JPH0227478A - Document management device - Google Patents

Abstract

PURPOSE:To attain ambiguous retrieve to express the magnitude of the relation between a key word and each document by selecting the document conforming to a retrieving request with the use of an expanded inverted file according to the retrieving request of a user, and managing an input/output. CONSTITUTION:An expanded inverted file 7 to hold the relation of the key word and each document by a numeric value is provided. For this reason, when the document is retrieved by a document selecting part 3, the magnitude of the relating degree between the key word group designated by a user 10 and each document obtained by prescribed calculation from the expanded inverted file 7 can be defined as file accuracy. Thus, the ambiguous retrieve to indicate a retrieve result in such a sequence is attained, and when the document is retrieved, a threshold value control type ambiguous retrieve, in which the file accuracy required by the user is set at a threshold value, and the document number control type ambiguous retrieve, in which the document required by the user is selected in the sequence at the larger file accuracy and it is made into the retrieve result, are attained.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a document management device, and particularly to a document management device in which search results are ordered by evaluation values having larger values as they are closer to a user's search request during document retrieval. This invention relates to a flexible and high-speed document management device.

(Prior art) In a conventional document management device, an inverted file expresses the relationship between a keyword and a document as a binary value of ``yes'' and ``no''.On the other hand, when handling actual documents, when a certain keyword There are many situations where the keyword is in the document but is not particularly important to the content of the document, but in this case, the conventional inverted file contains the keyword, so it is not possible to force a relationship between the keyword and the document. You must choose either to assume that there is a relationship between the keyword and the document, or to determine that there is no relationship between the keyword and the document, since the relationship is not very important. Therefore, conventional inverted files do not express the magnitude of the relationship between keywords and documents.

(Problem to be Solved by the Invention) In this way, a document management device using inverted files responds to a user's search request with a binary value of ``does'' or ``does not satisfy'' each document. It lacks flexibility as it can only answer questions. Therefore, in order to find a document that almost, but not completely, satisfies the search conditions specified by the user, it is necessary to perform the search again. Furthermore, if there is no document that satisfies the search conditions specified by the user, it is not possible to search for the document closest to the search conditions, which is extremely troublesome.

SUMMARY OF THE INVENTION An object of the present invention is to eliminate the drawbacks of the prior art and provide a document management device that expresses the magnitude of the relationship between keywords and documents.

(Means for Solving the Problems) In order to solve the above-mentioned problems, the present invention includes a document information management unit that stores registered documents and their bibliographic information, and a numerical value that maintains the relationship between keywords and each registered document. An extended inverted file management section that stores and manages the extended inverted 7 files; a document registration section that registers, updates, and deletes documents according to user registration requests; and an extended inverted file management section that stores and manages the extended inverted 7 files; It has a document selection unit that selects documents that match a search request using the search engine, and an input/output management unit that manages input and output.

(Function) According to the present invention, by providing an extended inverted file that holds the relationship between keywords and each registered document in numerical values, when searching for documents, the keyword group specified by the user and the extended inverted file are provided. File accuracy is defined as the degree of association between each document and a group of user-specified keywords, which is determined by a predetermined calculation from a searched file.
When searching for documents, use a threshold-controlled fuzzy search that uses the file accuracy requested by the user as a threshold and selects documents with a file accuracy greater than or equal to the threshold and uses them as search results, or searches for as many documents as the number of documents requested by the user. It becomes possible to perform a fuzzy search using document count control, which selects files in descending order of file certainty and uses them as search results.

(Example) Next, an example of the document management device of the present invention will be described with reference to the accompanying drawings.

In order to eliminate the drawbacks of the prior art, the document management device of the present invention uses an extended inverted file that holds the relationship between keywords and each registered document using numerical values. An evaluation value called file accuracy is introduced for this file, which has a larger value as the file is closer to the user's search request, and when searching for documents, the user can flexibly select documents that meet the request based on the size of the file accuracy. can.

The new concept of extended inverted files, which is a feature of the present invention, will be explained.

In conventional document management devices, keywords and documents are related by in-mart files. but,
In conventional inverted files, the relationship between a keyword and a document is expressed as a binary value of "Yes" or "No".However, in reality, even if a keyword exists in a certain document, there is a relationship between the keyword and the document. There are varying degrees of relationship.

Therefore, an inverted file is used that has the size of the relationship between keywords and documents as additional information. An inverted file having such additional information is called an extended inverted 7 file. In extended inverted files, the magnitude of the relationship is set in the range 0 to 1, where 0 indicates that there is no relationship between the keyword and the document, a value of 0 or more indicates that there is a relationship, and 1 indicates that the magnitude of the relationship is It shall be expressed as the maximum.

The expanded inverted file can also be viewed as a two-dimensional array. In that case, the relationship between the first document and the first keyword is WiCKj] or more simply W
It is expressed as ij.

Next, the relationship between keywords and documents is shown in Figure 2 (a
). An example of an extended inverted file at this time is shown in FIG. 2(b), and an example of a conventional inverted file is shown in FIG. 2(c).

In the conventional inverted file shown in FIG. 2(c), only the document numbers of documents related to each keyword are restored. On the other hand, in an extended inverted file, documents related to each keyword are restored as a set of document number and the size of the relationship between that keyword and the document, such as (document l, 1.0). ing.

Also in this extended inverted file, documents that are unrelated to a certain keyword are not entered in the keyword list. Therefore, a document that has not been restored for a certain keyword means that there is no relationship with that keyword, or that the relationship rate with that keyword is 0. For example, in the list of keyword l, document 3
Although FilJ is not stored, this means that the keyword 1 and document 3 are unrelated, or that their relationship speed is O.

FIG. 1 shows the configuration of the document management device of the present invention.

The input/output management section l includes an input section (not shown) that receives and receives requests from users, a request processing section that sends necessary information to the document registration support section 2 or document selection support section 3 according to the user's requests, and a request processing section that sends necessary information to the document registration support section 2 or the document selection support section 3 according to the user's requests. It consists of a display unit (not shown) that displays the processing results.

The document registration support section 2 includes a keyword creation support section (not shown) that extracts keyword candidates from registered documents, and an extended inverted file creation support section that supports the user in creating an extended inverted file. The process performed here is to instruct the user to update the document information database/extended inverted file when the user registers the document. Furthermore, it instructs to update the document information database 6 and extended inverted file 7 not only when registering a document but also when updating, changing, or deleting a document.

The document selection support unit 3 includes a file accuracy calculation unit (not shown) that calculates file accuracy using the user's search conditions and information in the expanded inverted file, and a file accuracy calculation unit (not shown) that arranges documents in the document information database in order of increasing file accuracy. a sorting section that changes, a threshold control section that performs threshold processing for document selection,
It consists of a document number control section that controls the number of documents for document selection. The processing of each of these parts will be explained in detail in the operation description.

The document information management section 4 manages a document information database 6 that holds registered documents and their bibliographic information, and registers, searches, updates, and deletes the contents according to requests from other sections.

The extended inverted file management unit 5 is an extended inverted file management unit 5.
Manage Ted File 7. The extended inverted file management section S registers, searches, updates, and deletes the contents of the extended inverted file 7 in accordance with requests from other sections. Furthermore, when the document opening speed of keywords in the expanded inverted file 7 is changed appropriately to make it easier for the user to use the extended inverted file 7 as a future expansion, the amount of change in the opening speed due to learning will also be calculated.

The processing of the document management device of this embodiment can be roughly divided into two: document registration processing and document selection processing.

However, document registration processing shall also include document update, correction, and deletion processing. The request processing section in the input/output control section analyzes which of these two processes the user is trying to perform, and the actual processing is performed by the document registration support section 2 if it is document registration processing, and by the document registration support section 2 if it is document selection processing. Document selection support section 3
It will be held in These processes will be explained below.

The purpose of document registration processing is to convert registered documents and their accompanying bibliographic information into a data page that is easy to use when selecting documents. The document registration process is performed by the document registration unit 2.

Examples of registered documents and their bibliographic information are shown in Figures 3 (,) to (c).
Shown below. FIG. 3(a) shows the registered document, and FIG. 3(b) shows its bibliographic information. Through the processing in the keyword creation support section of the document registration support section 2, keyword candidates are extracted from the document as shown in FIG. Through dialogue with the department, the user 10 refers to the keyword candidates and decides whether each keyword candidate should actually be used as a keyword, and if so, what size of opening speed should be used. (a) is a set of keywords actually added to the registered document by the user and the magnitude of their related speeds.Finally, the extended inverted 7 isle management unit 5 shows the keywords added and their related speeds. The set is added to the already existing extended inverted file 7.

Further, the registered documents and bibliographic information are stored in the document information database 6 through the document information management section 4.

The purpose of document search processing is to search and display appropriate documents according to user requests. The document selection process is performed by the document selection support section 3.

Search conditions are performed by specifying keywords and setting conditions for other bibliographic information. First, it is determined whether condition settings other than keywords for each document are satisfied. If the conditions are met, the file certainty of the document is calculated; if the conditions are not met, the file certainty of the document is not calculated and the value is set to zero.

Keywords are specified using keyword expressions that express the information the user wants to obtain. A keyword expression is one keyword or multiple keywords AND (product operation)
- It is a combination of OR (sum operation) and NOT (negation operation). A combination of multiple keywords is used when a single keyword expressing the information the user is trying to obtain does not exist in the document management device. For example, if the keyword ``used foreign car'' does not exist in the document management device, the keyword expression % expression %) using the existing keywords ``automobile'', ``domestic production'', and ``used'' represents negation. Further, "#" represents OR.

Next, a method of calculating each file probability for the keyword expression Quarry will be described (the actual calculation is performed by the file probability calculation section in the document selection support section 3).

Below, the i-th keyword is Ki and the first document is F.
ls The search formula specified by the user is KCLIST, and the (multiple) keywords included in document Fi are KFLIS.
Represented by T. Furthermore, the file accuracy of the i-th document is set at
The file accuracy calculation method is shown below.

A:) AND: Quarry=K l & K
In case of 2, Ci (Quary) =Wi [:1]
GDWi [:ni2] where ■ is a generalized product operation,
Calculated by one of the following methods a) OR: If Quarry=K 1 #K 2 then Ci [Qua ry] = Wt (Kl :I eWi
(K2] Here, e is a generalized sum operation and is calculated by one of the following methods) NOT: Quarry=! KCi
[Qua ry] = 1-Wi (K) Engineering) In general, any search expression can be expressed as a combination of three operations: AND10RI/NOT, and the file accuracy can also be calculated by a combination of the above three calculation methods. Calculated.

For example, suppose that Quarry=(!Kl)&(3 in K2S). At this time, C1(Quary)"(tWi (Kl:l)O(Wt
(K2)flBWi(K3:]) Next, a specific example will be shown. Consider the case where the extended inverted file is as shown in FIG. 2(b).

At this time, the file accuracy of documents 1 to 4 is
-Eng.) for the four search expressions.

(Algebraic product and algebraic sum were used as the generalized product and generalized sum, respectively.) The results are shown in FIG.

Once the file accuracy calculation unit of the document registration support unit 3 calculates the file accuracy, the sorting unit of the document registration support unit 3 sorts the documents in descending order of file accuracy. The results are then sent to the input/output management section 1 and displayed to the user 10.

Furthermore, when the user 10 selects the threshold-controlled fuzzy search, the sorted results are subjected to threshold processing by the threshold control section of the document selection support section 3 and sent to the input/output management section l. Threshold-controlled fuzzy search is effective when users have clear qualitative requirements for the documents they want. For example, if you don't want low-quality articles, you can set a high threshold. Alternatively, if the user selects document count control type fuzzy search, the sorted results will be sent to the document selection support unit 3.
The document number control unit performs document number control processing and is sent to the input/output management unit l. Such document count control type fuzzy search is effective when the user has specific qualitative requirements for the document he or she wants.

(Effect of the invention) As explained above, the present invention has the following effects.

(1) Fuzzy search is possible.

With the introduction of enhanced inverted files and the associated file certainty, users are now informed of whether or not each file matches the search criteria using a continuous value rather than a binary value of match or failure. be done. Therefore, users can select necessary files based on their own criteria/judgment.

(2) Threshold-controlled fuzzy search is possible.

Threshold-controlled fuzzy search allows a user to specify a necessary file as having a file certainty greater than or equal to a certain threshold, and display only files with a file certainty greater than or equal to that threshold as search results.

(3) Fuzzy search with controlled number of files is possible.

With file number control fuzzy search, the user can specify the number of files he or she needs and have that number of files or the number of files closest to that number displayed as search results.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an embodiment of a document management device according to the present invention, FIG. 2(a) is a diagram showing the relationship between keywords and documents, and FIG.
Figure (b) is a diagram showing an extended inverted file, the second
Figure (c) shows a conventional inverted file, Figure 3 (a) is an example of a registered document, Figure 3 (b) is an example of its bibliographic information, and Figure 3 (c) is an extracted file. FIG. 3(d) is a diagram showing an example of keyword candidates, and FIG. 3(d) is a diagram showing an example of keywords and their degrees of association. FIG. 4 is a diagram showing an example of calculating the file certainty of a document. l...Input/output management section 2...Document registration support section 3...Document selection support section 4...Document information V-embedding section 5...Extended inverted file management section 6...Document information Database 7...Extended inverted file

Claims (1)

[Scope of Claims] 1. Document information management means for storing and managing registered documents and their bibliographic information, and extended inverted files for storing and managing extended inverted files that numerically maintain relationships between keywords and each registered document. a document registration means for registering, updating, and deleting documents according to a user's registration request; and a document for selecting a document matching the search request using the extended inverted file according to the user's search request. It has a selection means and an input/output management means for managing input/output, and when searching for a document, calculates the degree of relevance between each document and a group of keywords specified by the user by performing a predetermined calculation from the expanded inverted file. A document management device characterized in that a file accuracy indicating a file accuracy is determined, and a search result is displayed based on the file accuracy. 2. In the document management device according to claim 1, when searching for a document, one of a document having the file accuracy equal to or higher than the file accuracy threshold requested by the user and a document within the number of documents requested by the user is searched. A document management device characterized in that the files are selected in descending order of file certainty.