JP2002183144A - System and method for document retrieval and recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a system and a method for document retrieval and a recording medium which improve retrieval precision by expanding a retrieval word by referring to a pattern gathered from an object document to be retrieved without normalizing an index-registered character string as to a different notation having a hyphen. SOLUTION: This system is equipped with a document storage means which stores pieces of document data described in a language whose words have clear breaks, a character string extracting means which extracts a character string sectioned with a specific delimiter character and appearing position information on the character string, etc., from the document data stored in the document storage means, and a normalization information storage means which stores the character string and a normalized character string in a normalization information memory means while making them correspond to each other when the extracted character string matches a specific pattern, and retrieves a document that a user desires from multiple documents.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a document retrieval system, a document retrieval method, and a recording medium for retrieving document data desired by a user from a document database, and more particularly, to a description in a language such as English where words are clearly separated. The present invention relates to a search system for targeted document data.

[0002]

2. Description of the Related Art In a conventional document search system, first, when performing full-text search on document data described in a language in which words are clearly separated, such as English, at the time of index registration, spaces, periods, commas, etc. are used. Each document is divided into character strings in word units using the delimiter character as a boundary, and the extracted character strings are identified by the identifier of the document in which the character string appears and the position of the character string in the document (the number of words from the beginning of the document) Together with the index information. Next, when searching, the document search system extracts a word character string from the query input by the user in text format in the same manner as at the time of index registration, creates a search condition based on the extracted character string, Perform a search.

[0003] In addition to periods and commas, there is a hyphen as one of the symbols treated as a delimiter. The hyphen is used when a plurality of continuous words are described as one unit (compound word). For example, "client-
In "server system" and "phosphate-rich water", two words connected by a hyphen (hereinafter referred to as "hyphen") qualify the following word as a group of words. In such a case, the hyphen is composed Individual words (eg "phosphat
e ") may be a search target, so by registering each index term before and after the hyphen,
Search omissions can be avoided.

[0004]

However, a hyphen may be used when a character string which should be described as one word or a character string which can be described as one word is divided and described. . For example, a line break in the middle of a word (for example, "edu-cation") or a case where a prefix and a stem are separated (for example, "pre-election") correspond to this. When the index is registered with a hyphen as a delimiter, "edu" and "cation" are index words in the former case, so even if the user enters "education" as a query, there is no match. In the latter case, "pre" and "election" are index words, but there is a problem that if the user's inquiry is "preelection", there is no match.

As a method for solving such a problem, there is a method of combining an alphabet character string immediately before a hyphen at the end of a line and an alphabet character string at the head of the next line, deleting the hyphen, and index-registering as one word. However, a hyphen at the end of a line does not always indicate a line break in the middle of a word. For example, if a hyphen happens to appear at the end of a line, such as "phosphate-rich", this method uses "phosp
haterich ".

In such a case, in a system using a large-scale dictionary, such as a machine translation system, a character string combined by end-of-line hyphen processing is looked up in a dictionary, and the character string is adopted only when registered in the dictionary. Is adopted. In the above case, a character string such as "phosphaterich" is not adopted because it is unlikely that it is registered in the dictionary.
cation "can be indexed. However, search systems usually do not use large dictionaries to improve processing efficiency, and such a method is not practical. I can not say.

On the other hand, according to the technique disclosed in Japanese Patent Application Laid-Open No. 7-65013, the fluctuation of the notation in the search target document and the user's inquiry is not unified at the time of index registration, but the expansion of the search word by a different notation is performed at the time of search. Is going.
According to this method, a search term is developed by referring to a different notation dictionary describing different notation candidates at the time of search. However, a line break in the middle of a word covers most English words, and there are multiple possible line breaks in one word (for example, in the case of "education", "ed-u-ca- In addition, it is difficult to construct a dictionary, and there is a problem of a decrease in search efficiency due to an excessive number of expansions of search words. Also, with regard to the continuation of prefixes and stems, there is a limit to solving problems with dictionaries because of the strong new word generation power.

The present invention has been made to solve the above-described problem, and refers to a pattern collected from a document to be searched without normalizing a character string to be indexed with respect to a different notation accompanied by a hyphen. An object of the present invention is to provide a document search system, a document search method, and a recording medium that improve search accuracy by expanding search words.

[0009]

According to a first aspect of the present invention, there is provided a document search system for searching a document desired by a user from a plurality of documents. Document storage means for storing a plurality of document data described in a language with a clear delimiter, and a character string delimited by a predetermined delimiter and the appearance of the character string from the document data stored in the document storage means A character string extracting means for extracting position information and the like, classifying the extracted character string, and converting the document data stored in the document storage means in which the character string appears and the appearance position information of the character string into the type Index storage means for associating with the converted character string and storing it in the index storage means, and when the character string extracted by the character string extraction means matches a specific pattern, A normalization information storage unit for associating a character string with a normalized character string of the character string and storing the normalized character string in the normalization information storage unit, and searches a document desired by the user from a plurality of documents It is characterized by the following.

According to a second aspect of the present invention, in the document search system according to the first aspect, the normalization information storage means stores a hyphen at the end of a line and an alphabetic character string before and after the hyphen at a specific pattern. Then, a hyphen, a line feed code, and a blank are deleted from this specific pattern, and a character string obtained by combining the preceding and following alphabetic character strings is generated as a normalized character string.

According to a third aspect of the present invention, in the document search system according to the first aspect, the normalization information storage unit specifies a prefix accompanied by a hyphen and an alphabet character string following the prefix. And a hyphen is removed from the character string of this specific pattern, and a character string obtained by combining the prefix and the subsequent alphabetic character string is generated as a normalized character string.

According to a fourth aspect of the present invention, in the document search system according to any one of the first to third aspects, a search text for inputting a search text for searching a document is provided. Input means, character string extraction means for extracting a character string delimited by a predetermined delimiter from the search text and appearance position information of the character string, and the like, and refer to the character string stored in the normalization information storage means. Extracting character string expanding means for expanding the extracted character string; and creating a search condition based on the character string extracted by the character string extracting means and the character string expanded by the extracted character string expanding means. And a search processing unit for searching for a document in the document storage unit based on the search condition and the index storage unit. And said that a document that.

According to a fifth aspect of the present invention, in the document search system according to any one of the second to fourth aspects, the extracted character string expanding means includes
If the character string extracted by the character string extraction means is a prefix with a hyphen and an alphabet character string following it, a hyphen is deleted, and a character string obtained by combining the prefix and the subsequent alphabet character string is used. The extracted character string is expanded.

According to a sixth aspect of the present invention, there is provided a document retrieval method for retrieving a document desired by a user from a plurality of documents, the plurality of documents having words separated by a clear language. From the document data stored in the document storage means for storing data, a character string delimited by a predetermined delimiter and information on the appearance position of the character string are extracted, and the extracted character string is categorized and The document data stored in the document storage unit where the column appears and the appearance position information of this character string are associated with the categorized character string and stored in the index storage unit, and the extracted character string is identified. When this pattern matches, this character string is associated with a normalized character string of this character string, and stored in the normalized information storage means.

According to a seventh aspect of the present invention, there is provided a document search method according to the sixth aspect, wherein a hyphen at the end of a line and an alphabetic character string before and after the hyphen are defined as a specific pattern, and the hyphen is converted from the specific pattern. And a line feed code and a space are deleted, and a character string obtained by combining the preceding and following alphabetic character strings is generated as a normalized character string.

The document search method according to claim 8 of the present invention is the document search method according to claim 6, wherein a prefix with a hyphen and an alphabetic character string following the prefix are used as a specific pattern. , A hyphen is removed from the character string, and a character string obtained by combining the prefix and the subsequent alphabetic character string is generated as a normalized character string.

According to a ninth aspect of the present invention, there is provided a document search method according to any one of the first to eighth aspects, wherein a search text for searching a document is input, and A character string separated by a predetermined delimiter and appearance position information of the character string are extracted from the search text, and the extracted character string is referred to by referring to the character string stored in the normalization information storage unit. Expanding, creating a search condition based on the extracted character string and the expanded character string, and searching the document in the document storage means by the search condition and the index storage means, It is characterized in that a document desired by the user is retrieved from the documents.

According to a tenth aspect of the present invention, in the document search method according to any one of the sixth to ninth aspects, a character string extracted from the search text includes a hyphen. In the case of a prefix and an alphabet character string following the prefix, hyphens are deleted, and the extracted character string is expanded by a character string obtained by combining the prefix and the subsequent alphabet character string.

Further, a recording medium according to claim 11 of the present invention is:
A document storage means for storing a plurality of document data written in a language in which words are clearly separated, and a character string delimited by a predetermined delimiter from a document data stored in the document storage means. Character string extracting means for extracting appearance position information and the like, classifying the extracted character string, and storing the document data and the appearance position information of the character string stored in the document storage means where the character string appears. An index storage unit that associates the character string with the categorized character string and stores the character string in the index storage unit; and if the character string extracted by the character string extraction unit matches a specific pattern, the character string and the character string By associating with the normalized string,
A normalization information storage unit for storing the normalization information storage unit in the normalization information storage unit;

Further, the recording medium according to claim 12 of the present invention is:
12. The recording medium according to claim 11, wherein search text input means for inputting a search text for searching for a document, a character string separated from the search text by a predetermined delimiter, and appearance position information of the character string. Character string extracting means for extracting the character string, extracting character string expanding means for expanding the extracted character string with reference to the character string stored in the normalized information storage means, and extracting the character string by the character string extracting means. Search condition creation means for creating a search condition based on a character string and a character string expanded by the extracted character string expansion means, and a document in the document storage means is searched by the search condition and the index storage means. Search processing means.

[0021]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The construction and operation of an embodiment of the present invention will be described below in detail with reference to the drawings.

(1) Configuration of Embodiment of the Present Invention FIGS. 1, 2 and 3 are block diagrams schematically showing the functional configuration of a document search system according to an embodiment of the present invention. The document search system 100 includes a control unit 110, a registration processing unit 120, a search processing unit 130, a document storage unit 140,
It comprises an index storage unit 150, a normalized information storage unit 160, and a prefix table 170. Further, the registration processing unit 120 includes a character string extraction unit 121, an index storage unit 12
2. Normalization information storage means 123. In addition, the search processing unit 130 includes the search text input unit 13.
1, character string extracting means 132, extracted character string expanding means 13
3. It comprises a search condition creating means 134, a search processing means 135, and an output means 136.

The control unit 110 controls the document search system 100
And a search processing unit 130 that searches the registered document for a user's desired document and outputs the result. The registration processing unit 120 stores the document data in the document storage unit 140 and creates the index storage unit 150 and the normalized information storage unit 160. Search processing unit 130
Is based on the search instruction from the user.
Search for 0 and output the result. Document storage means 14
No. 0 stores a large number of document data as a document database in a storage device in association with at least the document data and its identifier. The index storage unit 150 stores a character string included in each document data and document information in association with each other as an index. At this time, the character strings are classified and stored. The document information includes an identifier of the document data including the character string, and an appearance position where the character string appears in the document. When the character string extracted from the document data has a specific pattern, the normalization information storage unit 160 associates the normalized character string of the character string with the original character string. Store. At this time, a character string expanded from the normalized character string is also stored in correspondence with the normalized character string. The prefix table 170 stores a prefix that is referred to when the registration processing unit 120 or the search processing unit 130 processes a prefix word of document data or a search text.

(A) Outline of Registration Processing Unit 120 The registration processing unit 120 reads out document data one by one from the document storage unit 140 and sends the read document data to the character string extraction unit 121. Character string extraction means 121
Processes the transmitted data, extracts a character string serving as an index word, and returns it to the registration processing unit 120 together with the appearance position information. When the character string is extracted, the normalization information storage means 123 is called, and if the extracted character string matches a predetermined specific pattern, a normalized character string for the character string is generated. Then, both the extracted character string and the normalized character string are stored in the normalized information storage unit 160. At this time, if the specific pattern has a prefix, the prefix table 170 is referred to. The registration processing unit 120 sends the result received from the character string extraction unit 121 to the index storage unit 122. The index storage unit 122 classifies the character string extracted by the character string extraction unit 121 and stores the identifier of the document in which the character string appears and the position information of the character string in the index storage in association with the extracted character string. Means 1
Store in 50.

(B) Outline of Search Processing Unit 130 The search processing unit 130 allows a search text input unit 131 to input a text of a search inquiry from a user.
The input text is sent from the search processing unit 130 to the search condition creating unit 134. The search condition creating means 134 sends the text of the inquiry to the character string extracting means 132.
The character string extracting unit 132 processes the sent inquiry text in the same manner as the registration processing unit 120 to extract the character string of the search word, and returns the character string to the search condition creating unit 134. When this character string is extracted, the extracted character string expanding means 133 is called, and the normalization information storage means 160 is referred to. If there is a character string that matches the extracted search word, the normalized character string corresponding to the extracted character string is found. The character string is returned to the search condition creating means 134 together. Further, if the expanded character string for the prefixed word with a hyphen is obtained by referring to the prefix table 170, the expanded character string is also returned to the search condition creating means 134 together. . The search condition creating unit 134 converts the sent search word into an operator format, creates a search condition, and returns the search condition to the search processing unit 130.
The search processing unit 130 calls the search processing unit 135 based on the search conditions. The search processing unit 135 searches the index stored in the index storage unit 150 according to the search condition, and specifies a document satisfying the condition.
The output unit 136 displays information of a document based on the index searched by the search processing unit 135 on a display device such as a display. If necessary, the document storage unit 140 is accessed to output document data.

The following effects can be obtained by adopting such a configuration in the embodiment. A character string that matches a specific pattern is collected and stored from the target document data, so that a search word can be expanded by referring to the character string at the time of search, thereby improving search efficiency. -At the time of search, the end-of-line hyphen pattern that actually appears in the search target document can be referred to, so the search term entered by the user is expanded to a hyphenated character string that exists in the search target document to match the actual You can search. -When searching, it is possible to refer to the hyphenated prefix that actually appears in the search target document, so that the non-hyphenated prefix entered by the user is expanded into the hyphenated prefix and actually immediately You can search.

(2) Operation of the Document Retrieval System This embodiment will be described using a specific example of a document to be registered as shown in FIG. As shown in FIG. 4, document A has "The pre-election group" at the beginning of the document and "cooperatio" in the document.
The document ID is "1", which is English document data that includes a character string "n" and is an identifier that uniquely indicates document A. Document B has "research and edu-
The document ID is an English document data including a character string “cation system” (a line feed exists immediately after “edu-”), and a document ID that is an identifier uniquely indicating the document B is “2”. The document C is English document data including a character string “phosphate-rich water” (a line break exists immediately after “phosphate-”) in the document, and is an identifier that uniquely indicates the document C. The document ID is given “3.” The document D has “the professional education system” in the document.
The document ID is “4”, which is English document data including a character string “m” (a line feed exists immediately after “edica-”) and is an identifier that uniquely indicates the document D.

(A) Operation of Document Registration Process FIG. 5 is a flowchart schematically showing the operation of the document registration process in the registration processing unit 120. First, it is checked whether or not there is any document whose index has not been created in the document data stored in the document storage unit 140 (step S1). If there is no document for which an index (index) has not been created, since the index of all the document data to be searched has been created, the document registration process ends.

On the other hand, if there is a document for which an index has not been created, document data whose index (index) has not been created is read (step S2). A document ID uniquely indicating each document stored in the document storage unit 140 of the document data is acquired (step S3). This document data is sent to the character string extracting means 121 (step S4). When the character string information is returned from the character string extracting means 121 (Y in step S5), the character string information is registered in the index storage means 150 together with the document ID (S6), and the process returns to step S4 to return the character string information. Repeat the extraction process. At the time of registration, character string information is categorized (for example, the same character string is assigned to the same group and sorted by character code) and stored. As a result, the character string information extracted from the four documents in FIG. 4 is stored in the index storage unit 150 as shown in FIG. Character string extraction means 121
If no character string information is returned (N in step S5), if it is an end instruction (Y in step S7), it indicates that the document being processed has ended, so the process returns to step S1 to repeat the registration process. If it is not an end instruction (N in step S7), the process returns to step S4 to repeat the character string extraction process.

FIG. 6 is a flowchart schematically showing the operation when the character string extracting means 121 extracts a character string. First, it is checked whether or not the registration mode has been designated (step S10). If the registration mode has not been specified, processing in the search mode is performed (N in step S10). If the registration mode has been designated, the appearance position is set to zero (initialization) (step S11). Next, a start position is set (step S12). If the start position has reached the end of the document data (Y in step S13), the state of the end instruction is returned (step S14), and the process ends. If the end of the document data has not yet been reached (N in step S13), the character at the start position is a delimiter (step S13).
15 (Y in step S16) and if it is a hyphen (Y in step S16), a hyphen word process described later is performed (step S21).
Then, the process proceeds to step S17. Here, as the delimiters, the symbols shown in FIG. 7 are used in addition to spaces, tabs, and line feeds. If the character at the start position is a delimiter (Y in step S15) and is not a hyphen (N in step S16), the process proceeds to step S17. In step S17, the continuation of the delimiter is skipped from the current start position. After skipping the sequence of delimiters,
Alternatively, if the character at the start position is not a delimiter (N in step S15), a sequence of non-delimiters is extracted (step S18), and the extracted character string and appearance position are returned to the caller (step S19). Here, the character string information extracted as an index registration target from each of the documents A, B, C, and D (see FIG. 4) is extracted, for example, as shown in FIG. Note that all uppercase letters in the document are converted to lowercase letters before registration. Thereafter, the appearance position is advanced by one (step S20), and the process returns to step S12 to repeat the processing.

FIG. 10 is a flowchart schematically showing the operation of the hyphen processing (step S21 in FIG. 6). First, if the character string immediately before the hyphen is not a character string (English character string) composed of only alphabets (step S30)
N), the process ends. If the character string immediately before the hyphen is a character string (Y in S30) and the character string immediately after the hyphen is a line feed (Y in step S31), the head of the next line is not a character string (N in step S32), and the process ends. I do. If the beginning of the next line is an alphabetic character string (Y in step S32), the character strings before and after the hyphen are concatenated by removing the hyphen and the line feed (step S33), and the character string with the hyphen ( The line feed is to be deleted) and stored in the normalized character string storage unit 160 (step S34).
The process ends. At this time, an expandable pattern is generated from the normalized character string, and stored together with the normalized character string.

If the character string immediately before the hyphen is a character string and the character string immediately after the hyphen is not a line feed (N in step S31), the character string immediately before the hyphen is referred to the prefix table 170 (see, for example, FIG. 11). (Y in step S35) and if the character string immediately after is a character string (Y in step S36), the character strings before and after the hyphen are concatenated by removing the hyphen (step S33), and the concatenated character string with a hyphen The corresponding character strings are stored in the normalized character string storage unit 160 (step S34), and the process is terminated. If the character string immediately before the hyphen is not a prefix (N in S35), or if the character string immediately after the hyphen is not a character string (N in S36), the process ends.

FIG. 12 shows a normalized character string extracted from the four documents shown in FIG. 4 by hyphenation processing (see FIG. 10) and a character string based on the normalized character string. 3 shows the contents of the activation information storage means 160. For example, from document A, the hyphenated prefix "pre-election"
Is extracted, and an expansion pattern using the normalized character string "preelection" as a key is stored. From the document B, a character string “education” around the hyphen at the end of the line is extracted, and an expansion pattern using the normalized character string “education” as a key is stored. From document C, the string "pho" around the hyphen at the end of the line
sphate-rich "is extracted and its normalized string" phosphate
The expansion pattern with "rich" as a key is stored. Document D
Extracts the character string "educa-tion" before and after the hyphen at the end of the line, and stores an expansion pattern using the normalized character string "education" as a key.

(B) Operation of Search Processing Unit FIG. 13 is a flowchart schematically showing the operation of the document search processing in the search processing unit 130. First, the text input from the search text input unit 131 is sent to the search condition creation unit 134. The search condition creation means 134
If the sent instruction is an end instruction (Y in step S40), the process ends. If it is not an end instruction (N in step S40), if a search text is input (Y in step S41), the search text is input to the character string extracting unit 132.
(Step S42). When the character string information is returned from the character string extraction unit 132 (Y in step S43), the returned character string is not a hyphen (step S47).
N), and is converted into an OR operation (step S48). Also,
If the returned character string is a hyphen (step S47)
Y), after being converted to an adjacent operation (step S49),
Further, the operation is converted to an OR operation (step S48), and the process returns to step S42 to repeat the processing.

When the character string extraction process for the search text is completed (step S44), the search process is performed according to the created search condition (step S45), the search result is output (step S46), and an instruction from the next user is issued. Return to step S40 to receive.

FIG. 14 is a flowchart schematically showing the operation of the character string extracting means 132. First, it is checked whether or not the registration mode has been designated, and if it has been designated (Y in step S60), the processing of the registration mode is performed. If the registration mode is not specified (N in step S60), the start position is set (step S61). If the end of the search text is reached (Y in step S62), the state of the end instruction is returned (step S62). S63), the process ends. If the end of the search text has not been reached (step S6
2 N), the start position is a delimiter (step S6)
If Y is not a hyphen (N in step S65), the continuation of delimiters is skipped from the start position (step S66), and the process proceeds to step S67. If the start position is not a delimiter (N in step S64), the process proceeds to step S67. In step S67, a sequence of non-separable characters is extracted, and the extracted character string is returned (step S67).
68). Further, the extracted character string is stored in the normalized information storage unit 16.
If it is stored as a normalized character string of 0 (step S
69, Y), the development pattern corresponding to the normalized character string is returned (step S70), and the process returns to step S61 to repeat the processing. If it is not stored as a normalized character string (N in step S69), the process returns to step S61 to repeat the processing.

If the character at the start position is a delimiter (Y in step S64) and is a hyphen (step S64).
65, Y), a hyphen expansion process described later (step S7)
Perform 1). When the hyphen expansion processing returns a result, a character string of the processing result is returned (step S7).
3) Return to step S61 and repeat the process. At this time, the start position is advanced to the end position of the hyphen processing result. If the hyphen expansion does not return a result, the process proceeds to step S66.

FIG. 15 is a flowchart schematically showing the operation of the hyphen expansion process (step S71 in FIG. 14). First, referring to the prefix table 170, if the immediately preceding alphabetic character string is a prefix (Y in step S80) and the character string immediately afterward is (Y in step S81), the hyphen is deleted and the preceding and succeeding character strings are replaced. Returns the concatenated character string (step S
82) Further, a character string with a hyphen is returned (step S83), and the process ends. If the immediately preceding alphabetic string is not a prefix (N in step S80) or if the immediately following alphabetic string is not an alphabetic string (step S81), the process is immediately terminated because it is not the subject.

(3) Example of Creating Search Conditions A description will be given, with reference to FIGS. 16 and 17, of creating search condition expressions from search texts for the four documents described in FIG. FIG. 16 is an example of a search text for the document storage unit 140.

In query 1, "education sys
The text "tem" is input.
First, "education" is extracted. When this character string is used as a key to find the normalization information storage means 160 (see FIG. 10), the expansion pattern of "education" is "edu".
-cation "," educa-tion "found," education "," edu
-cation "," educa-tion "
The search condition creating means 134 converts the returned three character strings into an OR operation (here, in the form of #OR ()), but two of "edu-cation" and "educa-tion" Since each string is a hyphenated string, each of them is an adjacent operation (here, #NEXT ()
) And convert it into an OR operation term. Further, the character string extracting means returns "system" from the remaining character strings of the search text, and the search condition creating means adds this to the OR operation, and finally, the search condition 1 shown in FIG. 17 is created. When the document storage means 140 is searched using this search condition expression, the document I
D2 and D4 are identified, and document B and document D are output as search results.

In query 2, "preelection c
ampaign "is input. Character string extraction means 1
At step 32, "preelectin" is first extracted. Using this character string as a key, the normalized information storage unit 160 (FIG. 10)
Search), "pre-election" is found as an expansion pattern of "preelection", and "preelection", "pre-ele
ction ". The search condition creating means 134 converts the two returned character strings into an OR operation (here, in the form of #OR ()). -electio
Since "n" is a character string with a hyphen, it is converted to an adjacent operation (here, in the form of #NEXT ()) and then used as an OR operation term. The extraction means returns "campaign", and the search condition creation means adds the result to the OR operation to finally create search condition 2 shown in Fig. 17. Using this search condition expression, the document storage means 140 is used. Is searched, document ID1 is identified as a document that matches search condition 2, and document A is output as a search result.

In the inquiry 3, "co-operation"
Is entered. "Co-operation" is extracted by the character string extracting means 132, but because of the hyphen, it is subjected to hyphen expansion processing. In the hyphen expansion process, referring to the prefix table 170 (see FIG. 11), it is found that the alphabetic character string “co” immediately before the hyphen is a prefix, and “cooperation” in which the character strings before and after the hyphen are connected.
And "co-operation" which is a character string with a hyphen.
The search condition creating means 134 converts the returned character string into an OR operation. However, since the character string is a "co-operation" hyphenated character string, it is converted into an adjacent operation, and then converted into an OR operation item. Search condition 3 (see FIG. 17) is created. When the document storage means 140 is searched using this search condition expression,
Document ID 1 is identified as a document that matches search condition 3, and document A is output as a search result.

(4) Computer Example of the Present Invention Further, the present invention is not limited to only the above embodiment. FIG. 18 is a block diagram showing a hardware configuration when the document search system of the present invention is realized by a computer. That is, input device 1, output device 2, C
PU (Central Processing Uni)
t; central processing unit) 3, memory 4, storage device 5, and medium drive device 6 are connected by a bus 8.

The input device 1 includes a keyboard, a mouse, a digital still camera, a scanner, and the like, and is used for inputting information. The output device 2 includes a display and a printer device for outputting various output information, information input from the input device 1, and the like. The CPU 3 operates various programs. The memory 4 holds the program itself, and also holds information temporarily created when the program is executed by the CPU 3.
The storage device 5 holds data, programs, temporary information at the time of program execution, and the like. The medium drive device 6 is used to mount a recording medium storing programs, data, and the like, read them, and store them in the memory 4 or the storage device 5. Further, it may be used for directly inputting / outputting data or executing a program.

In such a computer device, each function (see FIG. 1, FIG. 2, and FIG. 3) constituting the document search system is programmed respectively, and a CD-ROM is previously prepared.
The CD-ROM is written in a recording medium such as a CD-ROM drive at each site.
The program is stored in the memory 4 or the storage device 5 by installing the program in the computer equipped with the program, and the functions of the document search system can be realized by executing the program. At this time, the document storage unit 140, the index storage unit 150, the normalization information storage unit 160, the affix table 170, and the like are stored in the storage unit 5, and the search text, the instruction to the document search system, etc. And a mouse are used. For output, the display of the output device 2 is mainly used for input and output of intermediate results, and for the final result, the printer of the output device 2 and the disk of the storage device 5 are used.

As a recording medium, a semiconductor medium (for example, ROM, IC memory card, etc.), an optical medium (for example, ROM,
Any of a DVD-ROM, MO, MD, CD-R, etc.) and a magnetic medium (eg, magnetic tape, flexible disk, etc.) may be used.

By executing the loaded program, not only the functions of the above-described embodiment are realized, but also the operating system or the like performs part or all of the actual processing based on the instructions of the program. The case where the function of the above-described embodiment is realized by the processing is also included.

(5) Operation on Network In the above-described embodiment, the document search system is a stand-alone environment. However, the present invention is not limited to this, and the terminal 200 and the document search server are used as shown in FIG. 210 may be used in a network environment. In this case, the terminal 200 or the document search server 21
The hardware device 0 is connected to the network 9 by adding the network connection device 7 to a general-purpose computer device (see FIG. 18). Further, a plurality of terminals 200 and document search servers 210 may be provided as necessary.

This network 9 is generally realized by a cable, and TCP / IP is used as a communication protocol. However, the transmission path is not limited to the cable, and may be any of wireless, wired and broadcast waves as long as the communication protocol between them is the same.
N (Local Area Network), WAN
(Wide Area Network), Internet, analog telephone network, digital telephone network (ISDN: I
ntegral Service Digital N
network, a PHS (Personal Handy System), a mobile phone network, a satellite communication network, or the like.

In such a network environment, the function of the document search system of the above-described embodiment is provided in the document search server 210, and an input of search conditions and the like from the input device 1 of the terminal 200 is transmitted to the document search server 210. The search result is returned to the requesting terminal 200, and the result is received, and the terminal 200 is configured to display the search result on the output device 2 such as a display. Further, the search result may be transmitted to another terminal 200 instead of the request source.

(6) Distribution of Medium A program for realizing the functions of the document search system of the present invention can be distributed in the form of a medium such as a CD-ROM. The program to be realized may be stored in a storage device such as a magnetic disk or the like, and may be provided by being distributed in a form such as download over a wired or wireless network, or distributed by broadcast waves.

[0052]

As described above, according to the present invention,
Search precision was improved by referencing patterns collected from documents to be searched for, without normalizing the character strings to be indexed, with regard to different notations with hyphens. .

[Brief description of the drawings]

FIG. 1 is a block diagram schematically showing an overall functional configuration of a document search system according to an embodiment of the present invention.

FIG. 2 is a block diagram schematically showing a functional configuration of a registration processing unit of the document search system.

FIG. 3 is a block diagram schematically showing a functional configuration of a search processing unit of the document search system.

FIG. 4 is a specific example of a registration target document.

FIG. 5 is a flowchart schematically showing an operation of a registration processing unit.

FIG. 6 is a flowchart schematically showing an operation of a character string extracting unit at the time of registration.

FIG. 7 is an example of symbols used as delimiters.

FIG. 8 is an example of character string information extracted by a character string extracting unit.

FIG. 9 is an example of contents stored in an index storage unit.

FIG. 10 is a flowchart schematically showing an operation of hyphen processing at the time of document registration.

FIG. 11 is an example of the contents of a prefix table.

FIG. 12 is an example of a normalized character string extracted by hyphen processing and its expansion pattern.

FIG. 13 is a flowchart schematically showing an operation of a search condition creating means.

FIG. 14 is a flowchart schematically showing an operation of a character string extracting unit at the time of a search.

FIG. 15 is a flowchart schematically showing an operation of a hyphen expansion process when a search condition is created.

FIG. 16 is an example of a search text for search.

FIG. 17 is an example of search conditions created by the search condition creation unit 13 for each inquiry text.

FIG. 18 is a diagram illustrating a hardware configuration of a computer device that implements the present invention.

FIG. 19 is a diagram for explaining an operation mode according to the network environment of the present invention.

[Explanation of symbols]

 REFERENCE SIGNS LIST 100 document search system 110 control unit 120 registration processing unit 121 color gamut selection unit 122 color gamut setting unit 123 representative color data storage unit 130 search processing unit 131 search text input unit 132 character string extraction unit 133 extracted character expansion unit 134 search Condition creation means 135 Search processing means 136 Output means 140 Document storage means 150 Index storage means 160 Normalization information storage means 170 Affix table 1 Input device 2 Output device 3 CPU 4 Memory 5 Storage device 6 Medium drive device 7 Network connection device 8 Bus 9 network 200 terminal 210 document search server

Claims (12)

[Claims] 1. A document retrieval system for retrieving a document desired by a user from a plurality of documents, a document storage means for storing a plurality of document data described in a language in which words are clearly separated, and said document storage. Character string extracting means for extracting a character string delimited by a predetermined delimiter and appearance position information of the character string from the document data stored in the means, and classifying the extracted character string, Index storage means for associating the document data stored in the document storage means in which the character string appears and the appearance position information of this character string with the categorized character string and storing it in the index storage means; If the character string extracted by the means matches a specific pattern, the character string is associated with the normalized character string of this character string, and stored in the normalization information storage means. A document retrieval system comprising: a normalization information storage unit for retrieving a document desired by the user from a plurality of documents. 2. The document search system according to claim 1, wherein the normalization information storage means uses a hyphen at the end of a line and an alphabetic character string before and after the hyphen as a specific pattern,
A document search system characterized in that a hyphen, a line feed code, and a space are removed from the specific pattern, and a character string obtained by combining preceding and succeeding alphabetic character strings is generated as a normalized character string. 3. The document retrieval system according to claim 1, wherein the normalization information storage means sets a prefix including a hyphen and a subsequent alphabetic character string as a specific pattern, A document search system characterized in that a hyphen is removed and a character string obtained by combining a prefix and a subsequent alphabetic character string is generated as a normalized character string. 4. The document search system according to claim 1, wherein a search text input means for inputting a search text for searching for a document, and a predetermined delimiter from the search text. Character string extracting means for extracting a character string delimited by the character string and appearance position information of the character string and the like, and expanding the extracted character string with reference to the character string stored in the normalization information storage means. Extraction string expansion means, search condition creation means for creating a search condition based on the character string extracted by the character string extraction means and the character string expanded by the extraction character string expansion means, Search processing means for searching for a document in the document storage means by the index storage means, and searching for a document desired by the user from the plurality of documents. Stem. 5. The document search system according to claim 2, wherein the extracted character string expanding unit includes a character string extracted by the character string extracting unit, the character string being prefixed with a hyphen. And a character string following the character string, in which the hyphen is deleted, and the extracted character string is expanded by a character string obtained by combining the prefix and the subsequent alphabetical character string. 6. A document retrieval method for retrieving a document desired by a user from a plurality of documents, wherein a document stored in document storage means for storing a plurality of document data described in a language in which words are clearly separated. From the data, a character string separated by a predetermined delimiter and appearance position information of the character string and the like are extracted, and the extracted character string is categorized and stored in the document storage unit where the character string appears. The document data and the appearance position information of this character string are associated with the categorized character string and stored in the index storage unit. If the extracted character string matches a specific pattern, this character string and this character string A document search method, wherein a character string is associated with a normalized character string and stored in a normalized information storage unit. 7. The document search method according to claim 6, wherein a hyphen at the end of a line and an alphabet character string before and after the hyphen are a specific pattern, and a hyphen, a line feed code, and a space are deleted from the specific pattern, A document search method characterized by generating a character string obtained by combining alphabetic character strings as a normalized character string. 8. The document search method according to claim 6, wherein a prefix accompanied by a hyphen and an alphabet character string following the prefix are used as a specific pattern, and the prefix is obtained by deleting the hyphen from the character string of the specific pattern. A document search method comprising: generating a character string obtained by combining a character string following the character string and a subsequent character string as a normalized character string. 9. A document search method according to claim 1, wherein a search text for searching for a document is input, and a character delimited by a predetermined delimiter from the search text is input. Extracting the string and the appearance position information of the character string, expanding the extracted character string with reference to the character string stored in the normalization information storage means, and extracting the extracted character string and the expanded character string. Create a search condition based on the text string
A document search method, wherein a document desired by a user is searched from a plurality of documents by searching the document in the document storage means using the search condition and the index storage means. 10. The document search method according to claim 6, wherein the character string extracted from the search text is a prefix accompanied by a hyphen and an alphabet character string following the prefix. Is a method of extracting an extracted character string by a character string obtained by combining a prefix and a subsequent alphabetic character string by removing a hyphen. 11. A document storage means for storing a plurality of document data described in a language in which words are clearly separated,
From the document data stored in the document storage unit, a character string extraction unit that extracts a character string delimited by a predetermined delimiter and appearance position information of the character string, and classifies the extracted character string, Index storage means for associating document data stored in the document storage means in which the character string appears and appearance position information of the character string with the categorized character string and storing the document data in the index storage means; If the character string extracted by the character string extracting means matches a specific pattern, the character string is associated with a normalized character string of the character string, and the normalized information stored in the normalized information storing means is stored. Storage means for reading a computer-readable recording program for causing a computer to function as a document search system for searching for a document desired by the user from among a plurality of documents Capacity recording medium. 12. The recording medium according to claim 11, wherein search text input means for inputting a search text for searching for a document, a character string separated from the search text by a predetermined delimiter, and the character string Character string extracting means for extracting appearance position information of the character string; extracting character string developing means for expanding the extracted character string by referring to the character string stored in the normalization information storage means; Search condition creating means for creating a search condition based on the character string extracted by the extracting means and the character string expanded by the extracted character string expanding means; and And a search processing means for searching for the document.