JP2005004545A - Full-text searching device, program for processing document data, and recording medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a full-text searching device corresponding to document data in a plurality of languages. <P>SOLUTION: A registering processing means 7 causes document data inputted from an input means 1 and their language information to be registered in a document data storage part 3 along with an identifier (document identifier) which represents the document data. Also, the number of document data of that language is increased and registered in a language-based number storage part 4 and the registering processing means 7 uses a plurality of dividing process means 6 for different languages to divide the document data into index units for normalization. The index units with different notations after the normalization are registered in a full-text index storage part 5 along with information on an appearance position within the document data and the document identifier of the document data. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a full-text search device, a document data processing program, and a recording medium that search document data including a specified character string from a plurality of document data, and in particular, a document management system, an electronic library system, a patent publication search system, The present invention relates to a full-text search apparatus, a document data processing program, and a recording medium, which are extremely suitable for managing such a large amount of document data.
[0002]
[Prior art]
With the recent development of information communication technology, an environment in which a large amount of electronic documents can be accessed is being prepared. Under such circumstances, there has been proposed a document retrieval apparatus that allows a user to retrieve a desired document with high accuracy and at a high speed. This document search apparatus uses a keyword search method or a full-text search method.
[0003]
The full-text search method is a method in which an arbitrary search character string and all documents to be searched are collated and a document including the search character string is extracted without omission. This eliminates the need for assigning keywords in advance to all documents to be searched as in the keyword search method. In addition, it is possible to acquire keywords that the user wants to acquire from the Internet without omission of search.
[0004]
Note that in order to create a full-text index in full-text search, a division processing unit that divides document data into index units is necessary. As the division processing means, there are a method of dividing a connection of N characters as a unit, a method of dividing a word unit using morphological analysis, and the like.
[0005]
A method of dividing N character sequences into units is to divide document data into N character units, search for an index based on document data information including each character string and character string position information, and between character strings. This is a method of examining the positional relationship between the two and extracting corresponding document data.
[0006]
The morpheme analysis method uses a dictionary for analyzing Japanese, decomposes input document data into word units, and extracts terms that are keywords such as nouns from the decomposed word units. Is the method.
[0007]
However, full-text search technology, which is the technology for creating a large-scale index, which is the most core, cannot avoid tradeoffs in various ways.
[0008]
Therefore, there is a new full-text search device that creates an index to be used for searching document data including sentences in a plurality of languages and searches the document data (see Patent Document 1).
[0009]
This patent document 1 stores document data including sentences in a plurality of languages, and performs morphological analysis on each of the stored document data corresponding to document data in a different language. Then, the keyword of the document data is extracted and registered as an index together with the identifier of the document data corresponding to the extracted keyword. Then, a word is cut out from the input search condition, the keyword of the cut-out word and the index are collated, and document data matching the search condition is read based on the collation result. Thereby, registration processing and search processing of multilingual document data according to the user's request can be performed.
[0010]
In addition, there is a novel full-text search device that manages document data by storing an index related to multilingual document data separately for each language, and performs document data search processing using the managed document data (Patent Document 2). reference).
[0011]
This patent document 2 identifies the language of multilingual document data including characters in a plurality of languages, creates an index related to the identified multilingual data for each language, and stores it for each language. By searching the multilingual document data using the index for each language, it becomes possible to distinguish and manage for each language.
[0012]
Further, as a prior art by the present applicant, there is a full-text search device that can manage or search a plurality of document data as a single document data (see Patent Document 3).
[0013]
This patent document 3 registers document data divided into partial documents as one document data, and searches for partial document data including a designated character string from the plurality of registered document data. Extraction in units of partial documents and display of appearances of search terms can be performed.
[0014]
[Patent Document 1]
JP-A-9-50442
[Patent Document 2]
JP 2001-67368 A
[Patent Document 3]
Japanese Patent Laid-Open No. 2001-249943
[0015]
[Problems to be solved by the invention]
However, although Patent Document 1 has few restrictions because one document data is composed of a plurality of languages, a plurality of morphological analysis processes are performed on one document data in a division processing unit (multilingual keyword extraction unit). Therefore, the processing speed is considered to decrease. In addition, since only keywords are extracted, full-text search with an arbitrary character string is impossible.
[0016]
Further, in Patent Document 2, since one document data is composed of a plurality of languages, there are few restrictions. However, since document data and an index are stored for each language, it is considered that the processing speed of registration processing and search processing is reduced. .
[0017]
In Patent Document 3, a plurality of document data can be managed and searched as a single document data, but no description about language information is mentioned.
[0018]
In addition, the division processing unit using N character units as a unit has no dependency on language information, but the number of divisions increases and the number of index units increases, so the area required for the full-text index storage unit becomes large. There may be problems such as a decrease in processing speed due to an increase in the amount of data to be processed in the processing and search processing.
[0019]
On the other hand, since the division processing means using morphological analysis divides into word units, the number of index units decreases, but there is a problem that the implementation depends on language information. For example, in the case of a language having a space between words, such as Western languages, the division processing means detects the break of characters such as a space, and a continuous portion of the characters can be used as a word, and the implementation is simple. However, it is difficult to identify the word itself in a language such as Japanese where there is no separation such as a blank in the document. Therefore, the division processing means needs to detect a word using a parsing technique and is difficult to implement. This is because the conventional full-text search apparatus has one language as a target of full-text search and has only one division processing means optimal for the language. In addition, due to the variety of Japanese notation (compound words, katakana different notation), search omissions occurred even when word indexes were created.
[0020]
The present invention has been made in view of the above circumstances, and an object thereof is to provide a full-text search device, a document data processing program, and a recording medium corresponding to document data in a plurality of languages.
[0021]
[Means for Solving the Problems]
In order to achieve this object, the present invention has the following features.
The invention according to claim 1 is a full-text search device for searching document data including a designated character string from a plurality of document data described in different languages, the document data and language information of the document data, Input means for inputting document data, document data storage means for storing document data and language information of the document data, number-of-languages storage means for storing the number of document data for each language information, and the document data A plurality of division processing means for performing word division processing and notation normalization processing of document data linguistic information and dividing the data into index units; a full-text index storage means for storing index information to be used for search regardless of language information; Document data and language information input from the input means are registered in the document data storage means together with document identification information indicating the document data, and the number of document data of the language information is increased. Register in the number-of-languages storage means and divide the document data into index units by using the language information division processing means, and the appearance position information in the document data with different notation after notation normalization processing And registration processing means for registering in the full text index storage means as index information together with the document identification information of the document data.
[0022]
The invention described in claim 2 is the full-text search device according to claim 1, wherein the search character string input using the input means is divided using a plurality of division processing means for each language information, A search expression is generated using the plurality of obtained normalized search character strings, and search processing means for searching from the full-text index storage means using the generated search expression is provided.
[0023]
According to a third aspect of the present invention, in the full-text search device according to the second aspect, the search processing means examines the linguistic information stored in the number-by-language number storage means, and uses only the linguistic information division processing means. A search expression is generated using a plurality of normalized search character strings obtained by division.
[0024]
According to a fourth aspect of the present invention, there is provided input means for inputting document data and language information of the document data, document data storage means for storing document data and language information of the document data, and document data for each language information. A number-by-language number storage means for storing the number of entries, a plurality of division processing means for performing word division processing and notation normalization processing of the language information of the document data on the document data, and dividing into index units; Specified from a plurality of document data described in different languages, and a full-text index storage means for storing index information used in the database regardless of language information and a processing means for registering and retrieving document data In a full-text search apparatus for searching for document data including a character string, a document data processing program to be executed, the document data and language information input from the input means, The document data is registered in the document data storage means together with the document identification information indicating the document data, the number of document data of the language information is increased and registered in the number-by-language number storage means, and the document data is stored using the language information division processing means. A registration process that divides the index unit into index units and registers the index units with different notation after notation normalization processing as index information together with the appearance position information in the document data and the document identification information of the document data in the full-text index storage means It is made to make a means perform.
[0025]
According to a fifth aspect of the present invention, in the document data processing program according to the fourth aspect, the search character string input using the input unit is divided using a plurality of division processing units for each language information, and the division is performed. A search expression is generated using a plurality of normalized search character strings obtained as described above, and the processing means is caused to execute search processing for searching from the full-text index storage means using the generated search expression. .
[0026]
According to a sixth aspect of the present invention, in the document data processing program according to the fifth aspect, the search processing examines the language information stored in the number-by-language number storage means and uses only the language information division processing means. A search expression is generated using a plurality of normalized search character strings obtained by dividing the search.
[0027]
The invention according to claim 7 is characterized in that the document data processing program according to any one of claims 4 to 6 is recorded on a computer-readable recording medium.
[0028]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.
[0029]
First, the configuration of the full-text search apparatus according to the present invention will be described with reference to FIG.
[0030]
The full-text search apparatus according to the present invention includes an input unit 1, an output unit 2, a document data storage unit 3, a number-by-language storage unit 4, a full-text index storage unit 5, a plurality of division processing units 6, and a registration The processing unit 7 and the search processing unit 8 are included. Hereinafter, the function of each part which comprises a full text search device is demonstrated.
[0031]
The input unit 1 is a processing unit that inputs document data and language information of the document data or search conditions.
[0032]
The output unit 2 is a processing unit that outputs the search result searched by the search processing unit 8.
[0033]
The document data storage unit 3 is a storage unit that stores document data and language information of the document data.
[0034]
The number-of-languages storage unit 4 is a storage unit that stores the number of document data for each language information.
[0035]
The full-text index storage unit 5 is a storage unit that stores index information used in search processing regardless of language information.
[0036]
The division processing unit 6 is a processing unit that performs word division processing and notation normalization processing of the language information of the document data, and divides the document data into index units.
[0037]
The registration processing unit 7 registers the document data input from the input unit 1 and its language information in the document data storage unit 3 together with an identifier (document identifier) indicating the document data. In addition, the number of document data of the language is increased and registered in the number-of-languages storage unit 4, and the document data is divided and normalized into index units using a plurality of division processing means 6 for each language, and the normalized data is obtained. This is a processing unit for registering index units having different notations in the full-text index storage unit 5 together with appearance position information in the document data and a document identifier of the document data.
[0038]
The search processing unit 8 divides a search condition input using the input unit 1 using a plurality of division processing units 6 for each language, and uses a plurality of normalized search character strings obtained by the division. A processing unit that generates a search expression and searches the full-text index storage unit 5 using the generated index expression.
[0039]
In the full-text search device, the notation of the index unit is normalized and stored in the full-text index storage unit 5 at the time of registration of document data, and the search processing is performed by normalizing the notation of the search word even during the search process. It becomes possible to obtain a search result that does not depend on the notation of the client. The normalization of the notation performed at this time is also language-dependent processing.
[0040]
In particular, stemming for Western languages is a language-dependent process. Stemming is a search based on the stem of a word. For example, English “walked”, “walker”, and “walking” become “walk” by stemming, and these words can be used as search results simultaneously. It can be done.
[0041]
For example, in English, normalization processing includes singular / plural normalization, inflection normalization, normalization of phonetic symbol notation, and normalization of uppercase and lowercase letters. These are also language-dependent processes. In the case of Japanese, normalization of Chinese characters and variants, normalization of katakana notation (such as “interface”, “interface”, “interface” are the same thing), normalization of symbol notation Processing such as normalization of full-width / half-width (such as parentheses, quotes, and without middle-black) is a normalization process.
[0042]
Next, the hardware configuration of the full-text search apparatus in the present embodiment will be described with reference to FIG.
[0043]
The full-text search device according to the present embodiment includes an input device 20, an output device 21, a main control device 22, a storage device 23, and an input / output control device 24.
[0044]
In the full-text search device having the above configuration, the input device 20 is realized by the input means 1. The output device 21 is realized by the output unit 2. The main control device 22 is realized by various processing means such as a division processing means 6, a registration processing means 7, and a search processing means 8. The storage device 23 corresponds to various storage units including the document data storage unit 3, the number-of-languages storage unit 4, and the full-text index storage unit 5. For example, when a full-text search according to the present invention is performed using one limited storage device, an area to be used is allocated depending on whether the search process is performed mainly or the registration process is performed mainly. This makes it possible to perform processing efficiently. The main controller 22 uses a CPU, a memory, or the like. Further, the input / output control device 24 controls the input device 20 and the output device 21 in accordance with a control signal from the main control device 22.
[0045]
Next, the hardware configuration of the client / server in this embodiment will be described with reference to FIG.
[0046]
The client side 25 includes an input device 20, an output device 21, a main control device 22, a network control device 26, and an input / output control device 24.
The server side 27 has a network control device 26, a main control device 22, and a storage device 23. The client 25 and the server 27 are connected via a network 28.
[0047]
In the hardware configuration of the client / server configured as described above, the input device 20 is realized by the input means 1. The output device 21 is realized by the output means 2. Further, the main control device 22 is realized by the processing means of the division processing means 6, the registration processing means 7, and the search processing means 8. The storage device 23 corresponds to each of the storage units of the document data storage unit 3, the number-of-languages storage unit 4, and the full-text search storage unit 5. The network control device 26 controls data transmission and the like between the client 25 and the server 27 via the network 28. Further, the input / output control device 24 of the client 25 controls the input device 20 and the output device 21 according to the control signal of the main control device 22.
[0048]
Here, the outline of the processing operation in the full-text search apparatus according to the present invention will be described.
First, as a first stage, search data is collected. This search data may be data on a local disk, or may be data collected by running a robot inside the intranet. In some cases, robots run across the Internet.
[0049]
As the second stage, the search data collected in the first stage is passed through the document file and indexed by the indexer.
[0050]
Next, as a third stage, which is an actual operation stage, a search request from the client is searched using the index of the search engine part, and the search result is transmitted to the client.
[0051]
Here, morphological analysis will be described as an example of a division processing unit that is indispensable for creating an index used in the third stage.
[0052]
In documents such as Japanese, there is no space between words and they are not separated, so a basic search index cannot be created for a document as it is. Therefore, using the morphological dictionary and grammatical knowledge about morphemes, the document is “word-written” in units of words, and morphological analysis is used to determine the syntactic role of each word. “Square writing” is a method of writing a space between words or words or between clauses when writing a document.
[0053]
A morpheme is a unit of a word that cannot be further divided, and generally refers to the smallest element that has meaning, and a sentence is composed of this morpheme. For example, by dividing the sentence “search document” into morphemes “document”, “to”, “search”, “to”, each morpheme is given meaning, and syntax analysis and semantic analysis Use for natural language processing such as context understanding.
[0054]
There are three important processes for performing morphological analysis: “separation”, “identification of word parts of speech”, and “processing of words not in the dictionary”. There are various combinations, and various methods on the part of speech occur in the cut out method.
[0055]
The main analysis methods will be described below.
[0056]
The longest match method is a method of scanning a given character string in order from the right and selecting the longest match among the words registered in the dictionary.
[0057]
The character type cutting method is a method of cutting by priority using character type breaks such as delimiters (such as punctuation marks), kanji, katakana, letters, numbers, hiragana.
[0058]
The phrase number minimum method is a method of cutting a document so that the number of phrases is minimized. The phrase means a group of “noun + particle”, “verb”, and the like.
[0059]
The connection rule method uses a connection table that describes whether or not the word B can be connected after the word A. Since the combinations of words are enormous, rules such as part-of-speech connections, character type connections, and connections that form phrases are applied.
[0060]
As described above, in the morphological analysis full-text search, morphemes and semantic information given to the morphemes are used to automatically extract keywords, create an index (search index), or implement a natural language interface.
[0061]
Next, the processing operation in the full-text search apparatus according to the present invention will be described.
[0062]
(Document data registration process)
First, document data registration processing in the full-text search apparatus will be described with reference to FIG.
[0063]
To execute document data registration processing, the client first creates document data (step S1), and registers the created document data and the language information of the document data in the registration processing unit 7 from the input unit 1. (Step S2). Next, the registration processing unit 7 registers the document data input from the input unit 1 and the language information of the document data in the document data storage unit 3, and simultaneously sets an identifier (document identifier) indicating the document data. (Step S3). A conceptual diagram of the document data storage unit 3 is shown in FIG. For example, in FIG. 5, the document data “... search processing means ... according to the search command ...” has the document identifier set to 100 and the language information set to language A. ing. In addition, the document data “... Sie succin das Holz nach dem fehlenden kind...” Has a document identifier 105 and language information B.
[0064]
Next, the registration processing means 7 increases the number of document data in that language and registers it in the number-by-language number storage unit 4 (step S4). In addition, the conceptual diagram of the number storage part 4 according to language is shown in FIG. The number-of-languages storage unit 4 includes language information and the number of document data, and the number of document data is stored for each language information.
[0065]
Next, the registration processing means 7 uses the language information division processing means 6 to divide and normalize the document data into index units (step S5). Then, the registration processing means 7 registers index units having different notations after normalization in the full-text index storage unit 5 together with appearance position information and document identifiers in the document data (step S6).
[0066]
A conceptual diagram of the full-text index storage unit 5 is shown in FIG. The full-text index storage unit 5 includes an index unit and a transposed list {document identifier, appearance count, appearance position}. For example, the transposed list for the index unit: “search”: {100, 2, (20, 60) )}, “100” indicates a document identifier (document ID), “2” indicates the number of appearances, and “(20, 60)” indicates an appearance position.
[0067]
Therefore, the transposed list {100, 2, (20, 60)} for “search” includes information that “search” appears twice in the document 100 and its appearance position is the 20th and 60th characters ”. Show. In addition, the transposed list {100, 3, (260, 280, 320)} for “search” appears “in the document 100,“ search ”appears three times, and its appearance position is the 260th, 280, 320th character. ". Similarly, the transposed list {105, 2, (10, 50)} for “suche” has information that “in the document 100,“ suche ”appears twice and its appearance position is the 10th and 50th characters”. Will be shown.
[0068]
As described above, in the full-text search apparatus, the registration processing unit 7 registers the document data input from the input unit 1 and its language information in the document data storage unit 3 together with an identifier (document identifier) indicating the document data. To do. In addition, the number of document data of the language is increased and registered in the number-of-languages storage unit 4, and the registration processing unit 7 divides the document data into index units by using a plurality of division processing units 6 for each language. The index unit having a different representation after normalization is registered in the full-text index storage unit 5 together with the appearance position information in the document data and the document identifier of the document data, thereby improving the efficiency of the search process. Is possible.
[0069]
(First search processing of document data)
Next, a document data first search process in the full-text search apparatus will be described with reference to FIG.
[0070]
To execute document data search processing, the client first inputs a search character string from the input means 1 to the search processing means 8 (step S10). Then, the search processing unit 8 uses the plurality of division processing units 6 for each language to divide and normalize the search character string input from the input unit 1 into index units (step S11). Next, the search processing means 8 uses a logical sum of a plurality of index units obtained by division as a search condition, and uses a full-text index to collect a set of document identifiers (Rs) of document data including the index unit (search character string). ) Is acquired from the full-text index storage unit 5 (step S12). Then, the search processing means 8 outputs a set of document identifiers (Rs) of document data including the index unit (search character string) acquired from the full-text index storage unit 5 to the client through the output means 2 (step S13).
[0071]
As described above, in the full-text search apparatus, the search processing unit 8 divides the search character string input from the input unit 1 by using the plurality of division processing units 6 for each language, and a plurality of pieces obtained by the division. By generating a search expression using the normalized search character string and searching from the full-text index storage unit 5, search processing can be executed without adding language information of the search character string, thereby reducing the load on the client It becomes possible to do.
[0072]
(Second search process of document data)
Next, the second search processing of document data in the full-text search device will be described with reference to FIG.
[0073]
To execute document data search processing, the client first inputs a search character string from the input means 1 to the search processing means 8 (step S20). Next, the search processing means 8 examines the linguistic information stored in the number-of-languages storage unit 4 and uses the linguistic information division processing means 6 to convert the search character string input by the input means 1 into an index unit. And normalize (step S21). Then, the search processing means 8 uses the obtained logical sum of a plurality of index units as a search condition, and uses the full-text index to set a set (Rs) of document identifiers of document data including the index unit (search character string) as a full-text index. Obtained from the storage unit 5 (step S22). Then, the search processing means 8 outputs a set of document identifiers (Rs) of the document data including the index unit (search character string) acquired from the full-text index storage unit 5 to the client through the output means 2 (step S23).
[0074]
In this way, in the full-text search device, the search processing means 8 examines the linguistic information stored in the language identification number storage unit 4 and inputs it by the input means 1 using only the linguistic information division processing means 6. The search string is divided into index units, and a search expression is generated using a plurality of normalized search strings obtained by dividing the search string, thereby minimizing the number of search string expansions. The processing can be performed efficiently, and the waiting time until the client acquires the search result can be shortened.
[0075]
The above-described embodiment is a preferred embodiment of the present invention, and various modifications can be made without departing from the gist of the present invention. For example, it is possible to construct the full-text search device in the present embodiment by causing the information processing device such as a computer to execute the processing operation in the full-text search device in the above-described embodiment as a program. In addition, the full-text search apparatus according to the present embodiment can be constructed by recording the program on a computer-readable recording medium and mounting the recording medium on an information processing apparatus.
[0076]
【The invention's effect】
As apparent from the above description, the present invention has the following effects.
The full-text search device, the document data processing program, and the recording medium according to the present invention can process document data without adding language information of the document data, thereby reducing the load on the client. it can.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a configuration of a full-text search apparatus according to the present invention.
FIG. 2 is a block diagram showing a hardware configuration in a stand-alone form.
FIG. 3 is a block diagram showing a hardware configuration in a client / server configuration.
FIG. 4 is a flowchart showing a procedure of document data registration processing in the full-text search apparatus according to the present invention.
FIG. 5 is a conceptual diagram of a document data storage unit constituting the full-text search device according to the present invention.
FIG. 6 is a conceptual diagram of the number-of-languages storage unit constituting the full-text search apparatus according to the present invention.
FIG. 7 is a conceptual diagram of a full-text index storage unit constituting the full-text search device according to the present invention.
FIG. 8 is a flowchart showing a procedure of first search processing of document data in the full-text search device according to the present invention.
FIG. 9 is a flowchart showing a procedure of second search processing of document data in the full-text search device according to the present invention.
[Explanation of symbols]
1 Input means
2 Output means
3 Document data storage
4. Number storage section by language
5 Full text index storage
6 Division processing means
7 Registration processing means
8 Search processing means
20 input devices
21 Output device
22 Main controller
23 Storage device
24 I / O control device
25 clients
26 Network controller
27 servers
28 network

Claims (7)

A full-text search device for searching document data including a designated character string from a plurality of document data described in different languages,
Input means for inputting document data and language information of the document data;
Document data storage means for storing document data and language information of the document data;
Number-of-languages storage means for storing the number of document data for each language information;
A plurality of division processing means for performing word division processing and notation normalization processing of language information of the document data and dividing the document data into index units;
Full-text index storage means for storing index information used in the search regardless of language information;
Document data and language information input from the input means are registered in the document data storage means together with document identification information indicating the document data, and the number of document data of the language information is increased to increase the number of document data by language. The document data is divided into index units by using the division processing means for the language information, and the index units having different notations after the notation normalization processing are displayed in the document data on the position information and the document. Registration processing means for registering in the full-text index storage means as index information together with document identification information of data;
A full-text search device characterized by comprising: The search character string input using the input unit is divided using the plurality of division processing units for each of the language information, and a search formula is used using the plurality of normalized search character strings obtained by the division. The full-text search apparatus according to claim 1, further comprising a search processing unit that generates a search from the full-text index storage unit using the generated search expression. The search processing means examines language information stored in the number-of-languages number storage means, and uses a plurality of normalized search character strings obtained by dividing the language information using only the division processing means. The full-text search device according to claim 2, wherein the search formula is generated. Input means for inputting document data and language information of the document data, document data storage means for storing the document data and language information of the document data, and the number of documents by language for storing the number of document data for each language information Storage means, a plurality of division processing means for performing word division processing and notation normalization processing of the language information of the document data for the document data, and dividing the index data into index units; A document including a specified character string from a plurality of document data described in different languages, and a full-text index storage unit that stores the document data and a processing unit that performs document data registration and search processing A document data processing program to be executed in a full-text search device for searching data,
Document data and language information input from the input means are registered in the document data storage means together with document identification information indicating the document data, and the number of document data of the language information is increased to increase the number of document data by language. The document data is divided into index units by using the division processing means for the language information, and the index units having different notations after the notation normalization processing are displayed in the document data on the position information and the document. A document data processing program for causing the processing means to execute registration processing for registering in the full-text index storage means as index information together with document identification information of the data. The search character string input using the input unit is divided using the plurality of division processing units for each of the language information, and a search formula is used using the plurality of normalized search character strings obtained by the division. 5. The document data processing program according to claim 4, wherein the processing unit is caused to execute a search process for searching from the full-text index storage unit using the generated search expression. The search process includes
Using the plurality of normalized search character strings obtained by examining the language information stored in the number-of-languages storage means and dividing only the division processing means of those language information, the search formula is 6. The document data processing program according to claim 5, wherein the document data processing program is generated. 7. A computer-readable recording medium on which the document data processing program according to claim 4 is recorded.

Images