JP2004199282A - Document retrieval device and documents registeration device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To execute efficient document retrieval by excluding any useless input/output data in retrieval processing when discriminating different notation characters or not discriminating them. <P>SOLUTION: This device is provided with a character string conversion part for converting respective characters in an inputted retrieval keyword into character codes exclusively for one character or character codes assigned to the set of different notation characters based on different notation conditions to designate whether to identify or discriminate the different notation characters, a position information reading part for reading the occurrence position information of the respective character codes in a document to be retrieved by referring to a retrieval database, a collating part for collating the occurrence position information of the respective character codes with the character code column and a collation result outputting part for outputting the occurrence position information of the retrieval keyword based on the collation result. <P>COPYRIGHT: (C)2004,JPO&NCIPI

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a document search device and a document registration device.
[0002]
[Prior art]
2. Description of the Related Art In recent years, a large amount of electronic documents have been accumulated, and a method for efficiently searching a desired document from a document database has been demanded.
[0003]
In a general document search device, when a document database is created, an index for searching for the appearance position information of each character in a document to be registered by a character code or the like is created and stored in a storage device or the like. At the time of search, the appearance position of the search keyword can be searched at high speed by referring to the index.
[0004]
On the other hand, in the document search, for example, when it is desired to search for a document including the keyword “whiskey”, it may be more appropriate to search for a document including the phrase “whiskey” or “whiskey” for the purpose of the user. If so, it may be better not to include it. Other examples of characters that can be regarded as the same depending on the example (hereinafter, referred to as “different notation characters”) such as “I / i / ヰ” include “side / side” in “Watanabe” and “Watanabe”. and so on.
[0005]
In a conventional document search method, a character string having two or more different notations, such as “whiskey”, “whiskey”, and “whiskey”, is fixedly distinguished from the different notations, or is searched. Don't search. In the search method in which the document is fixedly distinguished, a document including a word with a different notation is not to be searched, so that a user who also needs a document including a different notation of an input search keyword is omitted from the search. On the other hand, in the method of searching without distinguishing, all documents containing words with different notations are also searched, so that there is a problem that a user who wants to distinguish different notations can obtain unnecessary documents.
[0006]
In order to solve this problem, there is a method in which a user can select whether to distinguish different expressions of a search keyword during a search. For example, in the conventional document search device disclosed in Patent Document 1, in a document database index, different notation characters such as full-width and half-width characters and uppercase and lowercase English characters are treated as common characters. The index holds character type information for distinguishing between different characters such as uppercase and lowercase for each character, together with location information indicating the appearance position of each character in the document database. At the time of search, the search keywords are converted to common characters used in the index. When performing a search that does not distinguish between different notations, character position information is not referred to, and the appearance position information of the same common character as the search keyword converted to the common character is acquired from the index. On the other hand, when a search for distinguishing different notations is specified, not only the common character is searched, but also the character type information is referred to, and the appearance position information is acquired only when the character types match.
[0007]
[Patent Document 1]
JP-A-08-77188 (FIGS. 1 and 2)
[0008]
[Problems to be solved by the invention]
As described above, in the document search apparatus, the user can select whether to search for differently written characters or to search without distinguishing them according to the purpose of the user, so that the accuracy of the search is improved and the usability is good.
[0009]
However, in the above-described document search device, since characters in a keyword are converted into a common character at the time of search, even when a search for distinguishing different notations is performed, an index of unnecessary character type information is read out. There is a problem that data input / output occurs.
[0010]
Further, in the above-described document search apparatus, a process of referring to character type information is required when performing a search while distinguishing different notations by common characterization of the search keyword, and the number of steps of the search process increases, thereby reducing the CPU load. Increase.
[0011]
Further, in the above-described document search apparatus, the document index includes character type information in addition to the appearance position information, so that the data input / output amount when referring to the index at the time of document registration and search increases.
[0012]
The present invention solves the above problem by preventing unnecessary data input / output and performing processing steps in both search processing for distinguishing and not distinguishing differently written characters depending on the application. It is an object of the present invention to obtain a document search apparatus that performs an efficient document search with as little as possible.
[0013]
Another object of the present invention is to provide a document registration device for registering a search target document in a format that allows efficient search.
[0014]
[Means for Solving the Problems]
A document search device according to the present invention provides a different notation distinguishing condition that specifies one of a process of matching one character and a different notation character of the character as the same character and a process of matching these characters as another character. Based on the above, each character in the input search keyword is referred to a character code storage unit that holds a character code dedicated to one character or a character code assigned to a set of candidate characters of the different notation of the character. Reads the occurrence position information of each character code in the character code string from the character string conversion unit that converts the character code and generates the character code string, and compares the character code string with the read occurrence position information from the search target document database And a collation result output unit for outputting a collation result by the collation unit.
[0015]
The document registration device according to the present invention refers to a character code storage unit that holds a character code dedicated to one character or a character code assigned to a set of candidate characters of the different character of the character, and reads each character in the input document. A character code allocating unit that extracts a character code dedicated to the character for the character, and extracts a character code assigned to the set of the different notation character candidates when there is a different notation character candidate set including each character, A character code appearance position index generation unit for generating appearance position information in a document of all character codes extracted for each character in a document, and an index generated by the character code appearance position index generation unit in a storage device Is provided with an index storage unit for storing data in the index storage unit.
[0016]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments of the present invention will be described.
Embodiment 1 FIG.
FIG. 1 is a block diagram showing a configuration of a system including a document search device and a document registration device according to Embodiment 1 of the present invention.
When registering the search target document, the user inputs the document via the input processing unit 24 of the document registration device 20. The document registration device 20 includes an internal character code allocating unit 21 (character code allocating unit), an appearance position information creating unit 22 (character code appearance position index creating unit), and an index creating unit 23 (character code appearance position index creating unit, index storage). Section) processes the document and generates an index for searching. The generated index is stored in the index database 30 (search target document database). Further, the input document is also stored in a storage device or the like (not shown). The internal character code allocating unit 21, appearance position information creating unit 22, index creating unit 23, and input processing unit 24 represent modules of operations performed by a central processing unit of a computer according to a program. Constitutes a central processing unit integrally.
[0017]
At the time of document search, the user inputs a search keyword via the input processing unit 15 of the document search device 10. The document search device 10 performs a search process by the character string conversion unit 11, the position information reading unit 12 (collation unit), the collation unit 13, and the collation result output unit 14 based on the input search keyword. Then, a document containing the input search keyword is searched, and a search result is output. The character string conversion unit 11, the position information reading unit 12, the collation unit 13, the collation result output unit 14, and the input processing unit 15 represent the modules of the operations performed by the central processing unit of the computer according to the program. Actually constitutes a central processing unit as a single unit.
[0018]
The internal character code table 40 (character code storage unit) is realized by, for example, a relational database. Characters are used as keys and character codes dedicated to the characters are associated with each other. Further, when there is a set of different notation character candidates including the character, an internal character code assigned to the different notation character candidate set is also associated. As shown in FIG. 2, the internal character codes registered in the internal character code table 40 include an internal character code uniquely assigned to one character, and “I / Y / ヰ” and “side / There is an internal character code assigned to a set of different notation characters such as "side".
The user can arbitrarily register the set of differently written characters using an input means (a personal computer or the like) of a management device (not shown) of the internal character code table 40.
[0019]
The document search device 10 and the document registration device 20 may be different server devices or the like, or may be one device. Further, the index database 30 and the internal character code table 40 may be stored in the same storage device or may be stored in separate devices, and are connected to the document search device 10 and the document registration device 20 via a line. Connected. Alternatively, it may be stored in the same device as the document search device 10 or the document registration device 20.
[0020]
Next, the document registration process will be described. FIG. 3 is a flowchart of a document registration process executed by the document registration device 20 according to the first embodiment of the present invention. First, the input processing unit 24 receives an input of a registered document (step ST201). Here, it is assumed that the registration document 51 shown in FIG. 4 has been input.
[0021]
Next, the internal character code assigning unit 21 refers to the internal character code table 40 and assigns an internal character code to each character in the input registered document. That is, a character code dedicated to the character associated with each character is extracted, and if an internal character code of a set of different notation characters including the character is registered, the character code is also extracted (step ST202). For example, in the example of FIG. 4, the internal character code "A2" for "U" in the registered document 51, the character codes "A52" and "A101" for "i", and Extracts "A1" and "A101".
[0022]
Next, the appearance position information creating unit 22 creates appearance position information in the registered document 51 for each internal character code assigned to each character of the registered document 51 in step ST202 (step ST203). For example, for “A1” representing “A”, “7” is acquired because “A” appears in the seventh character of the registered document 51. Similarly, “2” is acquired for “A52” representing “i”. As for “A101” representing “I / i / ヰ”, “2” and “7” are acquired because “i” of the second character and “i” of the seventh character are targets.
[0023]
Next, the index creating unit 23 creates an index of the registered document 51 and stores it in the index database 30 (step ST204). FIG. 4 shows a part of the contents stored in the index database 30. The index uses the internal character code obtained in step ST202 as a heading, the character corresponding to the internal character code, and the appearance position information (registered document number-position in document) in the registered document obtained in step ST203. Number) is registered.
[0024]
Next, the document search processing will be described. FIG. 5 is a flowchart of a document search process executed by the document search device 10 according to the first embodiment of the present invention. First, the input processing unit 15 receives an input of a search keyword and a different notation condition (different notation discrimination condition) from the user (step ST501).
[0025]
Here, a case where the search keyword is “whiskey” will be described as an example. The different notation condition is a condition that the user specifies for each character in the search keyword, whether the corresponding character is distinguished in different notation, or not distinguished (the different notation character is regarded as the same). For each character, “0” is specified when different notation is distinguished, and “1” is specified when no distinction is made. For example, if "U", "S", "K", and "-" of the search keyword "Whiskey" are distinguished differently and "I" is not distinguished, the different notation condition is "01000". . As a result, "i" is regarded as the same word even if "i" and "@" are used, and "whiskey" and "whiskey" are also searched.
[0026]
Character string conversion unit 11 reads the input search keyword one character at a time (step ST502).
[0027]
Next, the character string converter 11 refers to the input different notation condition for the read character and determines whether or not to distinguish the different notation (step ST503).
[0028]
For the character determined to be distinguished in the different notation in step ST503, the character string conversion unit 11 refers to the internal character code table 40 and converts the character into an internal character code dedicated to the character (step ST504).
[0029]
On the other hand, for the character determined not to be distinguished in the different notation in step ST503, the character string conversion unit 11 refers to the internal character code table 40 and sets the character to the internal character code assigned to the set of the different notation characters. Conversion is performed (step ST505).
[0030]
Character string converter 11 determines whether all characters of the search keyword have been read (step ST506). Until the reading of all the characters is completed, the processing from step ST502 to step ST506 is repeated. When the reading is completed, the obtained conversion results are linked to generate a character code string (step ST507). The character string conversion processing of the search keyword will be described with a specific example. Since “U” of the keyword “whiskey” is distinguished in a different notation based on a different notation condition, it is converted to “A2” from FIG. On the other hand, “i” which does not distinguish between different notations is converted into “A101” representing “i / i / ヰ”. Finally, a character code string “A2A101A12A6A90” is obtained.
[0031]
Next, the position information reading unit 12 reads index data as shown in FIG. The position information reading unit 12 acquires the appearance position information of the corresponding internal character code from the index using each character code in the character code string obtained in step ST507 as a key (step ST508). For the character code string “A2A101A12A6A90”, the appearance position information of each internal character code as shown in FIG. 6 is obtained.
[0032]
Next, collating section 13 performs collation between the appearance position information acquired in step ST505 and the character code string acquired in step ST507 (step ST509). Specifically, the appearance position information of each character code is checked to determine whether or not there is a line that matches the character code string. In the example of FIG. 6, the arrangement of “1-1, 1-2, 1-3, 1-4, 1-5” corresponds.
[0033]
As a result of the collation, if there is continuous appearance position information having the same length as the character string code of the search keyword, collation result output unit 14 outputs the appearance position information and ends (step ST510). In the example of FIG. 6, "1-1" which is the head number of the part corresponding to the search keyword is output.
[0034]
On the other hand, when there is no matching arrangement, the matching result output unit 14 returns “no corresponding keyword” (step ST511).
[0035]
Other combinations of different notation characters to be registered in the internal character code table 40 include a combination of uppercase and lowercase letters of the alphabet, a set of full-width and half-width letters of the alphabet, the uppercase of hiragana and the uppercase of vowels, or the lowercase of vowels. Pair, modern hiragana and historical kana, pair of katakana long and uppercase or vowel lowercase, modern katakana and historical kana, full-width and half-width katakana, Arabic numerals A set of numbers and Chinese numerals, a set of full-width and half-width Arabic numerals, a set of regular and abbreviated Chinese characters, or a set of folk characters may be registered. Alternatively, a combination of two or more of these may be registered.
[0036]
As described above, according to the first embodiment, the character string conversion unit 11 of the document search device 10 converts the search keyword into a character code dedicated to one character or a set of different notation characters based on the different notation condition. Since the character codes are converted to the assigned character codes and collated with the index database 30, in both of the search processing for distinguishing and not distinguishing differently written characters, useless data input / output is prevented, and There is an effect that a document search can be performed in the processing step.
[0037]
Further, according to the first embodiment, the different notation condition is specified by the user when inputting the search keyword, so that the search condition can be specified according to the user's purpose of use.
[0038]
Further, according to the first embodiment, the internal character code allocating unit 21 of the document registration device 20 allocates each character included in the registered document to a set of a character code dedicated to one character and a different notation character. Since the generated character codes are assigned to each of them and an index is generated together with the appearance position information, the document can be registered in a format that allows efficient search.
[0039]
【The invention's effect】
As described above, according to the present invention, each character in an input search keyword is assigned to a character code dedicated to that character or a set of candidates for a different notation character of the character based on the different notation distinguishing condition. It is converted to a character code and collated to determine whether it exists in the search target database. This has the effect of preventing input / output and reducing the number of processing steps as much as possible, so that a document search apparatus for performing efficient document search can be obtained.
[0040]
According to the present invention, a character code dedicated to the character is assigned to each character in the document to be registered, and when there is a set of different notation character candidates including each character, the character code is assigned to the different notation character candidate set. Since the character code is assigned to the character and an index is created together with the appearance position information in the document, there is an effect that a document registration device that registers a document in a format that allows efficient search can be obtained.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a configuration of a system including a document search device and a document registration device according to a first embodiment of the present invention.
FIG. 2 is a diagram showing a part of contents stored in an internal character code table according to the first embodiment of the present invention.
FIG. 3 is a flowchart of a document registration process executed by the document registration device according to the first embodiment of the present invention.
FIG. 4 is a diagram showing a part of contents stored in an index database according to the first embodiment of the present invention.
FIG. 5 is a flowchart of a document search process executed by the document search device according to the first embodiment of the present invention.
FIG. 6 is a diagram showing output contents of a position information reading unit according to the first embodiment of the present invention.
[Explanation of symbols]
Reference Signs List 10 document retrieval device, 11 character string conversion unit, 12 position information readout unit (collation unit), 13 collation unit, 14 collation result output unit, 15 input processing unit, 20 document registration device, 21 internal character code allocation unit (character code Allocation unit), 22 appearance position information creation unit (character code appearance position index generation unit), 23 index creation unit (character code appearance position index generation unit, index storage unit), 24 input processing unit, 30 index database (search target document) Database), 40 internal character code table (character code storage unit), 51 registered document.

Claims (3)

Two or more different notation characters that specify one of a process of comparing one character and a different notation character of this character as the same character and a process of matching these characters as another character are distinguished or regarded as the same. A character that holds a character code dedicated to one character or a character code assigned to a set of candidate characters of the different notation based on the different notation distinguishing condition that specifies A character string conversion unit that converts the character code by referring to the code storage unit and generates a character code string;
A matching unit that reads, from the search target document database, the appearance position information of each character code in the character code string, and compares the character code string with the read occurrence position information;
A document search device including a collation result output unit that outputs a collation result by the collation unit. The character string conversion unit can receive, together with the previous term search keyword, a different notation discrimination condition specified for each character in the search search keyword, and convert each character into a character code in accordance with the above different notation discrimination condition. 2. The document search device according to claim 1, wherein the conversion is performed. Reference is made to a character code storage unit that holds a character code dedicated to one character or a character code assigned to a set of candidate characters for the different characters of the character, and for each character in the input document, the character code dedicated to that character is stored. A character code allocating unit that extracts and, when there is a set of allotted character candidates including the respective characters, extracts a character code assigned to the set of allotted character candidates;
A character code appearance position index generation unit that generates appearance position information in the document of all character codes extracted for each character in the document;
A document registration device comprising an index storage unit for storing an index generated by the character code appearance position index generation unit in a storage device.

Images