Thus, it is necessary to repeat from the sentence reading process to the base determination process until the word data corresponding to the base is reached. Therefore, step 203 is processed recursively. At this time, it is possible to measure the number of recursions by adding 1 to each time the word sentence data 102 is read recursively by setting the search hierarchy information of the word data divided from the first read word data 102 to “1”. The information to make is exploration hierarchy information.

  Step 207 is a search completed determination process. In step 207, the base determination unit 15 reads out the unsearched data 701 stored in the unsearched list 700 and the search hierarchy information 702 stored in association with the unsearched data 701, and stores the unsearched data 701. After deleting from the unexplored list 700, it is confirmed whether or not the unsearched data 701 is already included in the searched data (not shown) stored in the search information storage unit 14. As a result of this processing, if there is no matching word data in the searched data, the unsearched data 701 has not yet undergone the base determination process, so the unsearched data 701 is added to the searched data. To step 208.

  Step 208 is a base determination process. In step 208, the base determination unit 15 performs base determination processing on the unsearched data 701 read in the previous processing. The details of the base determination process are the same as in step 202 already described, and subsequent step 209 performs the same process as the base registration process in step 204 already described.

  Step 210 is a search end determination process. In step 210, the end condition determination unit 17 determines whether or not the exploration hierarchy information 702 (see FIG. 4) read from the exploration information storage unit 14 by the base determination unit 15 in the previous process is equal to or less than a predetermined numerical value. . Here, the predetermined numerical value is “5”. The purpose of limiting the number of recursively read word interpretation processes using exploration hierarchy information is to efficiently implement the present invention, and the predetermined numerical value is not limited to this. It is not an essential requirement of the invention.

  Step 211 is a headword confirmation process. In step 211, the end condition determination unit 17 performs a process of inquiring whether or not the unsearched data 701 used in the above process is included in the dictionary data 100 as the entry word data 101.

  Step 212 is an end determination process. In step 212, the end condition determination unit 17 performs an inquiry process as to whether or not one or more unsearched data 701 exists in the unsearched list 700 stored in the search information storage unit 14. If there is no unexplored data 701 as a result of this processing, the searched data is cleared and the value of the search hierarchy information is set to zero.

  Step 213 is all headword end determination processing as already described.

  A specific example of the processing from step 207 to step 212 will be described. In step 207, the base determination unit 15 stores the unsearched data 701 (“electronic computer” in FIG. 4) stored at the head position of the unsearched list 700 stored in the search information storage unit 14 and the unsearched data 701. The search hierarchy information ("1" in FIG. 4) stored in association with is read out. Next, the base determination unit 15 performs an inquiry process as to whether or not the read unsearched data 701 is included in the searched data.

  In the next step 208, the basis determination unit 15 performs a basis determination process for the word “electronic computer” that is the unexplored data 701. Since “electronic computer” is not included in the base list 18, it is determined that it is not a base (N in 208). In the next step 210, the end condition determining unit 17 determines whether the value of the search hierarchy information is “5 or more” which is the end condition. Since the exploration hierarchy information is “1” as described above, the termination condition is not satisfied (N in 210). In the next step 211, the termination condition determination unit 17 performs an inquiry process as to whether or not the “electronic computer” exists in the entry word data 101. Since the dictionary data 100 includes “electronic computer” as the entry word 107 (Y in 211), the word data “electronic computer” is passed to the dictionary data reading unit 11 and the process returns to step 203.

Next, the process of step 203 is performed again. In step 203, the dictionary data reading unit 11 reads the word sentence data 102 whose headword data 101 is “electronic computer” and adds 1 to the search hierarchy information. The word sentence data 102 read in FIG. 1 is the word sentence 108 (see FIG. 2). Next, in step 205, morphological analysis processing is performed, and the non-independent words “no”, “.” And the formal noun “ko” are excluded from “computer”, “no”, “ko”, “.” Divided into words. In step 206, the “computer” is subjected to the notation fluctuation correction process, and the “computer” as the processing result is additionally stored in the unsearched list 700 of the search information storage unit 14 together with the search hierarchy information (value “2”). FIG. 4 shows an example of the unexplored list 700 with “computer” added.

  Next, in step 207, an inquiry process is performed to determine whether or not the word data “different” stored at the head position of the unsearched data 701 in the unsearched list is included in the searched data. The basis determination process for the word data “different” is not included in the searched data, and the process proceeds to step 208. In step 208, the basis determination unit 15 determines whether or not the word data “different” is included in the basis list 18. Since “different” is not included in the base list 18, it is determined that the word data is not a base (N).

  Next, in step 210, search end determination processing is performed. The end condition determination unit 17 determines whether “different” search hierarchy information satisfies the end condition. The exploration hierarchy information is “1” and the termination condition is not satisfied. Therefore, the process proceeds to step 211.

  Next, in step 211, the dictionary data reading unit 11 checks whether or not “different” exists as the entry word 101 of the dictionary data 100. Since there is no “different” in the entry word data 101, the process proceeds to step 212. (N).

  Next, at step 212, the end condition determination unit 17 determines whether word data exists in the unsearched list 700 stored in the search information storage unit 14. Since word data exists in the unexplored list, the process proceeds to step 207 (N in 212). In this way, the above processing is recursively continued until there is no unsearched data 701 included in the unsearched list (until the contents of the unsearched list becomes empty). If there is no unsearched data 701 included in the unsearched list in step 207, the searched data is cleared and the process proceeds to step 213.

  In this way, basic determination processing for all headword data 101 included in the dictionary data 100 and basic determination processing for all word data obtained by dividing the word sentence data 102 stored in association with the headword data 101 are performed. As a result, the base database shown in FIG. 3 is generated. As shown in FIG. 3, the value “1” is added to each of the bases 402 (performed, equipment, color, taste... Xn) corresponding to the word 402 as shown in FIG. In other words, it is possible to express numerically what base 402 is associated with each word 401 (perform, interface, word 1, word 2... Word m). That is, since the relation with the base forming the semantic concept of each word can be expressed numerically, the semantic concept of the word can be easily processed by a mathematical method by using this base database.

  According to the embodiment described above, there is an effect that word semantic concept data can be automatically constructed from dictionary data.

  The present invention is not limited to the above-described embodiment, and the end determination condition can be variously modified without changing the gist of the invention.

  Another embodiment according to the present invention will be described with reference to the drawings. FIG. 8 is a functional block diagram showing the configuration of the base database generation apparatus according to the present invention. The base database generation device 1a is implemented by a computer equipped with a base database generation program 2a for executing the base database generation method according to the present invention. Each functional block shown in FIG. 8 is implemented when the base database generation program according to the present invention and the hardware of the computer cooperate to execute processing. The parts different from the base database generation apparatus in FIG. 1 described above are a search information storage unit 14a, a base determination unit 15a, and a non-basisable database storage unit 40. The other dictionary data storage unit 10, dictionary data reading unit 11, morphological analysis unit 12, notation fluctuation correction unit 13, end condition determination unit 17, base list 18, base registration unit 16, and base database storage unit 30 are the same as those in the first embodiment. It is a functional block similar to the base database generation device 1 used.

  The non-basisable database storage unit 40 includes headword data 101 registered in the dictionary data 100 shown in FIG. 2 and unsearched data 701 (see FIG. 4) obtained by dividing the word interpretation data 102. The base database generation method according to the present invention stores an unbasisable database that registers word data that cannot reach only one base. As long as the non-basisable database can be queried using the headword data 101 and the unsearched data 701, the creation method, recording method, and data structure are not limited.

  The search information storage unit 14a reads word data (hereinafter referred to as word data 101) associated with the word data divided from the word data 102 and the word data 102 from which the word data is generated (hereinafter referred to as word data 101). An unexplored list generated by associating “original word data” with the search hierarchy information is stored. A specific example of the unsearched list stored in the search information storage unit 14a is shown in FIG. In FIG. 13, the unexplored list 700a has a table structure that can store unsearched data 701, searched hierarchy information 702, and original word data 703 in association with each other. As shown in FIGS. 13A, 13B, and 13C, the unexplored list 700a is generated for each original word data 703 and stored. The unsearched list 700a is not limited to the format described above as long as it can handle a predetermined process by a program according to the present invention described later.

  In addition, searched data (not shown) is also stored in the search information storage unit 14a. The searched data is a file in which the headword data 101 and the unsearched data 701 that have been subjected to the base determination process in the base determination unit 15a are accumulated at any time. This searched data is used for avoiding duplicated base judgment processing for the same data. Therefore, the storage format and the like of the unsearched list and the searched data are not limited to the above as long as the program according to the present invention, which will be described later, can handle predetermined processing.

The base determination unit 15 a determines whether the entry word data 101 read by the dictionary data reading unit 11 corresponds to a base by an inquiry process using the base list 18.
The entry word data 101 determined to be included in the base list 18 based on the result of this processing is passed to the base registration unit 16. Further, the base determination unit 15 reads out the unsearched data 701 stored in the unsearched list 700a, deletes the unsearched data 701 from the unsearched list 700a, and then stores the unsearched data 701 in the search information storage. Processing for confirming whether or not the data is included in the searched data stored in the unit 14a is performed. If the unexplored data 701 is not included in the searched data as a result of the confirmation process, the unsearched data 701 is added to the searched data and the inquiry process is performed. As a result of the inquiry processing, if it is determined that the unexplored data 701 is included in the base list 18, the unsearched data 701 is transferred to the base registration unit 16. The unexplored data 701 is deleted from the unexplored list 700a. Further, a process of accessing the non-basisable database storage unit 40 and inquiring whether or not predetermined word data is stored in the non-basisable database is also performed. If the unsearched data 701 is included in the searched data in the confirmation process, a process of newly reading the unsearched data 701 from the unsearched list 700a is performed.

  FIG. 9 is a flowchart showing the processing flow of the base database generation program with respect to the processing flow of the base database generation method performed using the base database generation device 1a.

  Step 501 is a headword reading process similar to step 201 in the first embodiment. In step 501, the dictionary data reading unit 11 reads the entry word data 101 at the head position of the storage area of the dictionary data 100 stored in the dictionary data storage unit 10 and inserts “*” into the read flag 112. Reading of the headword data 101 is performed in the order stored in the dictionary data 100.

  Step 502 is a base database confirmation process. In step 502, the base determination unit 15 a performs a process for inquiring whether or not the headword data 101 read in step 501 is stored in the index of the base database 400. As a result of this process, if the headword data 101 does not exist in the index, the process proceeds to step 503, and if present, the process proceeds to step 521.

  Step 503 is a base determination process. Processing similar to that in step 202 in the first embodiment is performed. In step 503, the base determination unit 15a performs a process for inquiring whether or not the headword data 101 matches the word data included in the base list 18. If the headword data 101 is extracted from the base list 18 as a result of the inquiry process, the headword data 101 corresponds to the base, and the headword data 101 is transferred to the base registration unit 16.

  The above steps will be specifically described using the dictionary data 100 in FIG. 2 as an example. The headword 103 “Done” stored at the head position of the headword data 101 is read (step 501), and it is confirmed whether or not the headword 103 is stored in the base database. Since there is no word registered as an index in the base database 400 at the stage when the processing of this program is started, it is determined that the headword 103 (perform) does not exist in the index (N in Step 502), and in the subsequent Step 503, Since the headword 101 “do” is included in the base list 18 illustrated in FIG. 5, the base list 18 is determined to be a base, and the process proceeds to step 504.

  Step 504 is a base registration process, and the same process as in step 204 of the first embodiment is performed. Subsequent step 521 is all headword end determination processing, which is the same processing as step 213 in the first embodiment.

  In step 521, the end condition determination unit 17 performs an inquiry process as to whether or not there is a blank in the read flag 112 stored in association with the entry word data 101. In this embodiment, since the result of this process is blank, there is headword data 101 that has not been subjected to the base judgment process, and the base judgment process for all headword data has not been completed. Migrate to

  Next, in step 501, the headword data 101 “interface” (headword 105) stored next to the headword 103 is read, and in step 502, it is determined whether or not the headword 105 exists in the base database 400. . Since the entry word 105 does not exist in the base database 400, the process proceeds to step 503. In step 503, the base determination process for the headword 105 is performed. Since the headword 105 is not included in the base list 18, it is determined that it is not a base, and the process proceeds to step 505 (N in 503).

  Step 505 is a non-basisable determination process. In step 505, the basis determination unit 15 a performs an inquiry process using the entry word data 101 (entry word 105) with respect to the unbasisable database (not shown) stored in the non-basisability database storage unit 40. Since there is no word “interface” in the base database, the process proceeds to step 506 as a result of this processing.

  Step 506 is a sentence reading process. In step 506 and subsequent steps 507 and 508, processing similar to that in steps 203, 205, and 206 described above is performed. The unexplored data 701 generated in step 508 is stored as the unsearched list 700a in the search information storage unit 14a. As shown in FIG. 13A, the unsearched list 700a stores unsearched data 701, search hierarchy information 702, and original word data 703 of the unsearched data 701 in association with each other.

  Step 509 is a search completed determination process. In step 509, the base determination unit 15a reads the unexplored data 701 stored in the unsearched list 700a and the search hierarchy information 702 stored in association with the unsearched data 701, and uses the unsearched data 701 as the unsearched data 701. After deleting from the unexplored list 700, a process for confirming whether or not the unsearched data 701 is already included in the searched data (not shown) stored in the search information storage unit 14a is performed. As a result of this processing, if there is no matching word data in the searched data, the above-mentioned unsearched data 701 has not yet undergone the base determination processing, so this unsearched data 701 is added to the searched data. Then, the process proceeds to step 510.

  Step 510 is a base registration confirmation process. In step 510, the base determination unit 15 a performs an inquiry process as to whether or not the unsearched data 701 (electronic computer) stored in the unsearched list 700 is already stored in the word data 401 that is an index of the base database 400. Do. Since it is found that “electronic computer” does not exist in the base database 400, the process proceeds to step 512.

  Step 512 is a base determination process, and the same process as that of step 208 of the first embodiment described above is performed, and thus the description thereof is omitted. Since the above “electronic computer” is not included in the base list 18, the process proceeds to step 514.

  Step 514 is a non-basisability determination process, and the same process as step 505 described above is performed. In step 514, since the “electronic computer” does not exist in the non-basisable database, the process proceeds to step 515.

  Step 515 is search end determination processing, which is the same processing as step 210 in the first embodiment. In step 515, the exploration hierarchy information associated with the “electronic computer” is “1”, so the termination condition is not satisfied. Therefore, the process proceeds to step 516.

  Step 516 is a headword confirmation process, which is the same as step 211 in the first embodiment. In step 516, since the “electronic computer” exists in the entry word data 101, the process proceeds to 506 (N). In step 506, the sentence sentence data 102 corresponding to the “electronic computer” is read and 1 is added to the search hierarchy information.

  Steps 507 and 508 are performed using the word interpretation data 102 (word interpretation 108) read in step 506, and the unsearched list 700a shown in FIG. 13B is generated in the search information storage unit 14a.

  In step 509, the unexplored data 701 (computer) generated from the interpretation sentence 108 and stored in the unexplored list (FIG. 13B) and the exploration hierarchy information (related to the unexplored data 701) ( A value of “2”) is read out and a search completion determination process is performed. Since the word data “computer” is not included in the searched data as a result of the processing, the unsearched data 701 (computer) is added to the searched data, and the process proceeds to step 510. In the base registration confirmation process in step 510, it is determined that the “computer” does not exist in the base database 400, and the process proceeds to step 512.

  In the basic determination process in step 512, since the “computer” is not included in the basic list 18, this word data is added to the searched data, and the process proceeds to step 514.

  In step 514, since the “computer” does not exist in the non-basisable database, the process proceeds to step 515 (N). In step 515, since the value of the searched data is “2”, the end condition is not satisfied, and the process proceeds to step 516. In step 516, since it is confirmed that the “computer” exists in the entry word data 101, the process proceeds to step 506 (N). In step 506, the word sentence data 102 (the word sentence 110) corresponding to the headword 109 “computer” is read from the dictionary data 100, and the word data “electronic computer” is searched for the search hierarchy information (value is determined by the processing in step 507 and step 508). “3”) is stored in the unexplored list and the process proceeds to step 510. In step 509, since it is determined that the “electronic computer” exists in the searched data, the process proceeds to step 517.

  Step 517 is the same level determination process. In step 517, the base determination unit 15a stores the original word data (“computer”) of the word data (“electronic computer”) that is the current processing target in the unexplored list 700a stored in the search information storage unit 14a. Then, it is determined whether other word data having the same original word data is stored. Since there is no word data having “computer” as the caller word in the unexplored list at this stage, the process proceeds to step 518.

  In step 518, the base determination unit 15 a performs an inquiry process as to whether or not “computer” that is the original word data is already stored in the word 401 that is an index of the base database 400. As a result of this processing, the original word data “computer” does not exist in the index of the base database 400, so the processing moves to step 519. Step 519 is a non-basisable database registration process. In step 519, the basis determination unit 15a additionally stores the read source word data “computer” in the non-basisable database 40, and the process proceeds to step 520.

  Step 520 is an end determination process. In step 520, the end condition determination unit 17 performs an inquiry process as to whether or not any word data is stored in the unsearched list 700a stored in the search information storage unit 14a. In this embodiment, since word data is stored in the unexplored list, the process proceeds to step 517.

  In step 517, an inquiry process is performed to check whether word data having the same original word data as the word registered in the non-basisable database 40 in the non-basisability database registration process (step 519) exists in the unsearched list. In this embodiment, since there is no other word data having the word data “electronic computer” (FIG. 13C) as the caller word data in the unexplored list, the process proceeds to step 518.

  In step 518, collation processing is performed to determine whether or not the word data “electronic computer” is stored in the base database 400. Since the word data “electronic computer” is not stored in the base database 400, the process proceeds to step 519. In step 519, the basis determination unit 15a stores the word data “electronic computer” in the non-basisable database. Next, in step 520, end determination processing is performed.

  In step 520, since it is determined whether word data exists in the unexplored list, the process proceeds to step 517. In step 517, it is determined whether or not word data having the same caller word as the caller word registered in the non-basisable database registration process exists in the unsearched list. As a result of the process, it is determined that there is word data having the same read source word as the read source word data in the unsearched list, and the process proceeds to step 509.

  In step 509, the word data and the search hierarchy information at the storage start position of the unsearched list are read out, and a search completed determination process is performed. As a result of the processing, it is determined that the read word data “different” has not been searched. Next, in step 510, the base registration confirmation process is performed on the word data “different”. Since there is no “different” in the word 401 of the base database 400, the process proceeds to step 512. The basic judgment process of step 512 is performed for the “different”, and the word data “different” is additionally stored in the searched data. Since it is determined that “different” is not present in the base list 18 as a result of the determination process, the process proceeds to step 514.

  In step 514, a non-basisable determination process is performed on the word data “different”. As a result of the processing, it is determined that “different” is not stored in the non-basisable database, and the processing moves to step 515.

  Since the search hierarchy information of the word data “different” is 1, the process proceeds to step 516 by the search end determination process in step 515. In step 516, as a result of checking whether or not the word data “different” is stored in the headword data 101 of the dictionary data 100, it is determined that “different” does not exist in the headword data 101. Shifts to step 517.

  In step 517, collation processing is performed to determine whether another word data exists in the unsearched list using the read source word “interface” of the word data “different” as the read source word. As a result of this process, it is determined that the corresponding word data exists in the unexplored list, and the process returns to step 509. In step 509, the base determination unit 15 a reads the word data “device” stored at the storage start position of the unsearched list and performs a search completion confirmation process to determine whether it exists in the searched data. As a result of the processing, it is determined that “device” does not exist in the searched data, so this “device” is added to the searched data, and the processing moves to step 510.

  Next, in step 510, the base determination unit 15 a performs base determination processing on the word data “device”. Since it is determined that “device” is included in the base list as a result of the processing, the processing proceeds to step 513. In step 513, the base determination unit 15a adds “device” as the word 401 of the base database 400, adds “1” to the base 402 corresponding to “device”, and “0” for the base 402 not corresponding to “device”. "Is added.

  Next, in step 511, the base determination unit 15 a checks whether or not the read source word data “interface” of the word data “device” determined to be the base in the above processing is stored in the word 401 of the base database. If the result of the processing is not stored in the word 401 of the base database, that is, the index data, this read source word data “interface” is additionally stored as an index of the base database 400, and the additionally stored word 401 “interface” In the base 402, “1” is added to the part corresponding to the word data “device” determined to be the base, and “0” is added to the other part, and the process goes to step 517. Transition.

  As already described in step 517, it is determined whether or not there is word data in the unexplored list that uses “interface” that is the read source word of the word data “device” as the read source word data. As described above, this processing is recursively executed until the base determination processing is completed for the word data in the unsearched list.

  Next, a base list generation method according to the present invention will be described. FIG. 10 is a functional block diagram showing the configuration of the base list generation device used in the base list generation method. The base list generation apparatus is implemented by a computer equipped with a computer program for executing the base list generation method according to the present invention. Each functional block shown in FIG. 10 is realized by the cooperation between the base list generation program according to the present invention and the hardware of the computer executing the processing.

  In FIG. 10, the base list generation apparatus includes a base list generation program 3a, a dictionary data storage unit 10, a base list storage 20a, and a word frequency database 50. The dictionary data storage unit 10 is the same as the dictionary data storage unit used in the first and second embodiments, and stores the dictionary data 100 (see FIG. 2). The word frequency database is a database that accumulates information generated by processing of the base list generation program. The base list database 20a is a database that stores a base list generated by the implementation of the present invention.

  FIG. 12 shows an example of the structure of a word database stored in the word frequency database 50. 12, the word frequency database 800 is configured by a table structure having a field for recording a word 801 included in information read from the dictionary data 100 and a frequency 802 in which the word 801 is included in the dictionary data 100. However, the present invention is not limited to this, as long as the structure can record the frequency of appearance of each word, which will be described later.

  Next, an embodiment of the base list generation program will be described. FIG. 11 is a flowchart showing the flow of processing of the base list generation program. In FIG. 11, step 601 executed first is a headword reading process. In this process, the dictionary data reading unit 31 reads the entry word data 101 of the dictionary data 100 stored in the dictionary data storage unit 10. Step 601 is an interpretation sentence reading process. In this process, the word sentence data 102 stored in association with the headword data 101 read out above is read, the read word data 102 is passed to the morpheme analyzer 32, and 1 is set in the search hierarchy information. The addition processing proceeds to step 602. Here, the initial value of the exploration hierarchy information is zero.

  Step 603 is a morphological analysis process. In step 603, the morpheme analysis unit 32 performs a morpheme analysis process on the word interpretation data 102. The morpheme analysis process performed here is the same as the morpheme analysis process used in the first and second embodiments. Step 604 is a notation fluctuation correction process. In step 604, the morpheme analyzer 32 corrects the notation fluctuation for the word data generated from the word interpretation data 102. This notation fluctuation correction process is the same process as the notation fluctuation correction process used in the first and second embodiments. The word data generated by the process of step 604 is stored in the unexplored list stored in the search information storage unit 33 as word data as unsearched data together with the search hierarchy information stored in association with the unsearched data. The The structure of the unexplored list, the unexplored data, and the exploration hierarchy information are the same as those used in the first and second embodiments already described.

  Step 605 is a search completed determination process. In step 605, the word frequency registration unit 34 reads the unexplored data and the exploration hierarchy information from the unexplored list, performs an inquiry process as to whether or not the read unexplored data is included in the searched data, and stores the unexplored data. Delete from the unexplored list. As a result of this process, if the unexplored data is included in the searched data, the process proceeds to step 609. As a result of the collation process, if the read unexplored data is not included in the searched data, the process proceeds to step 606.

  Step 606 is a word frequency registration process. In step 606, the word frequency registration unit 34 performs a collation process for determining whether or not the word data that is the unexplored data is already recorded in the word frequency database 50. If it is not recorded as a result of the collation processing, the word data is added to the word 800 of the word frequency database 50, that is, an index, and a numerical value “1” is recorded in the frequency 802. If the unsearched data has already been recorded in the word frequency database 50 as a result of the matching process, 1 is added to the frequency 802 associated with the word 801 corresponding to the unsearched data and recorded. After the recording process to the frequency 802 is completed, the process proceeds to step 607.

  Step 607 is an end determination process. In step 607, the end condition determination unit 35 determines whether or not the exploration hierarchy information read from the unexplored list together with the unexplored data is equal to or greater than “4” as the end condition. If the result of this determination processing is that the exploration hierarchy information is less than 4, the processing moves to step 608. As a result of the determination process, if the exploration hierarchy information is 4 or more, the process proceeds to step 609.

  Step 608 is a headword confirmation process. In step 608, the dictionary data reading unit 31 reads unsearched data from the unsearched list stored in the search information storage unit 33, and performs an inquiry process as to whether data corresponding to the entry word data 101 exists. If the unexplored data exists in the entry word data 101 as a result of the inquiry process, the unexplored data is deleted from the unexplored list, 1 is added to the search hierarchy information, and the entry word data 101 is related. The stored sentence interpretation data 102 is read and passed to the morpheme analyzer 32.

  If the unsearched data does not exist in the entry word data 101 as a result of the inquiry process, the unsearched data is deleted from the unsearched list, and the process proceeds to step 609.

  Step 609 is an end determination process. In step 609, the termination condition determination unit 35 performs an inquiry process as to whether or not unsearched data exists in the unsearched list. If unexplored data exists as a result of this processing, the process proceeds to step 606. If unexplored data does not exist, the searched data is cleared and the value of the exploration hierarchy information is set to zero. Transition to 610.

  Step 610 is an all end determination process. In step 610, the end condition determination unit 35 performs an inquiry process as to whether or not there is a blank in the read flag 112 stored in association with the entry word data 101. As a result of this processing, if there is a blank, the entry word data 101 has not been processed, so the processing of this program is recursed to step 601. As a result of the above processing, if there is no blank in the read flag 112, the program is terminated.

  Step 611 is a base list extraction process. In step 611, the base list registration unit 36 performs a sort process according to the frequency stored in the word frequency database, extracts word data related to the word 801 located in the top 500 by the sort process, and extracts the extracted word data. Register in the base list database. The condition for extracting word data from the word frequency database may be extracted by a statistical method in addition to the extraction method by providing a constant threshold as described above.

  As described above, a base can be generated using the dictionary data 100.

  As described above, according to the present invention, it is possible to provide a word semantic concept extraction method that is the basis for constructing a highly accurate word thesaurus based on a word semantic concept when interpreting a natural language. .

It is a functional block diagram which shows embodiment of the base database production | generation apparatus which concerns on this invention. It is a figure which shows the example of the dictionary data used for the base database production | generation method concerning this invention. It is a figure which shows the example of the base database produced | generated by the base database production | generation method concerning this invention. It is a figure which shows the example of the unexplored list | wrist which can be used for this invention. It is a figure which shows the example of the base list | wrist which can be used for this invention. It is a flowchart which shows the example of a process of the computer program which implements the base database production | generation method concerning this invention. It is a figure which shows the example of the result by the morphological analysis process used for this invention. It is a functional block diagram which shows another Example of the basic database production | generation apparatus which concerns on this invention. It is a flowchart which shows the example of a process of another computer program which implements the base database production | generation method concerning this invention. It is a functional block diagram which shows embodiment of the base list production | generation apparatus which concerns on this invention. It is a flowchart which shows the process example of the computer program which implements the base list production | generation method concerning this invention. It is a figure which shows the structural example of the word frequency database produced | generated by the base list automatic creation method which concerns on this invention. It is a figure which shows another example of the unexplored list | wrist which can be used for this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Base database production | generation apparatus 100 Dictionary data 101 Headword data 102 Word interpretation data 200 Base list 400 Base database
401 word 402 base 600 analysis result data 700 word frequency database

Claims (9)

A computer comprising a first storage means for storing a word in association with one or more words explaining the meaning of the word, and a second storage means for storing the word in association with a base, a word base database A method of generating,
The computer
Reading the word from the first storage means;
A word reading step for reading the word sentence stored in association with the word read in the step from the first storage means;
A word dividing step for dividing the above sentence into words;
A base determination step for determining whether a word read from the first storage means or a word divided from the interpretation sentence corresponds to a base;
If it is determined in the basis determination step that the word corresponds to a basis, a base association step of associating the word with the base and storing it in the second storage means;
When it is determined that the word does not correspond to a base in the base determination step, the word reading step, the word division step, and the recursion step for performing the base determination step recursively using the word are included. Characteristic base database generation method. The recursion step includes a determination step for determining whether or not a basic determination process has been performed on all words divided from the interpretation sentence;
When the same word appears as a determination target in the basis determination step, a step of reading a basis already associated with the word from the second storage means;
2. The base database creation method according to claim 1, further comprising the step of associating the word used for reading the interpretation sentence with the read base and storing it in a second storage means. First storage means for storing a word and one or more word sentences explaining the meaning of the word in association with each other, second storage means for storing the word and the appearance frequency of the word in association with each other, and third storage means for storing a base list A base list for use in the base database of this word by a computer equipped with:
The computer
Reading a word from the first storage means;
A word reading step for reading a word sentence stored in association with the word read from the first storage means;
A word dividing step for dividing the read word sentence into words;
Registering the divided words in the second storage means;
Reading a sentence corresponding to the divided word from the first storage means;
Recursively performing word division processing, word registration processing, and word interpretation data reading processing on the read word sentence;
And a step of extracting only specific words using the word frequency information stored in the second storage means and storing them in the third storage means. First storage means for storing a word and one or more word sentences explaining the meaning of the word in association with each other; second storage means for storing the word and the base in association with each other;
Means for reading out the word from the first storage means;
A word reading means for reading a word sentence stored in association with the read word from the first storage means;
A word dividing means for dividing the read word sentence into words;
Base judgment means for judging whether a word read from the first storage means or a word divided from the interpretation sentence corresponds to a base;
A base associating means for associating the word with a base and storing it in the second storage means when the base judging means determines that the word corresponds to a base;
A recursion unit that recursively operates the word sentence reading unit, the word dividing unit, and the base determination unit using the word when the base determination unit determines that the word does not correspond to a base. A base database generation device characterized by that. A judging means for judging whether or not processing by the base judging means has been performed on all words divided from the interpretation sentence;
A reading means for reading a base associated with the word from the second storage means when it is determined that the word has already been processed by the determination means;
5. The base database generation apparatus according to claim 4, further comprising means for associating a word used for reading the interpretation sentence with the read base and storing it in a second storage means. First storage means for associating and storing a word and one or more interpretations explaining the meaning of the word;
Second storage means for storing the word and the appearance frequency of the word in association with each other;
Means for reading a word from the first storage means;
Means for reading an interpretation sentence stored in association with the word read from the first storage means;
Means for dividing the read-out sentence into words;
Means for registering the word in the second storage means;
If the word is already registered in the second storage means, means for adding 1 to the frequency information of the same word already registered;
And a third storage means for extracting and storing only specific words using the word frequency information stored in the second storage means. A computer comprising first storage means for storing a word and one or more word explanations explaining the meaning of the word in association with each other, and second storage means for storing the word in association with a base;
Means for reading out the word from the first storage means;
Means for reading out the memorized sentence stored in association with the read word from the first storage means;
A word dividing means for dividing the read word sentence into words;
Base judgment means for judging whether a word read from the first storage means or a word divided from the interpretation sentence corresponds to a base;
A base associating means for associating a word determined to be a base in the base determining means as a base corresponding to the read word and storing it in a second storage means;
When the basis judgment means determines that the divided word does not correspond to the basis, the word sentence corresponding to the word is read from the first storage means, and the word division means for the read word sentence and the basis judgment means for the word A computer program that operates as a recursive unit that recursively performs the above operation until a word divided from an interpretation sentence becomes a base. Computer
Determining means for determining whether or not the processing by the base determining means has been performed on the words divided by the dividing means;
A reading means for reading a base associated with the word from the second storage means when it is determined that the word has already been processed by the determining means;
8. The computer program according to claim 7, wherein the computer program is operated as means for storing in the second storage means in association with the word used for reading the word sentence that is the source of the divided word with respect to the read base. . Computer
A first storage means for associating and storing a word and one or more words explaining the meaning of the word;
Second storage means for storing the word and the appearance frequency of the word in association with each other;
Means for sorting according to the appearance frequency,
Third storage means for extracting and storing the rearranged words;
Means for reading a word from the first storage means;
Means for reading a sentence that is stored in association with the read word;
Means for dividing the read word sentence into words;
Means for registering the divided words in the second storage means;
Means for reading the sentence corresponding to the divided word from the first storage means;
A computer program that operates as a means for recursively performing the dividing process, the registering process, and the reading process of an interpretation sentence.

Images