JPH07146880A - Document retrieval apparatus and method - Google Patents

Abstract

(57) [Summary] [Object] To provide a document search device and method that can easily register a new document without disturbing the search process and can easily search the registered document. [Structure] When registering a new document in an index, register it in a sub-index that is smaller than the main index, access each index at the time of search, perform a search, and combine the results as the search result Registration can be done early and you can search immediately after registration. In particular, by providing the sub-index in the high-speed main storage device, there is no fear that the search processing speed will decrease. Also, if the secondary index becomes large, and if the secondary index is merged with the primary index to reduce the size of the secondary index when search processing is not being performed, the registration time for new documents can be shortened. There is no need to worry about obstructing the search process.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a document search apparatus and method, and more particularly to a document search apparatus and method for performing fuzzy search including incomplete matching of target character strings.

[0002]

2. Description of the Related Art Conventionally, when checking whether or not a certain character string is included in each document in a storage device in which a large number of documents are stored, it is usual that all the character strings are included. Although an exact match search was performed to check only whether or not there was a case where the search could not be performed due to a subtle difference in the description, especially when searching for a long foreign word written in katakana. Further, there is a problem that the search becomes significantly slow when the total amount of documents to be searched is large, such as the number of documents to be searched.

Therefore, in Japanese Patent Laid-Open No. 4-326164 filed by the same applicant as the applicant of the present application, a large-capacity fixed disk as an index file by mapping autocorrelation information of each character (code) at the same time when a document is stored. By storing it in a memory such as, and obtaining the autocorrelation information of each character of the search character string at the time of search and detecting the presence or absence of it,
There is disclosed a search system capable of easily and rapidly checking not only the presence / absence of a search character string in each search target document but also the matching degree. Such a system speeds up character string retrieval for each document.

However, in the above system, it is necessary to update the index file when a document is newly stored and can be searched. Since this index file is a map showing the correlation between each character,
To register a new document in this index, the information of the correlation between the characters of the new document is interrupted to the information of the existing correlation. Therefore, in order to update the index file, it is necessary to recreate the entire map, which requires a certain amount of time, and there is a problem in that search processing cannot be performed during that time.

Therefore, it is conceivable to provide the above-mentioned index in a main storage device composed of a semiconductor memory which can be accessed at high speed, but there is a problem that the memory capacity becomes insufficient because the index becomes huge when the number of documents increases. .

[0006]

SUMMARY OF THE INVENTION The present invention has been made in view of the above-mentioned problems of the prior art, and its main purpose is to easily register a new document without hindering the search processing. It is an object of the present invention to provide a document search device and method that can easily search even a registered document.

[0007]

According to the present invention, the above-mentioned object has an index formed by mapping autocorrelation information between characters in each document with respect to a plurality of documents stored in a storage device. A document search device for performing a fuzzy search also including incomplete matching of character strings, wherein the index is a main index formed by mapping autocorrelation information between characters in a document that is already stored. A sub-index formed by mapping autocorrelation information between characters of only a newly stored document and searching the character string for all the documents by accessing each index separately. And a main index formed by mapping autocorrelation information between characters in a document already stored in a storage device, and the main index. A document search method for performing a fuzzy search including an incomplete match of a character string with respect to all documents stored in the storage device with an index composed of a sub-index smaller than the index, Autocorrelation information between characters of a newly stored document is registered in the sub-index, and the main index and the sub-index are separately accessed to search all the registered documents for the character string. It is achieved by providing a document search method characterized by:

[0008]

By doing so, it is possible to register a new document at an early stage and to retrieve it immediately after registration. In particular, by providing the sub-index in the high-speed main storage device, there is no fear that the search processing speed will decrease. In addition, when the sub-index becomes large, if the sub-index is merged with the main index as appropriate when the search process is not performed, there is no fear of hindering the search process.

[0009]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT A preferred embodiment of the present invention will now be described in detail with reference to the accompanying drawings.

FIG. 1 is a block diagram showing a system configuration of a server / client type workstation to which the present invention is applied. This system includes a server 1 having a large-capacity auxiliary storage device 2 and a plurality of client machines 4 connected to the server 1 via a known network 3.
And have.

A large number of documents are stored in the storage device 2. In these documents, the autocorrelation information of each character (code) is mapped at the time of storage and stored as a main index file 31 described later together with the document (FIG. 5) and managed by the server 1.

As shown in FIGS. 2 and 5, the server 1 retrieves a certain character string from the above autocorrelation information,
In order to determine the degree of coincidence, a search unit 11 for collating the above character strings with all the documents to determine the degree of coincidence, if necessary, and when storing a new document in the auxiliary storage device 2 The main storage device 12 including a semiconductor memory in which a sub-index 32 for registering the autocorrelation information of each character is formed is provided, and after the document is searched by the character string, the result is displayed on the display unit 13 of the client machine 4. It is possible to do.

The outline of the operating procedure of this embodiment will be described below with reference to the flow chart of FIG.

First, a command is input from a client machine 4 to select whether to search or register a document (steps 1 and 2). When the document is to be registered, the process proceeds to step 3, where the document to be registered is stored in the auxiliary storage device 2 and the autocorrelation information of the document is registered in the sub index 32 of the main storage device 12. And
The result is output to the client machine 4 side (step 4), and the process ends.

On the other hand, when searching for a document, the client machine 4 searches for, for example, the character string "feedforward" for all the documents stored in the storage device 2. This sets the matching threshold of the threshold setting slider 21 displayed on the display unit 13 with a pointing device such as a mouse to, for example, 70% or more while inputting the character string as a "search key" from the client machine 4. And
This client machine 4 accesses the server 1, and the search unit 11 of the server 1 first searches using the main index (step 5), then searches using the sub index (step 6), and combines the results. (Step 7). A conceptual diagram of the processing is shown in FIG. here,
As described above, since the search is performed using the main index and the sub-index in which the autocorrelation information of each document is mapped in advance, it is only necessary to create the autocorrelation information for the character string "feedforward" and collate it with the above map. It is possible to perform high-speed search. The speed of this search hardly depends on the capacity of the entire document, but depends on the length of the character string to be searched.

The search result obtained by the search unit 11 is sent to the client machine 4 as it is, and the document titles having the degree of coincidence equal to or more than the above-mentioned set threshold value are collectively displayed on the display as shown in FIG. 4 (step). 4). Then, the operator, for example, clicks the "sort" key 22 in FIG. 4 with a pointing device to rearrange and display the images in descending order of coincidence.

Next, when the operator selects (clicks) one of the documents displayed in FIG. 4, for example, "levitation supporting device", the content is displayed and processed according to the operator's instruction. To finish.

Although not shown in the figure, the server 1 side chooses a time when there are few requests for various processes, and appropriately selects the sub index in the main index 31 depending on the time required for the search or the size of the sub index 32. It is designed to merge 32. As a result, the size of the sub-index 32 can be always suppressed and the registration time of a new document can be shortened.

[0019]

As is clear from the above description, according to the document search apparatus and method of the present invention, when a new document is registered in the index, it is registered in the sub-index smaller than the main index, and each document is searched in the index. By accessing the index and performing a search, and combining the results as a search result, a new document can be registered at an early stage and can be searched immediately after the registration. In particular, by providing the sub-index in the high-speed main storage device, there is no fear that the search processing speed will decrease. Also, if the secondary index becomes large,
If the sub-index is merged with the main index to suppress the size of the sub-index when the search process is not performed, the registration time of a new document is shortened and there is no fear of hindering the search process.

[Brief description of drawings]

FIG. 1 is a block diagram showing a system configuration of a server / client type workstation to which the present invention is applied.

FIG. 2 is a block diagram showing a part of a functional configuration of a server and a client machine in a server / client type workstation to which the present invention is applied.

FIG. 3 is a flowchart showing a document registration and retrieval procedure in a server / client type workstation to which the present invention is applied.

FIG. 4 is an explanatory diagram showing a display state of a display screen of a client machine.

FIG. 5 is a conceptual diagram of document search processing.

[Explanation of symbols]

 1 Server 2 Auxiliary Storage Device 3 Network 4 Client Device 11 Search Unit 12 Main Storage Device 13 Display Unit 21 Threshold Setting Slider 22 “Sort” Key 31 Main Index File 32 Sub Index

Claims (4)

[Claims] 1. A fuzzy search including an incomplete match of a certain character string is performed with an index formed by mapping autocorrelation information between respective characters in each document with respect to a plurality of documents stored in a storage device. And a main index formed by mapping autocorrelation information between characters in a document that is already stored, and a self index between characters in a newly stored document only. A document retrieving apparatus comprising: a sub index formed by mapping correlation information; and retrieving the character string for all documents by individually accessing each index. 2. The storage device has a small-capacity main storage device capable of high-speed data transfer, and a large-capacity auxiliary storage device which transfers data at a lower speed than the main storage device. The document search device according to claim 1, wherein the main index is stored in the auxiliary storage device and the sub-index is stored in the main storage device. 3. The storage device having an index composed of a main index formed by mapping autocorrelation information between characters in a document already stored in the storage device and a sub-index smaller than the main index. A document search method for performing a fuzzy search including an incomplete match of a certain character string with respect to all documents stored in, wherein the autocorrelation information between characters of a newly stored document is A document search method characterized in that the character string is searched for for all the registered documents by registering in a sub index and separately accessing the main index and the sub index. 4. The main index is merged with the main index to update the main index when the frequency of the search process is low according to the size of the sub index with respect to the main index and the search speed. The document search method according to claim 3.