JPH0950442A - Multilanguage document registration retrieval device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To suppress unregistered words to the minimum by registering extracted key word along with identifier as index, segmenting a word from inputted retrieval conditions, collating it with a key word of index and reading a document suited to the retrieval conditions. SOLUTION: A text data base part 3 stores the document including sentences of plural languages and a multilanguage key word extraction part 2 performs morpheme analysis corresponding to the sentence of the different language to the document and extracts the key word of the document. An index registration part 4 registers the extracted key word along with the identifier of the corresponding document as the index. Also, a retrieval condition input part 11 inputs the retrieval conditions, the word is segmented from the retrieval conditions to the index collation part, it is collated with the key word of the index and a text extraction part 14 reads the document by a collated result. Thus, the size of the index prepared at the time of registration is made compact as well.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a multilingual document registration / retrieval apparatus for retrieving a document by registering an index for retrieval with respect to a document including sentences described in a plurality of languages. Specifically, from the text database of documents written in multiple types of languages, keywords are extracted corresponding to multiple languages and registered as an index, and the registered index is used to search a multilingual text database. The present invention relates to a multilingual document registration / retrieval device.

[0002]

2. Description of the Related Art Conventionally, as a document retrieval apparatus for a multilingual document including sentences described in a plurality of types of languages, for example, a "document retrieval apparatus" described in Japanese Patent Laid-Open No. 21180/1992 is known. ing. This "document retrieval device"
A database with keywords created in your own language,
In a document search device that inputs the keyword and searches the database, has a dictionary of a native language and another language,
For a keyword input in another language, the keyword is converted into the native language, and the database is searched with the keyword in the native language. Then, the document name of the retrieved document and the whole sentence of the selected document are converted into another language and displayed on the screen.

[0003]

By the way, when performing a document search using an index in the document search process, the document to be searched is not limited to one language (for example, Japanese) such as a native language, and In many cases, the parts cited as cited documents also include languages of other countries (for example, English). Therefore, the index in the document retrieval device is compatible with multiple languages (multilingual). Otherwise, you will not be able to retrieve enough documents.

Conventionally, in a document search apparatus which uses an index for a document search process, a method of creating an index used for document search is to perform morphological analysis to cut out a word and create an index using the word as a keyword.
In that case, since the language for morphological analysis is usually one language (for example, only Japanese), words in languages of other countries could only be cut out as words of proper nouns. Therefore, in order to register the extracted words as a keyword in the index, it is not possible to restore the standard notation or the original form. I could only extract it and register it in the index.

Therefore, when the target document contains many languages of different countries, the number of keywords (proper nouns) of the words registered as the search index increases, and the index size becomes unnecessarily large. There was a problem.

Further, in such a document retrieval apparatus,
When performing a document search, there is a problem that only a character string registered as a keyword, for example, a character string having the same pattern as the index of a proper noun can be searched, and a search in an appropriate range cannot be sufficiently performed. In other words, when searching for documents,
There is a problem that a desired search document is not hit even if a search is performed unless a character string having exactly the same pattern as the word (proper noun) cut out by the morphological analysis is hit.

When a keyword in a certain language is registered and a search request is made in a language other than that of the country, such as the "document retrieval device" described in Japanese Patent Laid-Open No. 21180/1992 mentioned above. Can be configured to search for documents that include languages of other countries, for example, by translating it into a word in a language that has already been registered and has the same meaning. There is a problem in that the index is composed of keywords only in its own language, and consideration is not given to registering the index for documents composed of multiple languages.

The present invention has been made to solve these problems, and a first object of the present invention is to make documents containing sentences written in a plurality of languages different from each other as much as possible. Morphological analysis is performed for each language, words are extracted, keywords are extracted, and keywords can be registered so that unregistered words are minimized even for documents containing sentences written in multiple languages It is to provide a multilingual document registration / retrieval device that can be held down to the limit.

A second object of the present invention is to extract a word from a document including sentences written in a plurality of languages by morphological analysis corresponding to a plurality of languages, extract the keyword, and register the keyword. To provide a multilingual document registration / retrieval apparatus that avoids duplication of the range to be analyzed, minimizes the index size, and improves the retrieval accuracy when performing retrieval.

The third object of the present invention is to deal with a document including sentences written in a plurality of languages even when the inflection of a word in a language other than the native language and fluctuation of the notation are dealt with, and a search is performed. It is an object of the present invention to provide a multilingual document registration / retrieval device capable of improving the retrieval accuracy when performing the retrieval, minimizing the extraction of unnecessary unregistered words, and minimizing the index size.

Further, a fourth object of the present invention is that, when a word including sentences written in a plurality of languages is cut out by morphological analysis corresponding to a plurality of languages, the words corresponding to the respective plurality of languages are handled. Avoid duplicate analysis of target text by combining morphological analysis, perform morphological analysis as efficiently and optimally as possible, and minimize unregistered words for documents containing sentences written in multiple languages. To provide a multilingual document registration / retrieval device.

[0012]

In order to achieve the above-mentioned object, a multilingual document registration / retrieval apparatus, which is the first feature of the present invention, is used for retrieval of a document including sentences in a plurality of languages. In a multilingual document registration / retrieval device that creates and registers an index and retrieves a document by the index, a multilingual document storage unit (3) that stores a document including sentences in a plurality of languages is different from the document. A keyword extraction unit (2) for performing morphological analysis on each sentence of the language to extract a keyword of the document, and an index registration for registering the keyword extracted by the keyword extraction unit together with the identifier of the corresponding document as an index. Means (4), search condition input means (11) for inputting search conditions, and a word from the search conditions input by the search condition input means Cut, an index matching means for matching the keywords of the cut-out words and the index (12), characterized in that it comprises a reading means (14) for reading the documents relevant to the search criteria by matching results of the keyword and words.

A multilingual document registration / retrieval apparatus, which is a second feature of the present invention, creates an index in the multilingual document registration / retrieval apparatus that creates an index to be used when retrieving a document containing sentences in a plurality of languages. The multilingual document registration apparatus includes a plurality of word cutout means (21a, 22a, 23a) having different cutout target languages, a setting means (27) for setting the processing priority of the plurality of word cutout means,
Keyword extraction control means (26) for controlling the plurality of word cutting means according to the processing priority, cutting the words from the document and extracting the keywords, and the extracted keywords and the identifiers of the documents from which the words of the keywords are cut out. And an index registration means (28) for registering in the index in association with each other.

Further, in the multilingual document registration / retrieval device which is the third feature of the present invention, the keyword extraction control means processes the word which cannot be identified by the word cutting means of a certain processing priority, to the next processing. For the words that have been processed and processed by the priority word cutting means, the identifier of the word is used as a keyword, and for the words that cannot be identified to the end by a plurality of word cutting means, the word is used as a keyword. Characterize.

A fourth feature of the present invention, a multilingual document registration / retrieval apparatus, creates and registers an index used for retrieval for a document including sentences in a plurality of languages, and retrieves the document by the index. In the multilingual document registration / retrieval device that performs the above, the document registration device is provided with an input unit (1) for inputting a document to be registered and instructing keyword extraction, and a dictionary used for extracting words to extract a keyword of the document by morphological analysis. A keyword extracting means (2), a registering means (4) for registering the keyword extracted by the keyword extracting means together with an identifier of a corresponding document in an index, a document to be registered, an index and not registered in the dictionary file. Holding means (3, 5) for holding a word.

In the multilingual document registration / retrieval device, which is the fifth feature of the present invention, the keyword extracting means performs morphological analysis on a sentence of each language from a document composed of sentences of a plurality of languages. A plurality of word cutout means (21a, 22a, 23a) for cutting out words by means of a plurality of word cutout means, and a plurality of dictionary files (21b, 2) for storing dictionaries corresponding to the languages referred to by the plurality of word cutout means, respectively.
2b, 23b), an order setting means (27) for setting the order in which the plurality of word cutout means are applied, and a plurality of word cutout means are controlled from the document in the order set by the order setting means. And a control means (26) for controlling to cut out a word in a multilingual sentence.

In the multilingual document registration / retrieval device, which is the sixth feature of the present invention, an unregistered keyword that is temporarily held as an unregistered keyword candidate for a word judged by the word cutting means as an unregistered word. The candidate holding means (25) and the keyword candidate holding means (24) for temporarily holding as a keyword candidate the words extracted from other dictionaries, the control means (26, 27) By controlling the word cutting means in the first stage and inputting a document including sentences in a plurality of languages, the words are cut out by morphological analysis, and the words judged as unregistered words are temporarily unregistered keywords. The keyword stored in the unregistered keyword candidate storage means as a candidate, and the keyword candidate stored as a keyword candidate for the word extracted from the dictionary. Performing the process of holding in stages, sequentially controlling each word cutout means, inputting the unregistered word candidates held in the unregistered keyword candidate holding means by the word cutout means of the previous stage, and inputting the word by morphological analysis. The words that have been cut out and determined to be unregistered words are left as they are in the unregistered keyword candidate holding means, and the words extracted from the dictionary are deleted from the unregistered keyword candidate holding means, and the keyword candidate holding means. Is additionally held, and finally the keyword candidate held in the keyword candidate holding means is used as a keyword, and the unregistered keyword held in the unregistered keyword candidate holding means is used as an unregistered keyword. It is characterized by being registered in the index together with.

The multilingual document registration / retrieval device, which is the seventh characteristic of the present invention, creates and registers an index to be used for retrieval for a document including sentences in a plurality of languages, and retrieves the document by the index. In the multilingual document registration / retrieval device, the document retrieval device includes a retrieval condition input means (11) for inputting retrieval conditions, and a word extraction from the retrieval conditions input by the retrieval condition input means for matching with an index. Index matching means for language (12)
And extraction means (14) for extracting the corresponding document from the text database according to the matching result of the index matching means.

Further, in the multilingual document registration / retrieval device according to the eighth aspect of the present invention, the index collating means performs a morphological analysis on a sentence of a corresponding language from a document composed of a plurality of languages. A plurality of word cutout means (131a, 132a, 133a) for going and cutting out words
And an order setting means (137) for setting the order in which the word cutout means is applied by combining one or more word cutout means, and the search conditions input by the search condition input means in the order set by the order setting means. It has a control means (136) for controlling to cut out a word.

Further, in the multilingual document registration / retrieval device according to the ninth feature of the present invention, a word judged as an unregistered word by the word cutting means is temporarily held as an unregistered search word candidate. Search term candidate holding means (1
34) and a search word candidate holding means (135) for temporarily holding as a search word candidate for words extracted from other dictionaries, the control means (137, 1).
37) controls the word cutting means in the first stage, inputs a document including sentences in a plurality of languages, cuts out words by morphological analysis, and temporarily cuts words judged to be unregistered words. Is held in the unregistered search word candidate holding means as an unregistered search word candidate, and the word extracted from the dictionary is held as a search word candidate in the search word candidate holding means, and each word is sequentially The unregistered word candidate held in the unregistered search word candidate holding unit by the word cutting unit in the preceding stage is input as an input by controlling the cutting unit.
Words are cut out by morphological analysis, and words that are determined to be unregistered words are left as they are in the unregistered search word candidate holding means, and words extracted from the dictionary are deleted from the unregistered search word candidate holding means. Then, a process of additionally holding the search word candidate holding means is performed, and finally the search word candidate held in the search word candidate holding means is set as a search word, and the search word candidate holding means holds the search word candidate held in the unregistered search word candidate holding means. It is characterized in that the registered search word candidate is used as an unregistered search word, index matching is performed, the corresponding document is extracted by the text database unit, and the result information is output.

According to the multilingual document registration / retrieval apparatus of the present invention having such various characteristics, in the multilingual document registration / retrieval apparatus having the first characteristic, the multilingual document storage means (3) is provided. , Documents containing sentences in a plurality of languages are stored, and the keyword extracting means (2) performs morphological analysis on the documents corresponding to sentences in different languages to extract the keywords of the documents. The index registration means (4) registers the keyword extracted by the keyword extraction means as an index together with the identifier of the corresponding document.

When searching for a document, when the search condition is input by the search condition input means (11), the index collating means (12) cuts out a word from the search condition input by the search condition input means and cuts it out. Matches the word that was created with the keyword in the index. Then, the reading means (14) reads out a document that matches the search condition based on the matching result between the keyword and the word. In this manner, an index used for searching is created and registered for a document including sentences in a plurality of languages, and the document is searched by using the index.

Further, in the multilingual document registration / retrieval apparatus which is the second feature of the present invention, a plurality of word clipping means (21a, 22a, 23a) having different clipping target languages are used as the multilingual document registration apparatus for creating an index. ) Is provided, and the setting means (27) sets the processing priority of the plurality of word cutout means, the keyword extraction control means (26) controls the plurality of word cutout means in accordance with the processing priority. , Cut out words from the document,
Extract keywords. Index registration means (28)
Registers the extracted keyword in the index in association with the identifier of the document in which the word of the keyword is cut out. As a result, with respect to a document including sentences in a plurality of languages, a plurality of word segmentation units corresponding to respective target languages can perform morphological analysis to segment a keyword word. Therefore, it is possible to minimize the number of unregistered words in a document written in multiple languages and create an index to be used for searching.

Further, in the multilingual document registration / retrieval device, which is the third feature of the present invention, for a document including sentences in a plurality of languages, a plurality of word segmentation means are provided corresponding to respective target languages. In the case where a keyword word is cut out by performing morphological analysis, the keyword extraction control means processes a word that cannot be identified by the word cutout means having a certain processing priority, and processes the word cutout means having the next processing priority, For the cut out word, the identifier of the word is used as a keyword. Further, with respect to a word which cannot be identified to the end by a plurality of word cutting means, the word is used as a keyword. As a result, words can be cut out from a document written in a plurality of languages by morphological analysis corresponding to each language, and keywords can be extracted while avoiding overlapping analysis ranges. In this way, the index size when registering keywords can be minimized.

According to the multilingual document registration / retrieval device of the fourth aspect of the present invention, the document registration device includes an input unit (1), a keyword extraction unit (2), and a registration unit (4). , Holding means (3, 5).
When the input unit (1) inputs the document to be registered and instructs the keyword extraction, the keyword extraction unit (2) is provided with a dictionary used for extracting words, extracts the keyword of the document by morphological analysis, and the registration unit (4 ) Registers the keyword extracted by the keyword extracting means in the index together with the identifier of the corresponding document. As a result,
The holding means (3, 5) holds the document to be registered, the index, and the word not registered in the dictionary file. As a result, it is possible to create and register an index used for a search for a document including sentences in a plurality of languages, and search the document by using the index.

According to the multilingual document registration / retrieval device of the fifth aspect of the present invention, in the keyword extracting means,
A plurality of word cutting means (21a, 22a, 23a)
When a word is cut out from a document composed of sentences of a plurality of languages for a sentence of each language by morphological analysis, a plurality of dictionary files (21b, 22b, 23b) are used by the plurality of word cutting means (21a, 22a). , 23a) stores a dictionary corresponding to the language to be referred to, so that when the order setting means (27) sets the order in which the plurality of word cutting means is applied, the control means (2
6) controls a plurality of word cutout means in the order set by the order setting means to cut out words of a corresponding multilingual sentence from the document.

Further, in the multilingual document registration / retrieval device, which is the sixth feature of the present invention, the unregistered keyword candidate holding means (25) temporarily stores the word judged as the unregistered word by the word cutting means. Are stored as unregistered keyword candidates, and words extracted from other dictionaries are temporarily stored as keyword candidates by the keyword candidate storage means (24). In the case of extracting a keyword corresponding to multiple languages, the control means (2
6, 27) controls the word cutting means in the first stage to input a document including sentences in a plurality of languages and cut words by morphological analysis. As a result, the word judged as an unregistered word is temporarily held in the unregistered keyword candidate holding means as an unregistered keyword candidate, and the word extracted from the dictionary is held as the keyword candidate in the keyword candidate holding means. The process to hold is performed.

Then, each word cut-out means is sequentially controlled, and the word cut-out means in the preceding stage inputs the unregistered word candidates held in the unregistered keyword candidate holding means and cuts out the words by morphological analysis. The word determined to be an unregistered word is left as it is in the unregistered keyword candidate holding means, and the word extracted from the dictionary is deleted from the unregistered keyword candidate holding means to be stored in the keyword candidate holding means. Perform the process of additionally holding.

Finally, the keyword candidates held in the keyword candidate holding means are used as keywords, and the unregistered keywords held in the unregistered keyword candidate holding means are used as unregistered keywords and are indexed together with the identifiers of the corresponding documents. to register. In this way, words are cut out sequentially by morpheme analysis, so avoiding redundant analysis of the target text by combining multiple morpheme analyzes corresponding to multiple languages, and optimize morpheme as efficiently as possible. Analysis can be performed, and unregistered words can be minimized for a document including a sentence written in multiple languages.

According to the multilingual document registration / retrieval device, which is the seventh feature of the present invention, the document retrieval device includes a retrieval condition input means (11), an index collation means (12) corresponding to multiple languages. When the search condition inputting means (11) inputs a search condition, the multi-lingual index collating means (12) is provided with an extracting means (14) and the search condition inputting means (12) extracts the search condition from the search condition inputting means. Cut out words and match with index. The extraction means (14) extracts the corresponding document from the text database according to the matching result of the index matching means. As a result, when an index used for a search is created and registered for a document including sentences in a plurality of languages, the index can be used to search the document.

In the case of performing a document search, the eighth aspect of the present invention
According to the multilingual document registration / retrieval device characterized by the above, in the index collating means, the plurality of word cutting means (131a, 132a, 133a) are arranged in a corresponding language from a document composed of sentences in a plurality of languages. Morphological analysis is performed on the sentence to cut out words, so the order setting means (137) sets the order in which the word cutting means is applied by combining one or more word cutting means, and the control means (136). Control is performed to cut out words from the search condition input by the search condition input means in the order set by the order setting means, and the document is searched by the word of the search condition.

Further, in the multilingual document registration / retrieval device according to the ninth feature of the present invention, the unregistered search word candidate holding means (134) temporarily holds the word judged as the unregistered word by the word cutting means. The search word candidate holding means (135) temporarily holds the words extracted from other dictionaries as the search word candidates.

The control means (136, 137) controls the word cutting means in the first stage to input a document containing sentences in a plurality of languages, cut out words by morphological analysis, and identify unregistered words. The determined word is temporarily held in the unregistered search word candidate holding means as an unregistered search word candidate, and the word extracted from the dictionary is held in the search word candidate holding means as a search word candidate. Perform processing.

Then, each word cutout means is controlled in sequence, and the word registration means stored in the unregistered search word candidate holding means by the preceding word cutout means is used as an input to cut out words by morphological analysis. The word determined to be an unregistered word is left as it is in the unregistered search word candidate holding means, and the word extracted from the dictionary is deleted from the unregistered search word candidate holding means. A process of additionally holding the candidate holding means is performed.

Finally, the search word candidate held in the search word candidate holding means is used as a search word, and the unregistered search word held in the unregistered search word candidate holding means is used as an unregistered search word. Collation is performed, and the corresponding document is extracted by the text database unit and the result information is output.

In this way, an index used for searching is created and registered for a document including sentences in a plurality of languages, and the document is searched using the index. Therefore, it is possible to cope with the inflection of words other than the native language and the fluctuation of the notation, and improve the search accuracy. Further, unnecessary extraction of unregistered words is minimized, and the index size can be minimized. Furthermore, for documents written in multiple languages, avoiding redundant analysis of the target text by combining multiple morphological analyzes for multiple languages, and perform morphological analysis as efficiently and optimally as possible. It is possible to minimize unregistered words in a document described in multiple languages.

[0037]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be specifically described below with reference to the drawings. FIG. 1 is a block diagram showing the configuration of a multilingual document registration / retrieval apparatus according to an embodiment of the present invention. In FIG. 1, 1 is an input processing unit,
2 is a multilingual keyword extraction unit, 3 is a text database unit, 4 is an index registration unit, 5 is an index file unit, 11 is a search condition input unit, 12 is a multilingual index registration unit, 13 is a display unit, and 14 is a text extraction unit. It is a department.

In the text database section 3, for example,
Documents containing sentences written in multiple languages, such as sentences written in English and sentences written in Japanese (110: FIG. 1
1) is stored. When the input processing unit 1 receives a document registration input instruction from the user, the multilingual keyword extraction unit 2 responds to the document instructed by the user.
The process of extracting the keyword is performed. Here, the document for extracting the keyword is, for example, a document input from the input processing unit 1 and registered in the text database unit 3, or a document already registered in the text database unit 3. A document from which a keyword is extracted is specified for the document, and the keyword is extracted from the document.

As will be described later, the multilingual keyword extraction unit 2 is provided with a plurality of morphological analysis units (21a to 23a: FIG. 2) for performing a morphological analysis corresponding to the language of each country. The morphological analysis unit corresponding to the language of each country is sequentially controlled, and the morphological analysis is efficiently performed for each sentence of different languages in the document, and the process of extracting the word and extracting the keyword is performed. Multiple morphological analyzers,
A dictionary file is provided for each language corresponding to the language to be analyzed, and when morphological analysis is performed to extract words, the words corresponding to each language are the corresponding dictionary corresponding to the language of each country. Be compared.

The words registered in the dictionary file of a certain language are temporarily stored as keyword candidates in the keyword candidate storage unit, and the words not registered in the dictionary of a certain language are unregistered keyword candidates. It is temporarily held in the unregistered keyword candidate holding unit. Then, the unregistered keyword candidate word is subjected to morphological analysis by the morphological analysis unit of the next language. In this way, when the word is compared with the dictionary corresponding to the language of the next country, the morphological analysis is performed including the word of the previously unregistered keyword candidate, the word is cut out, and the keyword is extracted.

The keywords extracted in this way are
The index registration unit 4 registers it in the index table of the index file unit 5 as an index together with the identifier of the document corresponding to the keyword.

When a user searches for a desired document,
The user inputs the search condition from the search condition input unit 11.
When the search conditions are entered, the multilingual index matching unit 1
In step 2, a word is cut out from the input search condition, and the cut out word is collated with the keyword of the index of the index table. Then, according to the matching result, the text extracting unit 14 reads out a document that matches the search condition based on the matching result of the keyword and the word, and the display unit 13
At, the read document is displayed.

FIG. 2 is a block diagram showing the structure of the main part of the multilingual keyword extracting section. FIG. 2 shows the block of each element in the multilingual keyword extraction unit and the data flow thereof. In FIG. 2, 1 is an input processing unit, 21a is a first morphological analysis unit, 21b is a first dictionary file unit, 22a is a second morphological analysis unit, 22b is a second dictionary file unit, Reference numeral 23a is the Nth morpheme analysis unit. 24 is a keyword candidate holding unit, 25 is an unregistered keyword candidate holding unit, 26 is a keyword / unregistered keyword determination unit, 27 is an order setting unit,
Reference numeral 28 is an index registration unit, and 29 is an index file unit.

As shown in FIG. 2, the multilingual keyword extraction unit includes a plurality of morphological analysis units (21a) corresponding to respective languages for morphological analysis of sentences in each country.
23a) and a plurality of dictionary file units (21) corresponding to each language that supply dictionary data of each language to each morphological analysis unit.
b to 23b) are provided, and the morphological analysis of the multilingual document is performed in order to efficiently perform the morphological analysis of the multilingual sentence by controlling the plurality of morphological analysis units (21a to 23a). An order setting unit 27 that sets the order in which the above is performed, and a keyword candidate holding unit 24 that temporarily stores the analyzed words as keyword candidates as its working memory.
For a word that has not been analyzed by the morpheme analysis unit corresponding to one language, an unregistered keyword candidate that is temporarily registered is stored in order to perform morpheme analysis by the morpheme analysis unit corresponding to another language. A section 25 is provided. Then, when the morphological analysis is completed, the keyword / unregistered keyword determination unit 26 determines a word to be a keyword to be registered and a word to be an unregistered keyword, and the index registration unit 28 extracts the keyword from the document. It is registered in the index file unit 29 in association with the identifier.

FIG. 3 is a diagram for explaining the contents of the control table used when the keyword extraction process is performed. 3A shows a keyword extraction management table 30, FIG. 3B shows a morphological analysis management table 36, and FIG. 3C shows an analysis target character string type setting table 37. Shows.

The keyword extraction management table 30 is shown in FIG.
As shown in (a), it is a table for managing the usage state of each morphological analysis unit provided in the multilingual keyword extraction unit, and is a number field 31, a corresponding language type field 32, an order field 33, a usage flag field. 34 and an analysis target character string type field 35. The condition data of the morphological analysis unit used for each field is set. For example, in the second entry from the top, the condition of the morphological analysis unit for performing morphological analysis in Japanese is set, “2” is set in the number field 31, and “2” is set in the corresponding language type field 32. "Japanese" is set, "1" is set in the order field 33, and "O" is set in the use flag field 34.
N "is set. Further," text-ALL "is set in the analysis target character string type field 35,
The condition data here is set. In other words, from this condition data, "the second morphological analysis unit is used for Japanese, first performs morphological analysis of a multilingual document, and performs the analysis target character string range as the entire text. It means that.

The morphological analysis unit used in the keyword extraction process is managed by the morphological analysis management table 36. In the morphological analysis management table 36, FIG.
As shown in (b), data on the number n of morphological analysis units that can be used for each national language and the number i of the morphological analysis unit currently in use are managed. Further, in order to set the range of the analysis target character string, as shown in FIG. 3C, an analysis target character string type setting table 37 is provided.
In the analysis target character string type setting table 37, the range of the analysis target character string is defined according to the analysis target character string type. For example, when the analysis target character string type is “unregistered word group”, the range of the analysis target character string is set to “all keyword candidates in the unregistered keyword candidate holding unit or the unregistered search word holding unit”. When the analysis target character string type is “text-ALL”, the range of the analysis target character string is set to be “all text of registered document or search expression”.

Next, the multilingual document registration processing in the keyword extraction processing section as shown in FIG. 2 will be described using these control tables. FIG. 4 is a flowchart showing the overall processing flow of the multilingual document registration processing. The flowchart shown in FIG. 4 is an overall flow of processing for extracting a word as a keyword by performing morphological analysis corresponding to each language on a document described in one or more languages and registering it in an index. Is shown. FIG. 5 is a flowchart showing the processing flow of the keyword extraction management table condition setting processing in the multilingual document registration processing, and FIG. 6 shows the analysis target character string range in the multilingual document registration processing. FIG. 7 is a flowchart showing a processing flow of setting processing, and FIG. 7 is a flowchart showing a processing flow of keyword extraction processing in the multilingual document registration processing. Further, FIG. 8 is a flowchart showing a processing flow of unregistered keyword candidate processing in the multilingual document registration processing.

First, referring to the flow chart of FIG.
The overall processing of the multilingual document registration processing will be described. When the processing is started, first, in step 41, the input processing unit 1
The instruction to register the document is given by. Next, in step 42, a keyword extraction condition setting process (FIG. 5) is performed, and subsequently, in the next step 43, a character string set to be analyzed by morphological analysis is set (FIG. 6). That is, the condition of the morpheme analysis unit used in the keyword extraction process is set, and then the set of character strings to be analyzed is set using the morpheme analysis unit for which the condition is set. In the character string set setting process, for example, all of the text of the registered document or a set of unregistered keyword candidate character strings is set, and the character string set to be analyzed is specifically set. To do.

After the processing of these settings is completed, next, in order to control the keyword extraction processing, in step 44, the variable i indicating the number (order) of the morphological analysis unit currently used is set to "1". Then, the morphological analyzer used at the beginning of the document registration process is set. Next, in step 45, it is determined whether or not the use flag of the morphological analysis unit of the language corresponding to the variable i is on. If the usage flag is not on, the morphological analysis unit with the number of the variable i does not perform the morphological analysis process, so the process proceeds to step 52 to proceed to the process of the morphological analysis unit with the next number.

If the use flag of the morphological analysis unit of the language corresponding to the variable i is ON in the determination at step 45, that is, the keyword extraction management table 30
The language corresponding to the variable i indicating the order of
If the "use flag" of the morphological analysis entry of the language (where i is i) is ON, the process proceeds to the next step 46, and the i-th corresponding morphological analysis unit sequentially processes the character string set. The keyword extraction process (FIG. 7) is performed. In this keyword extraction process, as will be described later, the condition data of the keyword extraction management table 30 causes the first set of character string sets set in the type of the analysis target character string of the entry of the morphological analysis unit in the i-th order The i-th corresponding morpheme analysis unit extracts a keyword.

Next, in step 47, it is judged whether or not the extracted keyword is registered in the (i-th) morphological analysis dictionary. If it is not registered, the process proceeds to step 48, and the unregistered keyword is registered. It is stored in the unregistered keyword candidate holding unit together with the corresponding document ID as a candidate, and the process proceeds to step 51. If it is determined in step 47 that the extracted keyword is registered in the morphological analysis dictionary, the process proceeds to step 49,
It is stored in the keyword candidate holding unit together with the corresponding document ID as a keyword candidate. And then, step 5
At 0, unregistered keyword candidate processing (FIG. 8) is performed. In this unregistered keyword candidate processing, as will be described later, character string matching is performed on the word that was determined as an unregistered keyword candidate in the processing of the previous morpheme analysis unit, but was made a keyword candidate in the processing of the subsequent morpheme analysis unit. The matching word is removed from the unregistered keywords.
When this unregistered keyword candidate process is completed, the process proceeds to step 51.

In step 51, the i-th item of the registration document
It is determined whether or not the keyword extraction by the th morpheme analysis unit is completed. If keyword extraction has not been completed, the process returns to step 46 and the processing from step 46 is repeated. Further, if it is possible to confirm the end of the keyword extraction process by the morpheme analysis unit corresponding to the i-th by the determination process of this step 51, the keyword extraction process by the next morpheme analysis unit is performed, and therefore it is used in the next step 52. The variable i indicating the order of the morphological analysis unit is incremented, that is, the variable i is counted up (i = i + 1), and in the next step 53, the number n of the available morphological analysis units corresponding to each national language and the next It is compared with a variable i indicating the order of the morphological analysis unit used for.

As a result of this comparison, if n ≧ i, the keyword extraction processing by the i-th morpheme analysis unit has not been completed, so the processing returns to step 45 and the processing from step 45 is repeated. If n <i, it means that the analysis by the morpheme analysis unit which is in the usable state according to the conditions set in the keyword extraction management table has all been completed, so the process proceeds to step 54, and the keyword candidate group The process of creating an index from the unregistered keyword group is performed. This completes the registration process of one document, and then, in step 55, it is determined whether or not the registration of the document is ended. The process is returned to and the process from step 41 is repeated. Further, when the registration of the document is ended, the series of processes here is ended.

Next, the setting process of the conditions of the keyword extraction management table in the multilingual document registration process will be described. This process is a process executed in step 42 of the overall process flow of the multilingual document registration process of FIG. Please refer to the flowchart of FIG. When the processing here is started, the process proceeds to step 61, the number of morphological analysis units corresponding to each language stored in the keyword extraction management table is counted, and a variable n indicating the number of morphological analysis units is counted.
Set to. That is, the keyword extraction management table 30
If the number of morphological analysis units registered in is m, then n = m is set. Next, in step 62, the analysis order is analyzed by using the morpheme analysis unit corresponding to each language, in the order field 3 of the keyword extraction management table 30.
Set to 3. Then, in the next step 63, in order to set the morphological analysis corresponding to each language used when registering this document, it is determined whether the morphological analysis unit corresponding to each language is used or not, using flag of the keyword extraction management table 30. It is set by the ON / OFF flag in the field 34. As a result, the condition setting process of the keyword extraction management table ends.

Next, the setting process of the analysis target character string range in the multilingual document registration process in the multilingual document registration process will be described. As described above, this process is a process executed in step 43 of the overall process flow of the multilingual document registration process of FIG. Please refer to the flowchart of FIG. When this process is started, first, at step 71, a variable j stored in the keyword extraction management table 30 and indicating the entry number of the morphological analysis unit corresponding to each language is set to "1". Next, in step 72, the type of the analysis target character string for the morphological analysis unit corresponding to each language of the jth entry of the keyword extraction management table is selected from the analysis target character string type setting table. As described above, in the analysis target character string type setting table 37, the range of the character string to be subjected to the morphological analysis is set according to the type of the analysis target character string type. The type of analysis target character string is selected from the types of analysis target character string types in the column type setting table 37, and is set in the analysis target character string type field 35 of the keyword extraction management table 30.

Next, in step 73, it is determined whether or not the "order of analysis" of the j-th morpheme analysis unit is the first. If the order of the j-th morpheme analysis unit is the first, the process proceeds to step 74, and the previously set analysis target character string type is ignored and the "text-AL" is forced.
L ". Then, the process proceeds to step 75. As a result, the range of the solution target character string analyzed by the first morphological analysis unit is always the entire text of the registered document. If the morphological analysis unit is not the first in the determination in step 73, the process directly proceeds to step 75.

Subsequently, in order to set the analysis target character string type in the entry of the next morpheme analysis unit, next, in step 75, the variable j is incremented, and in the next step 76, j ≦ n. Or not. If j ≦ n, there is an entry corresponding to the morpheme analysis unit that has not been set yet, so the process returns to step 72 and the processing from step 72 is repeated. Also, j ≦
If not n, the processing here is terminated. That is, the variable j is compared with the number n of morphological analysis units stored in the keyword extraction management table 30, and if j is equal to or smaller than n, the process returns to step 72. If not, the process ends.

As a result, in the keyword extraction management table 30, the setting process of the character string set to be analyzed for each morphological analysis unit is completed. As the character string set to be analyzed here, for example, the process of setting all of the text of the registered document, the set of character strings of the unregistered keyword candidate group, etc. is completed, each morphological analysis unit, The keyword extraction processing is performed according to the setting contents.

Next, the keyword extraction process in the multilingual document registration process will be described. Please refer to the flowchart of FIG. As described above, this process is a process executed in step 46 of the overall process flow of the multilingual document registration process of FIG. The keyword extraction process here is started, and when the process proceeds to step 81, in step 81, morphological analysis is performed on the character string set set in the analysis target character string type of the keyword extraction management table, and words are cut out. That is, a word is cut out by morphological analysis from a position where the morphological analysis of the character string set is not completed.

Next, at step 82, it is judged whether or not the extracted word includes an unnecessary word. If the unnecessary word is not included, the process directly proceeds to step 88,
Immediately, select words other than unnecessary words as keywords,
This is the end of the process. If the extracted word includes an unnecessary word, the process proceeds to step 83 and it is determined whether the variable i is “1”. If the variable i is "1", the currently used morpheme analysis unit is the first morpheme analysis unit, so there is no processing for unregistered keyword candidates. Other words are extracted as keywords, and the process here ends.

If it is determined in step 83 that the variable i is not "1", then step 8
In step 4, it is determined whether the analysis target character string of the morphological analysis whose analysis order is the i-th is an unregistered word group. If the character string to be analyzed is not an unregistered word group, the process proceeds to step 85, and character strings are collated with the unregistered keyword candidates based on the cut out words. Then, in the next step 86, the result of the character string collation is determined. If the collation is successful, the process proceeds to step 87, and the word cut out from the unregistered keyword candidates or the word subjected to the character string collation is removed,
Next, in step 88, the processing here is terminated with words other than unnecessary words being extracted as keywords.

If it cannot be determined in step 86 that the character string has been collated, the process proceeds to step 88 without performing step 87.
The process ends here with the words other than unnecessary words extracted as keywords.

Next, the setting process of the unregistered keyword candidate process in the multilingual document registration process will be described. Please refer to the flowchart of FIG. As described above, this process is a process executed in step 50 of the overall process flow of the multilingual document registration process of FIG. When this unregistered keyword candidate process is started and the process proceeds to step 91, first, it is determined whether or not the variable i indicating the currently used morphological analysis order is “1”. If the variable i is "1", as described above, the morpheme analysis unit currently used is the first morpheme analysis unit, and therefore, there is no processing for the unregistered keyword candidate, and immediately this unregistered keyword is not processed. The processing of the candidate processing ends.

If it is confirmed in step 91 that the variable i is not "1", the flow advances to step 92 to determine whether the analysis target character string of the i-th morphological analysis is an unregistered word group. To determine. That is, in the keyword extraction management table 30, the analysis order is i-th.
It is determined whether the setting of the analysis target character string type field 35 of the entry corresponding to the th morpheme analysis unit is “unregistered word group”. If the result of this determination is that the analysis target character string type is “unregistered word group”, in step 95,
The keyword extracted from the unregistered keyword candidates is removed, and this processing ends.

If it is determined in step 92 that the analysis target character string type is not "unregistered word group", step 93
In step 94, the extracted keyword is subjected to character string matching with the unregistered keyword candidate, and in the next step 94, the result of this character string matching is determined. If the result of this determination is that the character strings can be collated, the processing proceeds to step 95,
The keyword extracted from the unregistered keyword candidates is removed, and this processing ends. If the character string collation cannot be performed, this process is ended as it is.

In this way, the processing for the unregistered keyword candidates is performed, and as a result, the unregistered keyword candidates are processed as the unregistered keyword candidates in the previous processing of the morphological analysis unit, but are processed as the keyword candidates in the processing of the subsequent morphological analysis unit. The character string matching is performed on the selected word, and the matching word is removed from the unregistered keywords.

Next, in the case of registering a document including a sentence described in a plurality of languages, a document including a sentence described in a plurality of languages will be specifically exemplified and an operation example thereof will be described. FIG. 9 is a diagram showing an example of a multilingual document. As shown in FIG. 9, the multilingual document 99 here is a document in which Japanese and English sentences exist, and a case where the multilingual document 99 is newly registered will be described. In this case, the entire document is first subjected to morphological analysis in Japanese, and then the unanalyzed portion is subjected to morphological analysis in English, keyword extraction is performed, and the keyword (document Registration number).

In this multilingual document registration / retrieval device, "English", "Japanese", "Chinese", and "Arabic"
When a morphological analysis unit capable of performing morphological analysis corresponding to each of the corresponding four languages is provided, the keyword extraction management table 30 that defines the conditions for keyword extraction, as shown in FIG. The conditions for controlling the morphological analysis unit are set. Therefore, in this case, the number of morphological analysis units registered in the keyword extraction management table 30 is counted as “4” (n = 4) (step 61: FIG. 5), and as shown in FIG. It is temporarily stored in the morphological analysis management table 36 in a record format (or as a variable).

Further, in setting the conditions of the keyword extraction management table 30, the order of analyzing the documents by each morphological analysis unit is, for example, "Japanese", "English", "Arabic", "Chinese". Therefore, in the order field 33 of the keyword extraction management table 30, the order is “2”, “1”, “4”, “in order from the top in correspondence with the morphological analysis unit corresponding to each language. 3 ”(step 62: FIG. 5).

Further, since the types of languages of the morphological analysis unit used in the keyword extraction process in the document registration process are “Japanese”, “English”, and “Chinese” here, the keyword extraction is performed. In the management table 30, the use flag field 34 of the entry corresponding to “Japanese”, “English” and “Chinese” is set to “ON”, and the use flag field 34 of the entry corresponding to “Arabic” is set to “OFF”. And

Further, in order to efficiently execute the keyword extraction process by specifying the range of the character string to be analyzed by each morphological analysis unit in the morphological analysis in the keyword extraction process in the document registration process, 4 In the morphological analysis processing set for the morphological analysis unit corresponding to each of the two languages, the second and subsequent ones do not always cover the entire document in which the character string group to be analyzed is always registered. Specifies the target text or range of character strings or a set of them.

Therefore, as shown in FIG. 3C, in the analysis target character string type setting table 37, the analysis target character string type corresponding to the range of the character string to be subjected to the morphological analysis defined in advance is defined as a keyword. Extraction management table 30
Is set in the analysis target character string type field 35. In this example, in the analysis target character string type field 35 of the keyword extraction management table 30, “unregistered word group”, “text-ALL”, “unregistered word group”,
"Text-range specification" is set, and the Japanese morphological analysis unit targets the entire document for analysis, but the English morphological analysis unit and the Chinese morphological analysis unit target the analysis target. Unregistered word group. In this case, the first morphological analysis unit that performs the morphological analysis is forcibly set to "text-ALL" by default so that the entire document of the registered document is always analyzed first (step 73-Step 74: FIG. 6).

In this way, when conditions such as the order of each morphological analysis unit used in the keyword extraction management table 30 and the range of character strings to be analyzed are set, each morphological analysis unit is controlled according to the set conditions. Then, the keyword extraction process is executed. When the keyword extraction process is started, first, a morphological analysis is performed using the morphological analysis unit whose order is set first. In this example, after confirming that the "usage flag" is "ON" by the morphological analysis unit for "Japanese corresponding" in the first order (step 45), the morphological analysis unit is supported. Then, keyword extraction is performed on the set of character strings.
That is, in this case, since "text-ALL" is set, keyword extraction is performed for all the texts of the registered document (step 46).

In the keyword extraction process (FIG. 7), the extracted words other than those which are judged as unnecessary words are used as keywords. Since the "Japanese" morphological analysis processing is the first analysis processing, unnecessary words are not processed for the unregistered keyword candidates. Then, it is determined whether or not the keyword extracted next is registered in the dictionary for Japanese morphological analysis (step 4).
7) The registered information is stored as a keyword candidate together with the document ID (step 48).

10 (a) and 10 (b) show the contents of the keyword candidate holding unit and the unregistered keyword candidate holding unit at the stage when the keyword extraction processing by the Japanese-compatible morpheme analysis unit is completed in comparison. It is a figure. For example, if the identification number (document ID) of the document to be registered (FIG. 9) is “202
04 ”, when the registration process of all the keyword candidates by the morphological analysis of“ Japanese ”is completed, FIG.
As shown in (a), the keyword candidates 102 for the document (fileID) 101 in the keyword candidate holding unit 100 include the words “Iraq” and “Kuwait” extracted by morphological analysis as keyword candidates for the document with document ID = 20204. , "United Nations", ..., "Bomber" is stored. On the other hand, if the unregistered keyword candidates are not registered in the morphological analysis dictionary, the process of registration of all unregistered keyword candidates by the "Japanese" morphological analysis is completed, as shown in FIG.
As shown in FIG. 1, the unregistered keyword candidate 1 for the document (fileID) 104 of the unregistered keyword candidate holding unit 103
In FIG. 05, as the unregistered keyword candidates for the document with the document ID = 20204, similarly, the words “Patriot”, “The”, “Ministr” extracted by the morphological analysis are used.
“Y”, “of”, ..., “Recently” are stored.

In this way, when the process of morphological analysis of "Japanese" is completed, the process of morphological analysis of "English" of the next corresponding language is started. In this case, as in the case described above, according to the conditions of the keyword extraction management table 30, it is confirmed that the "usage flag" is also "ON" by the morphological analysis unit of the second "English" in the order. After that (step 45), keyword extraction is performed on the character string set set corresponding to this morphological analysis unit. That is, in this case, "unregistered word group" is set as the analysis target character string type in the analysis target character string type field 35 of the keyword extraction management table 30, so that it is shown in FIG. As described above, the document (fi
The keyword extraction processing is performed on the character string stored in the unregistered keyword candidate 105 for (leID) 104. That is, the document ID that is the current registration target document
= 20204, the keyword extraction processing is performed on the character string previously extracted as the unregistered keyword candidate without performing the English-based morphological analysis of all the documents (step 46).

Processing of keyword extraction in this case (FIG. 7)
In the same manner as in the case described above, keywords other than the ones that are judged as unnecessary words among the cut out words are used as the keywords. In other words, by this processing, for example, “The” and “of” that are determined to be unnecessary words are excluded from the unregistered keyword candidates as unnecessary word processing for the unregistered word keyword candidates. Then, it is determined whether or not the extracted keyword is registered in the English morphological analysis dictionary (step 47), and the registered one is stored as a keyword candidate together with the document ID (step 48).

11 (a) and 11 (b) compare the contents of the keyword candidate holding unit and the unregistered keyword candidate holding unit at the stage when the next keyword extraction processing by the English-compatible morphological analysis unit is completed. FIG. As described above, if the identification number (document ID) of the document to be registered here (FIG. 9) is “20204”, all the registration processing by the morphological analysis of “English” is completed for the unregistered keyword candidate. 10B, the document (fileID) 104 of the unregistered keyword candidate holding unit 103 shown in FIG.
Character strings “Ministry”, “Education”, ..., “sai stored in the unregistered keyword candidate 105 corresponding to
For "d" and "recently", a morphological analysis is performed by an English-compatible morphological analysis unit, and as a result, among the extracted words, the words registered in the English morphological analysis dictionary are As shown in FIG. 11A, it is additionally stored in the keyword candidate 112 for the document (fileID) 111 of the keyword candidate holding unit 110. That is, document ID = 202
As a keyword candidate for the 04 document, "ministry", "education", ..., "sa" are added to the corresponding entries.
It is additionally stored as "y" and "recent".

Although not particularly mentioned in the morphological analysis processing described in this description, when cutting out a word by the morphological analysis, processing for unifying several expressions of the word into a standard form is also performed at the same time. In other words, "Ministry" → "ministr
The process of unifying uppercase letters to lowercase, such as “y” and “Education” → “education”, and the process of unifying to the original form, such as “said” → “say”, are performed. In this way, words that are unregistered keyword candidates in one language morphological analysis are extracted as keyword candidates by performing morphological analysis in another language, and the keywords extracted as the keyword candidates are unregistered. The process of removing from the registered keyword candidates is performed.

In this way, the data which is not registered in the dictionary for morphological analysis corresponding to English, in the case of this example, "Mo
Since "nbushou" remains, it is stored as an unregistered word keyword candidate. At the stage when all the keyword extraction processing by the morphological analysis of “English” is completed, as shown in FIG.
In FIG. 5, “patriot” and “Monbushou” are stored in the corresponding entries as unregistered keyword candidates for the document of document ID = 20204.

When the processing by the morphological analysis of "English" is completed in this way, the processing by the morphological analysis unit corresponding to the language of the third order is subsequently started. That is, the processing is started by the morphological analysis unit corresponding to the next "Arabic" in the corresponding language. However, in the keyword extraction management table 30, the "use flag" of the morphological analysis unit for "Arabic" is "OFF". In this case, as in the case described above, according to the conditions of the keyword extraction management table 30, in this case, the “use flag” of the morphological analysis unit corresponding to the third “Arabic” is “ON”. "Cannot be confirmed (step 45), and the processing by the morphological analysis unit corresponding to" Arabic "is skipped.

In this way, when the processing by the morphological analysis unit corresponding to the language of the third order is skipped,
Then, the process by the morphological analysis unit corresponding to the fourth order language is started. In this case as well, as in the case described above, the process proceeds according to the conditions of the keyword extraction management table 30. In this case, the “usage flag” of the morphological analysis unit corresponding to the fourth “Chinese” in the order is “ON”.
Since it can be confirmed (step 45),
By the morpheme analysis unit corresponding to this "Chinese language", the keyword extraction is performed on the character string set set to the correspondence. In this case, since "unregistered word group" is set as the analysis target character string type in the analysis target character string type field 35 of the keyword extraction management table 30, the unregistered keyword candidates shown in FIG. Holding part 1
The morpheme analysis is then performed on the character strings stored in the unregistered keyword candidates 115 for the 13 documents (fileID) 114, and the keyword extraction processing is performed.
That is, the document ID of the current registration target document = 202
For the document No. 04, the keyword extraction processing is continuously performed on the character string previously extracted as the current unregistered keyword candidate (step 46).

Processing for keyword extraction in this case (FIG. 7)
In the same manner as in the case described above, the keywords other than the ones that are judged as unnecessary words among the cut out words are the keywords, but there is no corresponding keyword, and the unregistered keyword candidate holding unit 113 does not register them. Since there is nothing corresponding to “Chinese” in the character string stored as the keyword candidate 115, even if the processing of keyword extraction by the morphological analysis of “Chinese” is completed, FIG. As shown in FIG. 11B, the keyword candidate storage unit 11
There is no change in the contents of 0 and the unregistered keyword candidate storage unit 113.

In this way, when the keyword extraction processing by morphological analysis for all languages is completed, the index table (120: FIG. 12) is created according to the contents of the keywords extracted by the processing so far. As shown in FIG. 12, the index table 12
0 is an index for multilingual document search,
A document (fileID) 122 and an unregistered word flag 123 of the document corresponding to each extracted keyword 121.
Is a registered table. The index registered in the index table 120 here is based on the word of the keyword 121, and the corresponding document (file
The document ID of (ID) 122 is sorted, and when there is an unregistered keyword for the document of the document ID, an unregistered flag to that effect is set. When such an index table 120 is created, the document registration work of the multilingual document is completed here.

Next, a case where a multilingual document search is performed using the index of the index table 120 thus created will be described.

When a user searches for a desired document,
As described above, the user inputs the search condition using the search condition input unit (11: FIG. 1). When the search condition is input, the multilingual index matching unit (12: FIG. 1) cuts out a word from the search condition expression of the input search condition, and matches the cut out word with the keyword of the index of the index table. By this matching result,
The text extraction unit (14: FIG. 1) reads out a document that matches the search condition based on the matching result between the index and the word, and the display unit (13: FIG. 1) displays the read document.

FIG. 13 is a block diagram showing the structure of the main part of the multilingual index matching unit. FIG. 13 shows a block of each element in the multilingual index matching unit and a data flow in the case of performing a morphological analysis from a search condition to determine a search expression. In FIG.
3 is a text database part, 11 is a search condition input part,
14 is a text extraction unit, 131a is a first morphological analysis unit, 131b is a first dictionary file unit, 132a
Is a second morpheme analysis unit, 132b is a second dictionary file unit, and 133a is an Nth morpheme analysis unit. Reference numeral 134 is a search word candidate holding unit, 135 is an unregistered search word candidate holding unit, 136 is a search formula determining unit, and 137 is an order setting unit.

As shown in FIG. 13, the multilingual index collating unit has a plurality of languages corresponding to respective languages for performing morphological analysis of sentences in the languages of the respective countries of the search condition expression input as the search conditions. Morphological analysis unit (131a to 133)
a), and a plurality of dictionary file units (131b to 131b) corresponding to each national language, which supply dictionary data of each language to each morphological analysis unit.
133b) is provided, and in order to efficiently perform the morpheme analysis of the multilingual search condition by controlling these plural morpheme analysis units (131a to 133a), the morpheme of the multilingual search condition is included. An order setting unit 137 that sets the order of analysis, a search word candidate holding unit 134 that temporarily stores the words of the analyzed search conditions as search word candidates as its working memory, and one language support For the words of the search condition that are not analyzed by the morpheme analysis unit, the unregistered search word candidate holding unit 13 that is temporarily registered in order to perform the morpheme analysis by the morpheme analysis unit corresponding to another language.
5 are provided. Then, when the morphological analysis of the search condition expression of the search condition is completed, the search expression determination unit 136 determines the word to be used as the search expression, and the text extraction unit 14
Thus, the document matching the search condition is extracted from the text database unit 3 by the document ID of the index.

FIG. 14 is a flowchart showing the overall processing flow of multilingual document search processing. The flowchart shown in FIG. 14 cuts out a word to be used as a search word by performing a morphological analysis on the search condition expression in the search condition described in one or more languages corresponding to each language, and performs the search. It shows the overall flow of processing for creating a formula and searching for a document. FIG. 15 is a flowchart showing the processing flow of the search word extraction processing in the multilingual document search processing, and FIG. 16 shows the processing flow of the unregistered search word candidate processing in the multilingual document search processing. It is a flowchart.

First, the overall processing of the multilingual document search processing will be described with reference to the flowchart of FIG. When the process is started, first, in step 141, the search expression input unit 11 inputs a multilingual search condition expression for document search. Next, in step 142, a search term extraction condition setting process is performed, and subsequently, in the next step 143, a character string set to be analyzed by morphological analysis is set. That is, in this case, similarly to the keyword extraction management table in the keyword extraction processing described above, the condition of the morphological analysis unit for cutting out the search word is set by the search word extraction management table (not shown), and Then, the set of character strings to be analyzed is further set using the morphological analysis unit for which the conditions have been set. In the character string set setting process, for example, all of the text of the search condition expression or a set of character strings of the unregistered search word candidate group (unregistered word group) is set, and the morphological analysis is specifically performed. Set the target character string set. It should be noted that the keyword extraction management table 30 described above may be used as the search word extraction management table here without providing a search word extraction management table.

After the processing of these settings is completed, next, in order to control the extraction processing of the search word for creating the search expression, in step 144, the variable i indicating the order of the morphological analysis unit currently used is set to " 1 "is set, and the morphological analysis unit used at the beginning of the processing of the morphological analysis of the search condition expression is set. Next, in step 145, it is determined whether the usage flag of the morphological analysis unit of the language corresponding to the variable i is on. If the use flag is not on, the morphological analysis unit with the number of the variable i does not perform the morphological analysis process, so that the process proceeds to the process of the morphological analysis unit with the next number.
Proceed to 2.

When the use flag of the morphological analysis unit of the language corresponding to the variable i is turned on in the determination of step 145, that is, the language corresponding to the variable i indicating the order in the search word extraction management table (“analysis If the "use flag" of the morphological analysis entry for the language whose "order to perform" is i) is ON, the process proceeds to the next step 146, and the i-th corresponding morphological analysis is performed for the character string set. Sequential extraction processing of search words (FIG. 15) is performed. In this search word extraction process, as will be described later, with respect to the character string set set by the type of the analysis target character string of the entry of the ith morphological analysis unit in the i-th order, based on the condition data of the search word extraction management table. The search word (keyword) is extracted by the i-th corresponding morphological analysis unit.

Next, in step 147, it is determined whether or not the extracted search word is registered in the (i-th) morphological analysis dictionary. If the result of determination in step 147 is that it is not registered, the process proceeds to step 148, where it is stored in the unregistered search word candidate holding unit together with the corresponding search condition expression as an unregistered search word candidate, and the process proceeds to step 151.
Further, when it is determined in step 147 that the extracted search word is registered in the morphological analysis dictionary, the process proceeds to step 149, and the search condition candidate holding unit together with the corresponding search condition expression as the search word candidate. Remember. Then, in step 150, unregistered search word candidate processing (FIG. 16) is performed. In this unregistered search word candidate process, as will be described later, the word of the unregistered search word candidate that was the unregistered search word candidate in the process of the previous morpheme analysis unit Column matching is performed, and the matched word is removed from the unregistered search words. When this unregistered search word candidate process is completed, the process proceeds to step 151.

In step 151, it is determined whether or not the retrieval word extraction by the i-th morpheme analysis unit is completed. If the search word extraction is not completed, step 146
Then, the process from step 146 is repeated. Further, if it is possible to confirm the end of the search word extraction process by the morpheme analysis unit corresponding to the i-th by the determination process of step 151, the search word extraction process is performed by the next morpheme analysis unit. The variable i indicating the order of the morphological analysis units to be used is incremented, that is, the variable i is counted up (when i = i + 1), and in the next step 153, the number n of morphological analysis units corresponding to each national language that can be used is n. And a variable i indicating the order of the morphological analysis unit to be used next.

If n ≧ i as a result of this comparison, it means that the search word extraction processing by the i-th morpheme analysis unit has not been completed, so the processing returns to step 145 and the processing from step 145 is repeated. Further, if n <i, it means that all the analysis by the morphological analysis unit which is in the usable state according to the condition set in the search word extraction management table has been completed, so the process proceeds to step 154 and the search word candidates A process of creating a search expression from the group and the unregistered search word group is performed. As a result, a search expression for multilingual document search is created. Next, in step 155, the created search expression is compared with the keyword of the index to extract the corresponding document. The process ends.

Next, the search word extraction process in the multilingual document search process will be described. Please refer to the flowchart of FIG. As described above, this process is a process executed in step 146 of the overall process flow of the multilingual document search process of FIG. The search word extraction processing here is started, and when proceeding to step 161, in step 161, a morphological analysis is performed on the character string set of the search condition expression set in the analysis target character string type of the search word extraction management table. Cut out words. That is, a word is cut out by a morphological analysis process from a position where the previous morphological analysis of the character string set is not completed.

Next, in step 162, it is judged whether or not the cut-out word includes an unnecessary word. If the unnecessary word is not included, the process proceeds to step 168 as it is, and a word other than the unnecessary word is immediately set as a search word,
This is the end of the process. If the extracted word includes an unnecessary word, the process proceeds to step 163 and the variable i
Is "1". If the variable i is "1", the morpheme analysis unit currently used is the first morpheme analysis unit, so there is no processing for unregistered search word candidates. The process here is terminated as a search word obtained by extracting words other than words.

If it is determined in step 163 that the variable i is not "1", then step 164
Then, it is determined whether the analysis target character string of the morphological analysis whose analysis order is the i-th is an unregistered word group. If the analysis target character string is not an unregistered word group, the process proceeds to step 165, and the unregistered search word candidates are subjected to character string matching by the cut-out words. Then, in the next step 166,
The result of the character string matching is determined. If the collation is successful, the process proceeds to step 167, the word cut out from the unregistered search word candidates or the word subjected to the character string collation is removed, and then, in step 168, a word other than the unnecessary word is extracted as a search word, This process ends.

If it cannot be determined in step 166 that the character string can be collated, the process proceeds to step 168 without performing the process of step 167, and the search word obtained by extracting words other than unnecessary words is searched. Then, the processing here is ended.

Next, the unregistered search word candidate process in the multilingual document search process will be described. Please refer to the flowchart of FIG. As described above, this process is step 15 of the overall process flow of the multilingual document search process of FIG.
This is the process executed at 0. When this unregistered search word candidate process is started and the process proceeds to step 171, first, the variable i indicating the order of the currently used morphological analysis unit is “1”.
Is determined. If the variable i is “1”, as described above, the morpheme analysis unit currently used is the first morpheme analysis unit, so there is no processing for the unregistered search word candidate, and this unregistered immediately. The processing of the search word candidate processing ends.

If it is confirmed in step 171 that the variable i is not "1", step 17
The process proceeds to 2 and it is determined whether the analysis target character string of the morphological analysis whose analysis order is the i-th is an unregistered word group. That is,
In the control table of the search word extraction condition, it is determined whether the setting of the analysis target character string type field of the entry corresponding to the i-th morpheme analysis unit in the analysis order is “unregistered word group”. If the result of this determination is that the analysis target character string type is “unregistered word group”, in step 175 the search word extracted from the unregistered search word candidates is removed,
This process ends.

If the analysis target character string type is not "unregistered word group" in the determination of step 172, step 1
Proceeding to 73, the extracted search word is character string collated with the unregistered search word candidate, and in the next step 174,
The result of this string matching is determined. If the result of this determination is that the character string collation has succeeded, the routine proceeds to step 175, the search word extracted from the unregistered search word candidates is removed, and this processing ends. If the character string collation cannot be performed, this process is ended as it is.

In this way, the processing for the unregistered search word candidate is performed, and as a result, the unregistered search word candidate is determined as the unregistered search word candidate in the processing of the previous morpheme analysis unit, but is searched for in the processing of the subsequent morpheme analysis unit. Character string matching is performed on the candidate words, and processing for removing the matched words from the unregistered search words is performed.

Next, in the case of retrieving a multilingual document with a retrieval condition expression of retrieval conditions including sentences described in a plurality of languages, the operation condition will be described by exemplifying the retrieval condition expression. . FIG. 17 is a diagram showing an example of a search condition expression of a search condition including a sentence described in multiple languages. As shown in FIG. 17, the search condition 179 here is a search condition expression of a sentence in which Japanese and English sentences exist.
A case where a search word is cut out from the sentence of 79 and a document search is performed will be described. In this case, first perform a morphological analysis in Japanese on the entire search condition expression written in multiple languages, then
Morphological analysis in English is performed to extract a search word, a search expression is generated from the extracted search word, and the corresponding document is searched.

In this multilingual document registration / retrieval device, "English", "Japanese", "Chinese", and "Arabic"
When a morphological analysis unit that can perform morphological analysis corresponding to each of the four corresponding languages is provided, the search word extraction management table that defines the conditions for the search word extraction is shown in FIG.
It is assumed that the control conditions of each morphological analysis unit are set with the same contents as the keyword extraction management table 30 shown in (a). Here, although the contents of the search word extraction management table are not shown in particular, the keyword extraction management table 30 shown in FIG. 3A is referred to as the same as the search word extraction management table, if necessary. Therefore, the number of morphological analysis units registered in the search word extraction management table in this case is counted as “4” (n = 4), and similarly, the morphological analysis management table (36: FIG.
In (b)), it is temporarily stored in a record format (or as a variable).

In setting the conditions of the search word extraction management table, the order in which the documents are analyzed by each morphological analysis unit is
For example, in order of "Japanese", "English", "Arabic", and "Chinese", the order of "parsing order" in the order field of the search word extraction management table (keyword extraction management table 30) is: Corresponding to the morphological analysis unit corresponding to each language, the order is “2”, “1”,
Set "4" and "3".

Since the types of languages of the morphological analysis unit used in the search word extraction process are "Japanese", "English", and "Chinese" here, the search word extraction management table (keyword extraction In the management table 30),
The use flags of the entries corresponding to “Japanese”, “English”, and “Chinese” are set to “ON”, and the use flags of the entries corresponding to “Arabic” are set to “OFF”.

Further, in order to efficiently execute the search word extraction processing, the range of the character string to be analyzed by each morphological analysis unit in the morphological analysis processing in the search word extraction processing in the document search processing is specified. In the morphological analysis processing set for the morphological analysis unit corresponding to each of the four languages after the second, the character string group to be analyzed is not always covered by the entire sentence of the search condition expression. Instead, specify the range of text or character strings to be analyzed or a set of them.

Therefore, as in the case of keyword extraction, as shown in FIG. 3C, in the analysis target character string type setting table 37, it corresponds to a range of character strings to be subjected to morphological analysis which is defined in advance. The analysis target character string type is set as the analysis target character string type of each entry in the search word extraction management table (keyword extraction management table 30). In this example, as the analysis target character string type of the search word extraction management table, "unregistered" is associated with each of the morphological analysis units corresponding to "English", "Japanese", "Chinese", and "Arabic". Set "word group", "text-ALL", "unregistered word group", and "text-range specification".

Therefore, in this case, the Japanese-compatible morphological analysis unit targets the entire document as an analysis target, whereas the English-compatible morphological analysis unit and the Chinese-compatible morphological analysis unit target the analysis target as an unregistered word group. I am trying. Note that the first morphological analysis unit that performs morphological analysis is forcibly set to "text-ALL" by default to always analyze the entire registered document at first.

In this way, when the conditions such as the order of each morphological analysis unit to be used, the analysis target character string range, etc. are set by the search word extraction management table (keyword extraction management table 30), according to the set conditions. Each morphological analysis unit is controlled to execute the search word extraction process. When the search word extraction process is started, first, the morphological analysis is performed using the morphological analysis unit whose order is set first. In this example, the first order is "Japanese"
The "use flag" is "ON" by the morphological analysis unit of
After confirming that (step 145),
Search word extraction is performed on the character string set set corresponding to this morphological analysis unit. That is, in this case, since "text-ALL" is set, the search word is extracted for all the texts of the search condition expression (step 14).
6).

In the retrieval word extraction process (FIG. 15),
A process is performed in which words other than those that are determined to be unnecessary words among the cut out words are used as search words. Since the "Japanese" morphological analysis processing is the first analysis processing, unnecessary words are not processed for unregistered search word candidates. Then, it is judged whether or not the extracted search word is registered in the dictionary for Japanese morphological analysis (step 14).
7) The registered one is stored as a search word candidate together with the corresponding search condition expression (step 14).
8).

18 (a) and 18 (b) compare the contents of the search word candidate holding unit and the unregistered search word candidate holding unit at the stage when the search word extraction processing by the Japanese morphological analysis unit is completed. FIG. For example, for a multilingual sentence “withdrawal of Iraqi troops and Patriot missile and Ministry of Education” of a search condition expression 179 as shown in FIG. 17, registration processing of all search word candidates by morphological analysis of “Japanese” is performed. At the end stage, as shown in FIG. 18A, the search word candidates 181 in the search word candidate holding unit include, as search word candidates, the words “Iraq”, “unit”, and “withdrawal” of the words cut out by the morphological analysis. , And “missile” are stored, while those that are not registered in the dictionary for morphological analysis are displayed when the registration processing of all unregistered search word candidates by morphological analysis of “Japanese” is completed. As shown in 18 (b), the unregistered search word candidate 1 in the unregistered search word candidate holding unit 1
Similarly, in 82, "patriot", "Ministry", "of", which are cut out by morphological analysis of "Japanese",
“Education” is stored as an unregistered search word candidate.

In this way, when the process of morphological analysis of "Japanese" is completed, the process of morphological analysis of "English" of the next corresponding language is subsequently started. In this case, as in the case described above, the processing is performed by the morphological analysis unit of the second "English" in the order according to the conditions of the search word extraction management table (keyword extraction management table 30). The "use flag" is also "O"
After confirming that it is N "(step 14
5) The search word is extracted from the character string set set corresponding to this morpheme analysis unit. That is, in this case, the search word extraction management table (keyword extraction management table 3
Since "unregistered word group" is set as the analysis target character string type of 0), the character string of the unregistered search word candidate 182 in the unregistered search word candidate holding unit is set as shown in FIG. 18B. Then, the process of extracting the search word is performed (step 14).
6).

In the retrieval word extraction process (FIG. 15) in this case, as in the case described above, the extracted words other than those which are judged as unnecessary words are used as retrieval words. That is, by this process, as an unnecessary word process for the unregistered word search word candidate, “of” that is determined to be an unnecessary word
Is excluded from the unregistered search word candidates. Then, it is determined whether or not the extracted search word is registered in the English morphological analysis dictionary (step 147), and the registered one is stored as a search word candidate together with the corresponding search condition expression (( Step 148).

19 (a) and 19 (b) show the contents of the search word candidate holding unit and the unregistered search word candidate holding unit at the stage when the search word extraction processing by the next English morphological analysis unit is completed. It is a figure shown in contrast. At the stage where all the registration processing by the morphological analysis of "English" is completed for the unregistered search word candidate 182 of FIG. The morpheme analysis is performed by the morpheme analysis unit, and as a result, the words registered in the English morpheme analysis dictionary are extracted as search word candidates, as shown in FIG. 19 (a). In the search term candidate 191 in the holding unit, “mi
"nistry" and "education" are additionally stored. Also, as in the case described above, in the morphological analysis process of this description, although not particularly touched, when cutting out a word by morphological analysis, a process of unifying uppercase letters to lowercase letters to standard forms But at the same time. In this way, in the morpheme analysis of one language, the words that are unregistered search word candidates are extracted as the search word candidates by performing the morpheme analysis in another language, and are extracted as the search word candidates. Processing for removing a search word from unregistered search word candidates is performed.

As a result, a word that is not registered in the dictionary for morphological analysis corresponding to English, that is, "Patriot" remains in this example, and is stored as an unregistered word search word candidate. At the stage when the processing of extracting all the search words by the morphological analysis of "English" is completed, as shown in FIG.
“Patriot” is stored as the unregistered search word candidate 192 in the unregistered search word candidate holding unit.

When the processing by the morphological analysis of "English" is completed in this way, the processing by the morphological analysis unit corresponding to the language in the third order is subsequently started. That is, the processing is started by the morphological analysis unit corresponding to the next "Arabic" in the corresponding language. However, in the search term extraction management table (keyword extraction management table 30), the morphological analysis unit of the "Arabic""Useflag" is "O"
In this case, as in the case described above, the morpheme analysis unit corresponding to the third "Arabic" corresponding to the condition of the search word extraction management table (keyword extraction management table 30) is used. "O" of "Use Flag"
N "cannot be confirmed (step 145), and the processing by the morphological analysis unit corresponding to" Arabic "is skipped.

In this way, if the processing by the morphological analysis unit corresponding to the language of the third order is skipped,
Then, the process by the morphological analysis unit corresponding to the fourth order language is started. Also in this case, as in the case described above, the process proceeds according to the conditions of the search word extraction management table (keyword extraction management table 30). In this case, it can be confirmed that the "use flag" of the morpheme analysis unit corresponding to the fourth "Chinese" in the order is "ON" (step 45). The analysis unit extracts the search word from the character string set set to the correspondence. In this case, since the analysis target character string type of the search word extraction management table (keyword extraction management table 30) is set to “unregistered word group” as the analysis target character string type, FIG. The unregistered search word candidate 192 in the unregistered search word candidate holding unit shown
Then, the morpheme analysis is performed on the character string stored in, and the search word is extracted (step 14).
6).

In the process of extracting the search word in this case (FIG. 15) as well, similar to the above-described case, a process is performed in which the extracted words other than those which are judged as unnecessary words are used as the search words. , "Chinese" does not apply to the character string stored as the unregistered search word candidate 192 in the unregistered search word candidate holding unit, and therefore "Chinese" does not apply. 19A and 19B, there is no change in the contents of the search word candidate storage unit and the unregistered search word candidate storage unit, even if the processing of the search word extraction by the morphological analysis is completed. .

In this way, when the process of extracting the search word by the morphological analysis for all languages is completed, the process shown in FIG.
As shown in, the search expression 200 is created. Search formula 20
0 is an array of a search key 201 and an unregistered word flag 202 for a multilingual document search, and the search key 201 and the unregistered word flag 202 are changed according to the search formula 200.
For example, an index table 12 as shown in FIG.
The index (keyword, document ID, unregistered word flag) of 0 is collated, and the multilingual document (FIG. 9) is read from the corresponding document ID.

[0123]

As described above, according to the multilingual document registration / retrieval apparatus of the present invention, there are documents or search conditional expressions that include sentences and are described in languages of one or more countries, and those documents are registered. In addition, when performing a search using a search condition expression,
As much as possible, the processing of the morphological analysis unit for the written language is combined to make it possible to extract words with the highest accuracy. Therefore, the size of the index created at the time of registration can be made compact. Further, also in the document search, it is possible to improve the accuracy (recall rate) of collation with the index. In addition, by embedding control information of conditions for keyword extraction such as “order of analysis”, “use flag”, and “type of analysis target character string” in the keyword extraction management table, the target text is simply duplicated in each morphological analysis. It becomes possible to analyze the data optimally and efficiently according to the situation, instead of performing the analysis. By customizing these conditions by the user,
It is possible to arbitrarily register and search multilingual documents according to users.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a multilingual document registration / retrieval device according to an embodiment of the present invention,

FIG. 2 is a block diagram showing a configuration of a main part of a multilingual keyword extraction unit,

FIG. 3 is a diagram for explaining the contents of a control table used when performing keyword extraction processing;

FIG. 4 is a flowchart showing the overall processing flow of multilingual document registration processing;

FIG. 5 is a flow chart showing a processing flow of a keyword extraction management table condition setting process in the multilingual document registration process;

FIG. 6 is a flowchart showing a processing flow of setting processing of an analysis target character string range in the multilingual document registration processing;

FIG. 7 is a flowchart showing a processing flow of keyword extraction processing in the multilingual document registration processing,

FIG. 8 is a flowchart showing a processing flow of unregistered keyword candidate processing in the multilingual document registration processing,

FIG. 9 is a diagram showing an example of a multilingual document,

10 (a) and 10 (b) show the contents of the keyword candidate holding unit and the unregistered keyword candidate holding unit at the stage when the keyword extraction processing by the morphological analysis unit for Japanese is completed. Figure,

11 (a) and 11 (b) compare the contents of the keyword candidate holding unit and the unregistered keyword candidate holding unit at the stage when the keyword extraction processing by the next English morphological analysis unit is completed. Figure showing,

FIG. 12 is a diagram showing an example of a created multilingual index table,

FIG. 13 is a block diagram showing a configuration of a main part of a multilingual index matching unit,

FIG. 14 is a flowchart showing the overall processing flow of multilingual document search processing;

FIG. 15 is a flowchart showing a processing flow of a search word extraction processing in the multilingual document search processing;

FIG. 16 is a flowchart showing a processing flow of unregistered search word candidate processing in the multilingual document search processing;

FIG. 17 is a diagram showing an example of a search condition expression of a search condition including a sentence written in multiple languages;

FIG. 18A and FIG. 18B show the contents of the search word candidate holding unit and the unregistered search word candidate holding unit at the stage when the search word extraction processing by the Japanese morphological analysis unit is completed. Figure shown in contrast,

19 (a) and 19 (b) show the contents of the search word candidate holding unit and the unregistered search word candidate holding unit at the stage when the search word extraction processing by the next English morphological analysis unit is finished. Figure showing contrasting,

FIG. 20 is a diagram showing an example of a search expression generated by a multilingual search condition expression.

[Explanation of symbols]

1 ... Input processing unit, 2 ... Multilingual keyword extraction unit, 3 ... Text database unit, 4 ... Index registration unit, 5 ...
Index file part, 11 ... Search condition input part, 12
... multilingual index registration unit, 13 ... display unit, 14 ... text extraction unit, 21a ... first morphological analysis unit, 21
b ... 1st dictionary file part, 22a ... 2nd morphological analysis part, 22b ... 2nd dictionary file part, 23
a ... Nth morphological analysis unit. 24 ... Keyword candidate holding unit, 25 ... Unregistered keyword candidate holding unit, 26
... Keyword / unregistered keyword determination unit, 27 ... Sequence setting unit, 28 ... Index registration unit, 29 ... Index file unit, 30 ... Keyword extraction management table, 31
... Number field, 32 ... Corresponding language type field, 3
3 ... order field, 34 ... use flag field, 3
5 ... Analysis target character string type field, 36 ... Morphological analysis management table, 37 ... Analysis target character string type setting table, 99 ... Multilingual document, 100 ... Keyword candidate storage unit, 101 ... Document (fileID), 102 ... Keyword candidate , 103 ... Unregistered keyword candidate storage / holding unit, 104
... document (fileID), 105 ... unregistered keyword candidate,
110 ... Keyword candidate storage unit, 111 ... Document (fileI
D), 112 ... Keyword candidates, 113 ... Unregistered keyword candidate storage / holding unit, 114 ... Document (fileID), 11
5 ... Unregistered keyword candidate, 120 ... Index table, 121 ... Extracted keyword, 122 ... Document (fileID), 123 ... Unregistered flag, 131a ... First
Th morpheme analysis unit, 131b ... first dictionary file unit, 132a ... second morpheme analysis unit, 132b ...
2nd dictionary file part, 133a ... Nth morphological analysis part, 134 ... Search word candidate holding part, 135 ... Unregistered search word candidate holding part, 136 ... Search formula determining part, 137 ... Order setting part, 179 Is a search condition, 181 ... Search word candidates, 1
82 ... Unregistered search word candidates, 191 ... Search word candidates, 192
... unregistered search word candidates, 200 ... search formula, 201 ... search key, 202 ... unregistered word flag.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Kazuo Aihara 3-2-1, Sakado, Takatsu-ku, Kawasaki-shi, Kanagawa KSP R & D Business Park Building at Fuji Xerox Co., Ltd. (72) Tatsuomi Kita Takatsu-ku, Kawasaki-shi, Kanagawa 3-2-1, Sakado KSP R & D Business Park Building, Fuji Xerox Co., Ltd. (72) Inventor Naomi Hiraoka 3-2-1 Sakado, Takatsu-ku, Kawasaki City, Kanagawa Prefecture KSP R & D Business Park Building, Fuji Xerox Co., Ltd. (72) Inventor Yuko Matsuo 3-2-1 Sakado, Takatsu-ku, Kawasaki-shi, Kanagawa KSP R & D Business Park Building Fuji Xerox Co., Ltd. (72) Inventor Hiroshi Yamaguchi 3-2-1 Sakado, Takatsu-ku, Kawasaki-shi, Kanagawa KSP R & D Business Park Building Fuji Xerox Co., Ltd. (72) Inventor Shinji Kawamoto 32-1 Sakado, Takatsu-ku, Kawasaki City, Kanagawa KSP R & D Business Park Building Fuji Xerox Co., Ltd.

Claims (9)

[Claims] 1. A multilingual document registration / retrieval apparatus for creating and registering an index used for a search for a document including sentences in a plurality of languages, and performing a document search using the index includes a sentence in a plurality of languages. Multilingual document storage means for storing documents, keyword extraction means for performing morphological analysis on the documents corresponding to sentences of different languages, and extracting document keywords, and keywords extracted by the keyword extraction means Index registration means for registering as an index with the identifier of the corresponding document, search condition input means for inputting the search condition, a word cut out from the search condition input by the search condition input means, and the cut out word and the keyword of the index Depending on the index matching means that matches with and the matching result of keywords and words A multilingual document registration / retrieval apparatus comprising: a reading unit that reads out a document that matches a retrieval condition. 2. A multilingual document registering and searching apparatus for creating and registering an index used for a search for a document including sentences in a plurality of languages, and searching the document by the index, a multilingual document for creating an index. The registration device includes a plurality of word cutout units having different cutout target languages, a setting unit configured to set processing priorities of the plurality of word cutout units, and a control unit configured to control the plurality of word cutout units according to the processing priority, and to extract words And a keyword extraction control means for extracting a keyword to extract a keyword, and an index registration means for registering the extracted keyword and the identifier of the document in which the word of the keyword is extracted in an index in association with each other. Language document registration and retrieval device. 3. The multilingual document registration / retrieval device according to claim 2, wherein the keyword extraction control unit determines a word that cannot be identified by a word cutting unit having a certain processing priority as a word having a next processing priority. For the words cut out by the cutting means, the identifiers of the words are used as keywords, and for the words that cannot be identified to the end by a plurality of word cutting means, the words are used as keywords. Language document registration and retrieval device. 4. A multilingual document registration / retrieval device that creates and registers an index to be used for a search for a document including sentences in a plurality of languages, and performs a document search by the index, in which the document registration device registers. An input means for inputting a document to instruct keyword extraction, a keyword extraction means for extracting a keyword of a document by morphological analysis, which includes a dictionary used for extracting words, and a keyword extracted by the keyword extraction means for a corresponding document A multilingual document registration / retrieval apparatus comprising: a registration unit that registers an identifier together with an identifier; and a holding unit that holds a document to be registered, an index, and a word that is not registered in the dictionary file. 5. The multilingual document registration / retrieval device according to claim 4, wherein the keyword extracting unit cuts out a word from a document composed of sentences of a plurality of languages by morphological analysis for a sentence of each language. A plurality of word cutout means, a plurality of dictionary files storing dictionaries corresponding to languages referred to by the plurality of word cutout means, and an order setting means for setting an order in which the plurality of word cutout means are applied; A multilingual document registration search, comprising: a control unit that controls a plurality of word cutting units in the order set by the order setting unit to cut out words of a corresponding multilingual sentence from the document. apparatus. 6. The multilingual document registration / retrieval device according to claim 5, further comprising: unregistered keyword candidate holding means for temporarily holding an unregistered keyword candidate for a word judged by the word cutting means as an unregistered word. , And a keyword candidate holding unit for temporarily holding as a keyword candidate the words extracted from other dictionaries, and the control unit controls the word cutting unit in the first stage to select a plurality of languages. With a document containing a sentence as an input, the words are cut out by morphological analysis, and for words that are determined to be unregistered words, they are temporarily held in the unregistered keyword candidate holding means as unregistered keyword candidates, and from the dictionary. The extracted words are retained as keyword candidates in the keyword candidate retaining means, and each word is sequentially analyzed. Controlling the output means, with the unregistered word candidates held in the unregistered keyword candidate holding means by the word cutting means in the preceding stage as an input, the words are cut out by morphological analysis, and the words judged to be unregistered words Is left in the unregistered keyword candidate holding means as it is, the word extracted from the dictionary is deleted from the unregistered keyword candidate holding means, and additionally held in the keyword candidate holding means, and finally the keyword A multilingual document characterized in that the keyword candidate held in the candidate holding means is used as a keyword, and the unregistered keyword held in the unregistered keyword candidate holding means is registered as an unregistered keyword in the index together with the identifier of the corresponding document. Registration search device. 7. A multilingual document registration / retrieval device that creates and registers an index to be used for retrieval for a document including sentences in a plurality of languages, and retrieves the document by the index. A search condition input means for inputting, a multilingual index matching means for cutting out a word from the search condition input by the search condition input means and matching it with an index, and a matching document by the matching result of the index matching means. A multilingual document registration / retrieval device, which comprises: 8. The multilingual document registration / retrieval device according to claim 7, wherein the index collating means performs morphological analysis on a sentence of a corresponding language from a document composed of a plurality of languages to find a word. A plurality of word cutout means to be cut out, an order setting means for combining one or more word cutout means to set the order of applying the word cutout means, and a search condition input means in the order set by the order setting means A multilingual document registration / retrieval apparatus comprising: a control unit that controls to cut out a word of a retrieval condition. 9. The multilingual document registration / retrieval device according to claim 8, wherein an unregistered search word candidate is temporarily held as an unregistered search word candidate with respect to a word judged by the word cutting means as an unregistered word. And a search word candidate holding means for temporarily holding a word extracted from a dictionary other than that as a search word candidate, wherein the control means controls the word cutting means in the first stage. , Inputting a document including sentences in a plurality of languages, cutting out words by morphological analysis, and regarding a word judged as an unregistered word, the unregistered search word candidate holding means is temporarily regarded as an unregistered search word candidate. The word extracted from the dictionary is stored as a search word candidate in the search word candidate holding means, and each word cutout means is sequentially controlled so that the word With the unregistered word candidates held in the unregistered search word candidate holding means by the extraction means as input, words are cut out by morphological analysis, and the words judged to be unregistered words are the unregistered search words as they are. Leave it in the candidate holding means,
The word extracted from the dictionary is deleted from the unregistered search word candidate holding means, additionally held in the search word candidate holding means, and finally the search word candidate held in the search word candidate holding means. Is used as a search word, the unregistered search word held in the unregistered search word candidate holding means is used as an unregistered search word, index matching is performed, the corresponding document is extracted by the text database unit, and the result information is output. Characteristic multilingual document registration and retrieval device.