JP3693734B2 - Information retrieval apparatus and information retrieval method thereof - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an information retrieval apparatus that semantically retrieves data from an accumulated document database and an information retrieval method thereof.
[0002]
[Prior art]
2. Description of the Related Art Conventionally, there has been known an information search apparatus that stores document data in an information processing apparatus such as a personal computer and detects a word / phrase described in the document data by keyword search. An information retrieval apparatus that searches a document for a phrase that includes a part of a character string provided by a user in order to increase the accuracy of retrieval has been put into practical use. In addition, as in Japanese Patent Application Laid-Open No. 6-266770, an information search apparatus that converts a character string provided by a user into a synonym and performs a search with the synonym to detect a synonym in the document has been proposed.
[0003]
[Problems to be solved by the invention]
However, in the search using the above-described information search apparatus, the search accuracy is poor only with the complete match, partial match, and synonyms for the character string given by the user, and the search accuracy is further improved from the user. There is a growing demand for this. For example, if a synonym search is performed to search for a corresponding document, sentence, paragraph, or the like and 500 search results are obtained, the user must confirm the contents of the 500 points on the display screen. The operation for this is very complicated for the user.
[0004]
Therefore, in view of the above-described points, an object of the present invention is to provide an information search apparatus and an information search method thereof that improve search accuracy using a search method that has not been available so far.
[0005]
[Means for Solving the Problems]
In order to achieve such an object, an invention according to claim 1 is an information retrieval apparatus that includes a document database having a plurality of documents and retrieves a document having a character string related to a given keyword. A word association dictionary describing the synonyms, a semantic attribute system dictionary describing the words and their semantic attributes, means for obtaining synonyms for the keywords from the word association dictionary, and semantic attributes for the keywords Means for acquiring from a semantic attribute system dictionary, and a document having the acquired synonym and having a word having the same acquired semantic attribute within a predetermined length range including the synonym. And a means for searching in a database.
[0006]
According to a second aspect of the present invention, in the information search device according to the first aspect, the keyword includes a plurality of words, and further includes means for removing a predetermined word as an unnecessary word from the keyword. And
[0007]
The invention according to claim 3 is the information search device according to claim 1, wherein the range of the predetermined length is formed of a predetermined number or more of character strings, and is a word string that semantically constitutes one word, The searching means acquires the semantic attribute of the word located at the end of the word string from the semantic attribute system dictionary, compares it with the semantic attribute of the keyword, and determines whether or not there is a match.
[0008]
According to a fourth aspect of the present invention, in the information retrieval apparatus according to the first aspect, in the retrieval process, the retrieval unit retrieves a synonym related document including only the synonym in the document database. And
[0009]
The invention of claim 5 further comprises means for displaying information for notifying the document obtained by the search of claim 1 and the synonym-related document obtained by the search of claim 4. It is characterized by.
[0010]
According to a sixth aspect of the present invention, in the information retrieval device according to the fifth aspect, the means for displaying is related to the document obtained by the retrieval according to the first aspect and the synonym relation obtained by the retrieval according to the fourth aspect. It is characterized by distinguishing and displaying information for informing a document.
[0011]
According to a seventh aspect of the present invention, in the information retrieval device according to the fifth aspect, to notify the document obtained by the retrieval according to the first aspect and the synonym related document obtained by the retrieval according to the fourth aspect. This information is a part of the document and synonym related documents.
[0012]
The invention of claim 8 includes a document database having a plurality of documents, and describes a word and its synonyms in an information search method of an information search apparatus for searching for a document having a character string related to a given keyword. A word association dictionary and a semantic attribute system dictionary describing a word and its semantic attributes are preinstalled in the information retrieval device, and the information retrieval device acquires a synonym for the keyword from the word association dictionary. , A semantic attribute for the keyword is acquired from the semantic attribute system dictionary, has the acquired synonym, and the acquired semantic attribute is the same within a predetermined length including the synonym The document database is searched for the document in which the document exists.
[0013]
The invention according to claim 9 is the information search method of the information search apparatus according to claim 8, wherein a document including a word that matches the keyword is searched in the document database, and there is no corresponding document. It shifts to the search of description, It is characterized by the above-mentioned.
[0014]
The invention according to claim 10 is the information search method of the information search device according to claim 8, wherein a sentence including the synonym of the document obtained as a search result is displayed on a display screen, and the synonym and the semantic attribute are displayed. The same word is displayed so that it can be identified.
[0015]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 shows an example of the system configuration of an information retrieval apparatus to which the present invention is applied. An information processing apparatus such as a personal computer or a workstation can be used as the information retrieval apparatus. In FIG. 1, the information retrieval apparatus has a CPU 1, a ROM 2, a RAM 3, an input interface 4, a hard disk storage device 7, a floppy disk drive, and an output interface 9. The CPU 1 uses system programs stored in the ROM 2 or the hard disk storage device 7 to perform system processing, for example, read / write information to / from the memory or storage device, and character code-image conversion for displaying information on the display device 11 (to be described later). Called), display control, control of the printer 10, and reception of input information from the keyboard 5 and mouse 6. The ROM 2 stores a system program and related information for executing the above system processing.
[0016]
The RAM 3 temporarily stores information used for information processing by the CPU 1 and information processing results. As information related to the present invention, document information acquired from the document database of the hard disk storage device 7 is temporarily stored. The input interface 4 transfers input information from a keyboard 5 and a coordinate input device (hereinafter referred to as a mouse) 6 such as a mouse to the CPU 1.
[0017]
The keyboard 5 is mainly used for character input. The mouse 6 is used for designating a position on the display screen. In addition to the system program, the hard disk storage device 7 stores an information search program according to the present invention, a document database to be searched, a word association dictionary and a semantic attribute system dictionary used for information search. Since the word association dictionary and the semantic attribute system dictionary are related to the present invention, they will be described in detail later. The document database of this embodiment is composed of a plurality of documents, and this document has at least a document name (file name) and document data in a character code form.
[0018]
The floppy disk drive 8 receives a floppy disk and reads / writes information from / to the floppy disk. The floppy disk is used to register documents created elsewhere in the document database. The output interface 9 transfers information to be displayed or printed and control information to the printer 10 and the display device 11.
[0019]
In such a system configuration, the present embodiment performs a synonym search on the document database, and examines a sentence including the synonym when a plurality of corresponding search results are obtained. It is characterized in that when a word having the same semantic attribute as the keyword has is included in the sentence, the sentence is handled as a highly weighted search result.
[0020]
A detailed description for this will be given below. FIG. 2 shows a part of the contents of the word association dictionary used for the above information search, and FIG. 3 shows a part of the contents of the semantic attribute system dictionary. The word associative dictionary is composed of a plurality of records, and one record is a character string (hereinafter collectively referred to as a word) 51 such as one character kanji, a word, or a phrase, an identification code assigned to the word, and a synonym (including a similar word). ) 53 and an identification code (referred to as a semantic attribute code) indicating a semantic attribute related to the word. If necessary, other information such as grammatical information indicating nouns, adjectives, etc., and word reading (notation) can be added.
[0021]
The semantic attribute system dictionary is also composed of a plurality of records, each record having a word, an identification code (semantic attribute code) indicating the semantic attribute of the word, and a character string of the word indicating the semantic attribute. The above-mentioned words (including one-character kanji) described in the semantic attribute system dictionary are included in one or more word strings, and a word that is frequently used as the last word is prepared in advance. The Chinese character “person” shown in FIG. 3 is a word used at the end of the word string of accompanying persons, managers, and the like. FIG. 3 shows that a semantic attribute such as person or organization is given to this word, and an identification code of 1001 is given to this semantic attribute.
[0022]
Information search processing using such a word association dictionary and semantic attribute system dictionary will be described with reference to FIGS. 4 and 5. FIG. FIG. 4 functionally shows the processing contents of the information retrieval processing program stored in the hard disk storage device 7. FIG. 5 shows detailed processing in the processing of FIG. When the execution of search is instructed from the keyboard 5, the information search program of FIG. 4 is started and the information search process is started in the CPU 1. As an example of information search, an example will be described in which a character string of the word “Natsu no Koshien” is given and a corresponding sentence in a document in which a word related to this word is described is searched.
[0023]
When the user inputs “Summer Koshien” as a keyword from the keyboard 5, the CPU 1 accepts the input word, that is, “Summer Koshien”, and temporarily stores it in the RAM 3 (step S10).
[0024]
Next, the CPU 1 searches the document database for a word that completely matches the keyword. Since this process is the same as the conventional search process, detailed description will not be required. When a corresponding word is found, a sentence or document including the word, or a document name is displayed (steps S20 → S30 → S35).
[0025]
On the other hand, when a completely matched word is not detected, the CPU 1 proceeds to synonym search. The input word “Natsu no Koshien” in the RAM 3 is decomposed into the minimum unit words “Natsu no Koshien” and “Koshien”, and words not used in the search are removed (step S40). For example, this criterion can make unnecessary words a word that is not described in the word association dictionary or a predetermined word that does not have a synonym described in the word association dictionary. Also, words that are not normally used in words, such as? , @, Etc. should also be removed.
[0026]
In this example, “summer” is removed, and “Koshien” is obtained as a synonym search keyword. It should be noted that the technique for word decomposition is well known in the name of parsing and will not require detailed explanation. The simplest method is to divide the character string in units of words described in the word association dictionary. The CPU 1 acquires the synonym “high school baseball” and the semantic attribute code “2004” by referring to the word association dictionary based on the obtained keyword “Koshien” (see step S50, FIG. 2).
[0027]
Next, the CPU 1 searches the document database of the hard disk storage device 7 to search for a document in which the acquired synonym is described (step S60). When a corresponding document is found (YES in step S70), the degree of association between the keyword (Koshien) given by the user and the synonym (high school baseball) is weighted, that is, the degree of association. Is calculated (step S80). A detailed procedure for this is shown in FIG.
[0028]
In FIG. 5, the CPU 1 extracts a sentence including synonyms from a document obtained as a result of the search. A simple method is to detect a character code representing a sentence delimiter such as “.” Before and after the synonym and a space character, and take out a character string sandwiched between the character codes as a sentence (step S100). The taken-out sentence is decomposed into words, more specifically, excluding predetermined unnecessary words, the sentence is decomposed into words (including phrases and the like) composed of four or more character strings (step S110). Then, the word including the synonym (with the smallest unit length and located at the end is extracted. For example, the extracted sentence is “This year's National High School Baseball Players ... ", A character string of" National High School Baseball Tournament "is taken out as a word of four or more characters (a word string composed of a predetermined number of character strings and semantically constituting one word). "Meeting" is taken out as a word of minimum unit length.
[0029]
For this word “meeting”, the CPU 1 refers to the semantic attribute system dictionary of FIG. 3 and acquires “2004” as the semantic attribute code. When such processing is performed on all search results (documents including synonyms), a semantic attribute code is obtained for each document (steps S120 → S130). As described above, since the semantic attribute code for the user keyword “Koshien” has already been checked and stored in the RAM 3, the CPU 11 uses a plurality of keywords derived from the semantic attribute code (2004) of the user keyword and all the search results documents. Are compared with the meaning attribute codes (step S140). A document in which a match is found is treated as a highly relevant document, and a document that does not match is treated as a less relevant document. These documents are stored in the corresponding areas on the RAM 3 (steps S150 and S155).
[0030]
Returning to FIG. 4, the CPU 1 extracts the corresponding part in the document as a search result from the RAM 3 and displays it according to the degree of relevance (weight) (step S <b> 90). An example of the display is shown in FIG. In this display example, the sentence with the matching semantic attribute code is displayed as the final search result, and as a reference, that is, the sentence that includes the synonym but the semantic attribute code does not match is displayed as the reference, and the related This is an example in which the degree is divided into two.
[0031]
As described above, in this embodiment, in addition to the synonym search, a sentence in which a word having a keyword and a semantic attribute is present near a synonym (similar word) is detected. You can get something close to the sentence you are looking for.
[0032]
In addition to this example, the following example can be implemented.
[0033]
1) In this embodiment, the search target is a sentence described in the document, but a paragraph, the entire document, or the like may be the search target. In addition, the type of sentence, paragraph, document, etc. may be designated from the keyboard. Furthermore, instead of the content of the sentence, the title name or file name registered in the document database may be used as a search result, or the information and the corresponding sentence may be displayed together. In other words, the document described in claim 1 includes the document itself, a part of the document, the document title, the file name, and the like.
[0034]
2) In the present embodiment, in order to detect a word in a document near a synonym having the same semantic attribute of a keyword, the longest word (including a compound word) including the synonym in the document is extracted, Checking semantic attributes. This is because the inventor realized that when the last word of the longest word is Japanese, it expresses the semantic attribute best, but if you want to further improve the search accuracy even at the expense of search time. The following detection method may be used. After extracting a sentence including a synonym, the word is extracted from the synonym as a starting point toward the beginning and end of the sentence. The semantic attribute code of the extracted word is extracted from the semantic attribute system dictionary and compared with the semantic attribute code of the keyword. If a match is found, the sentence is treated as a sentence with high relevance. In addition, if you want to change the degree of search accuracy, the search range for words that have the same semantic attributes as keywords is not limited to sentences containing synonyms, but can be selectively expanded to paragraphs, pages, or entire documents. Good. This selection is sentence, paragraph. . . Thus, the search range may be automatically expanded until a word having the same semantic attribute code is found, or the user may manually specify the search range from the keyboard.
[0035]
3) In this embodiment, the degree of association between the keyword and the sentence to be searched is divided into two stages, but it can be made two or more stages. In this case, the number of words having the same meaning attribute included in a predetermined length sentence including synonyms, for example, one page is counted. Corresponding to this numerical value, the search accuracy can be divided into multiple levels.
[0036]
4) As a display form of the search result, in addition to the sentence including the synonym, the longest word including the synonym is displayed in a rectangle as shown in FIG. 6, or the word having the same synonym and semantic attribute is displayed in another color. It is possible to inform the user of words related to the keyword such as. In addition, the form of notification may be a color, a figure such as a rectangle, a typeface change, an underline, a character size change, or the like.
[0037]
5) As a method of using the information search apparatus, it may be used alone, or it may be connected to a communication line, accepting a keyword from another computer of the search request, and returning the search result to the computer of the search request. . In this case, the information retrieval apparatus may be provided with a communication means.
[0038]
6) The storage location of the document database mounted on the information retrieval apparatus is not limited to the hard disk storage device, and any suitable known storage means such as a magneto-optical storage device may be used.
[0039]
【The invention's effect】
As described above, according to the inventions of claims 1 and 8, a document having both a synonym and a word having the same semantic attribute with respect to a keyword, and the two words are close to each other. Since the search is performed, it is possible to obtain a document related to the keyword as compared with the case where the search for only the synonym or the semantic attribute is continuously performed.
[0040]
According to the invention of claim 2, the user can use a keyword composed of a plurality of desired words without preparing a single word.
[0041]
According to the third aspect of the present invention, it is possible to take out a word that best relates to a synonym and expresses a semantic attribute.
[0042]
According to the fourth aspect of the present invention, the search failure due to the narrowing down of the keyword is eliminated by acquiring the search result of only the synonym search.
[0043]
According to the invention of claim 5, the user can know the contents and location of the document by displaying the title name of the document, the document itself, and a part of the document.
[0044]
In the invention of claim 6, the user can know the location of the document according to the degree of relevance of the keyword.
[0045]
According to the invention of claim 7, by displaying a part of the document as the search result, the user can confirm the degree of association with the keyword.
[0046]
According to the ninth aspect of the present invention, the search time and the search accuracy can be balanced by executing the synonym search according to the present invention after performing the keyword matching search.
[0047]
In the invention of claim 10, the user can know the synonyms and semantic attribute words used in the automatic search.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a system configuration of an information search apparatus to which the present invention is applied.
FIG. 2 is an explanatory diagram showing a part of the contents of a word association dictionary.
FIG. 3 is an explanatory diagram showing a part of the contents of a semantic attribute system dictionary.
FIG. 4 is a flowchart showing a processing procedure related to information retrieval.
FIG. 5 is a flowchart showing a processing procedure related to information retrieval.
FIG. 6 is an explanatory diagram showing a display example of search results.
[Explanation of symbols]
1 CPU
2 ROM
3 RAM
4 Input interface 5 Keyboard 6 Mouse 7 Hard disk storage device 8 Floppy disk drive 9 Output interface 10 Printer 11 Display device

Claims (10)

In an information retrieval apparatus for loading a document database having a plurality of documents and retrieving a document having a character string related to a given keyword,
A word association dictionary that lists words and their synonyms;
A semantic attribute system dictionary describing words and their semantic attributes;
Means for obtaining a synonym for the keyword from the word association dictionary;
Means for obtaining a semantic attribute for the keyword from the semantic attribute system dictionary;
And means for searching the document database for a document having the acquired synonym and having a word having the same acquired semantic attribute within a predetermined length including the synonym. An information retrieval apparatus characterized by that. The information search apparatus according to claim 1, further comprising means for removing a predetermined word from the keyword as an unnecessary word, the keyword being composed of a plurality of words. 2. The information search apparatus according to claim 1, wherein the range of the predetermined length is a word string composed of a predetermined number or more of character strings, semantically constituting one word, and the means for searching includes the word An information search apparatus characterized in that a semantic attribute of a word located at the end of a column is acquired from the semantic attribute system dictionary and compared with the semantic attribute of the keyword to determine whether or not there is a match. 2. The information search apparatus according to claim 1, wherein the searching unit searches the document database for a synonym related document including only the synonym in a search process. An information retrieval apparatus further comprising means for displaying information for notifying the document obtained by the search according to claim 1 and the synonym related document obtained by the search according to claim 4. . 6. The information search apparatus according to claim 5, wherein the display means is information for notifying the document obtained by the search according to claim 1 and the synonym related document obtained by the search according to claim 4. Information search apparatus characterized by distinguishing and displaying 6. The information search apparatus according to claim 5, wherein information for notifying the document obtained by the search according to claim 1 and the synonym related document obtained by the search according to claim 4 is the document and the synonym. An information search apparatus characterized by being part of a word-related document. In an information search method of an information search apparatus for loading a document database having a plurality of documents and searching for a document having a character string related to a given keyword,
A word association dictionary describing a word and its synonyms and a semantic attribute system dictionary describing a word and its semantic attributes are preinstalled in the information search device, and the information search device is a synonym for the keyword From the word association dictionary,
Obtaining a semantic attribute for the keyword from the semantic attribute system dictionary;
The document database is searched for a document having the acquired synonym and having a word having the same acquired semantic attribute within a predetermined length including the synonym. Information retrieval method of information retrieval apparatus. 9. The information search method of the information search device according to claim 8, wherein a document including a word that matches the keyword is searched in the document database, and when there is no corresponding document, the search is shifted to the search according to claim 8. The information search method of the information search device characterized by the above-mentioned. 9. The information search method of the information search device according to claim 8, wherein a sentence including the synonym of the document obtained as a search result is displayed on a display screen, and the word having the same synonym and semantic attribute can be identified. An information search method for an information search apparatus, characterized by:

Images