JP2002049638A - Document information retrieval device, method, document information retrieval program and computer readable recording medium storing document information retrieval program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To retrieve a document similar to a document that is not registered in a retrieval database by a simple operation quickly. SOLUTION: This document information retrieval device retrieves document information in a server 10 and replies on the basis of a retrieval request from a client 12, and when a retrieval condition designating part 26 in the client 12 designates a document file as a retrieval condition, transmits the designated contents of the file through a network. A document retrieving part 30 in a retrieval machine 20 installed on the server 10 side generates a key word from the contents of the file transmitted from the retrieval condition designating part 26, and retrieves a similar document by an index (important word string extracted from a retrieval subject document 25).

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a document information retrieving apparatus and method for quickly retrieving a required document from a large amount of document data, and a computer-readable recording medium storing a document information retrieving program. The present invention relates to a document information search device and method for searching for documents having similar contents by a simple operation of designating a document file itself as a search condition, and a computer-readable recording medium storing a document information search program.

[0002]

2. Description of the Related Art Conventionally, in a document management system using a network environment, document information which allows a user to search for a required document from a large amount of document data existing on the Internet or Ethernet (R) and to quickly refer to the document information. Provides a search device.

[0003] In this case, in the document search, one or a plurality of appropriate words or character strings that are considered to be included in a document required by the user are designated as keywords, and a document containing the word of the designated keyword is searched. Search from the search database and display a list of documents as search results.

In this document information search apparatus, for a search target document existing on a network, an important word is extracted from the contents and an enumerated index is created for each document and stored in a search database. When a user issues a search request specifying a keyword, an index of a search database is searched to obtain a search result of a document list.

Further, the conventional document information search apparatus searches for a document that is deemed necessary from a list of documents searched by a user by specifying a keyword, and then specifies a similar document search for a document selected from the list of documents. A term frequently appearing in a search document is automatically extracted and added to the previously executed search condition by a logical sum condition, so that a similar document can be searched.

[0006]

However, if a user obtains a document of interest via e-mail or the Internet and wants to search for a document similar in content to this document,
At present, a word or a character string included in an obtained document is selected and designated as a keyword, and a document list is first obtained as a search result. Next, a similar document must be searched by selecting a document from the searched document list and specifying a similar document search.

That is, even if a similar search for a document obtained by e-mail or the Internet is to be performed, the conventional document information search apparatus only searches for a document already registered in the search database, and searches for a similar document that specifies the document as a search condition. Cannot be searched, and a user cannot directly search for a similar document by using a document obtained by e-mail or the Internet as a search condition.

For this reason, it is necessary for the user to select a keyword deemed necessary for document search from the documents obtained by e-mail or the Internet and input it as a search condition. If there are many keywords, it takes time to input. .
If the keyword is not specified enough, search omission will occur,
In some cases, expected search results cannot be obtained.

Further, the number of searches obtained as a document list may be enormous, and it may take a great deal of trouble to open a document considered to be related from the document list and search for a necessary document.

SUMMARY OF THE INVENTION The present invention provides a document information search apparatus and method capable of quickly searching for a document similar to a document not registered in a search database by a simple operation, and a computer-readable recording medium storing a document information search program. The purpose is to:

[0011]

FIG. 1 is a diagram illustrating the principle of the present invention. The present invention relates to a document information search apparatus that searches and responds to document information on a search side such as a server 10 based on a search request from a client 12 or the like via a network. When a file is specified as a condition, a search condition specifying unit 26 that transmits the specified file content via a network is provided. A search side such as the server 10 transmits a keyword based on the file content transmitted from the search condition specifying unit 26 to the server 10 or the like. , And a search machine 20 for searching for similar documents is provided.

For this reason, when a document containing the content of interest is obtained by e-mail, the Internet, or the like, and a user wants to search for a document having a content similar to this document, the uploaded file is specified as a search condition by specifying the document. By doing so, documents with similar contents can be searched. Therefore, even a document that is not registered in the database can be freely designated as a search condition, and it is not necessary to input a keyword based on the troublesome document content, and a similar document can be found easily and quickly.

The search condition specifying unit 26 of the search request source transmits the first file portion of the specified file contents. Normally, since important keywords required for document search are often present at the beginning of a document, only the beginning of the file contents, for example, the first 1 KB portion, is transmitted as a search condition. Further, since the size of the document file used for the search condition varies, the communication load and the processing on the search side are reduced by determining the size of the file to be transmitted as the search condition.

The search condition specification section 26 includes HTML files and Excel files in files specified as search conditions. Of course, other file formats include any file format as long as the text document can be extracted.

The search machine 20 on the server 10 side is provided with a database 22 in which index information listing key words extracted from documents to be searched is stored for each document.
The file designation search unit 30 of the search machine 20 includes a text extraction processing unit 36 that extracts a text sentence from the file content received in response to the search request, a morphological analysis unit 38 that extracts a noun by morphological analysis of the text sentence, A keyword generation unit 40 that extracts keywords from among the keywords and generates a keyword connected by a logical sum, and a search execution unit 42 that searches for a similar document by searching the search database 22 using the keyword and notifies the client of the search result. Is provided.

The keyword generation section 40 counts the number of appearances H of each document in the index of each search document stored in the search database 22 and determines the number of occurrences H in a predetermined range. Select a number of words to generate keywords.

When the number of documents in the index is N, the keyword generation unit 40 selects, for example, the top 10 words having the number of appearances H in the range of 2N / 3 ≧ H ≧ 1 and selects the keyword as a keyword. Generate. This narrows down important words necessary for similarity search of existing documents registered in the index of the database, and improves the accuracy of similarity search.

Further, the keyword generation section 40 causes the property information extracted from the file received along with the search request to be included in the keyword and searched. In this case, the property information includes the creator of the file received with the search request,
Document title, etc. By adding the property information of the file to the search condition in this way, it is possible to appropriately narrow down the similar documents when it is desired to specify the creator, for example.

The search condition specifying unit 26 of the search request source is provided by the WWW browser 16 of the client 12, transmits the file contents specified on the search request screen of the WWW browser 16 to the WWW server 18 via the network, and Hand over to

The present invention also provides a search machine 20 serving as a document information search device on the search side such as a server. The document information search device as the search machine 20 includes a search database 22 storing index information listing key words extracted from a search target document for each document, and a search request from a network specifying a document file as a search condition. A text extraction processing unit 36 that extracts a text sentence from the file content received by the morphological analysis unit 28, a morphological analysis unit 28 that extracts a noun by morphological analysis of the text sentence, and extracts a key word from the noun to generate a keyword connected by a logical sum And a search execution unit 42 that searches for a similar document by searching a search database using a keyword and notifies the request source of the search result.

The present invention provides a document information search method for searching and responding to document information on a search machine such as a server based on a search request via a network from a search request source such as a client. In this document information search method, index information in which important words extracted from a search target document are listed is stored in a search database of a server for each document; when a document file is specified as a search condition, a search request for the specified file content is made. With the searcher via the network; the searcher extracts the text sentence from the file content received in response to the search request, and extracts the noun by morphological analysis of the text sentence. The method is characterized in that keywords are extracted and a keyword connected by a logical sum is generated, a similar document is searched by searching the search database using the keyword, and a search result is notified to a client. The details of this document information search method are basically the same as the device configuration.

Further, the present invention provides a computer-readable recording medium storing a document information search program, the document information search program receiving a search request specifying a document file as a search condition; Extracting a text sentence from the file content received in response to the search request, extracting a noun by morphological analysis of the text sentence, extracting an important word from the noun and generating a keyword connected by a logical sum And a step of searching for a similar document by searching the database using a keyword and notifying the request source of the search result.

Further, the present invention provides a document information search program, the program comprising: a step of receiving a search request in which a document file is specified as a search condition; Extracting a text sentence from the content, extracting a noun by morphological analysis of the text sentence, extracting an important word from the noun and generating a keyword connected by a logical sum, and searching the database by the keyword And searching for similar documents and notifying the requester of the search result.

[0024]

FIG. 2 shows a system configuration of a document information retrieval apparatus according to the present invention, taking as an example a case where the document information retrieval apparatus is constructed as a server-client type retrieval system using the Internet or Ethernet (R).

In FIG. 2, a client 12 on the user side is connected to the server 10 via the Internet / intranet 14. The client 12 is provided with a WWW browser 16 for searching.
A search request for document information is made to the server 10 using the WW browser 16, and the search result on the server 10 side is displayed.

The server 10 includes a WWW server 18, a search machine 20, and a document database 24.
The search machine 20 stores a search database 22. The document database 24 has a search target document 25.
Is stored. Further, external document management servers 44 and 48 are connected to the WWW server 18, and the document management servers 44 and 48 are also provided with document databases 46 and 50, respectively, and store the search target documents 25.

The WWW server 18 provided in the server 10
Receives the search request from the browser 16 and requests the search machine 20 for the search. The search result returned from the search machine 20 is returned to the browser 16 for display.

The search database 22 functions as a repository for managing an index made up of a set of important words described in a document to be searched in order to process a full-text search at high speed. The index records the document name of the document and its storage location. When a search request is received from the browser 16, the search machine 20 executes a search process on the index of the search database 22.

The document database 24 stores search target documents 25 collected from the document management servers 44 and 48, and an index of the search database 22 is created for the document database search target documents 25.

In such a server-client type search system, using the browser 16 of the client 12, the search condition specified by the user is transmitted to the WWW of the server 10 via the Internet / intranet 14.
Send to server 18. The specified search condition included in the search request received by the WWW server 18 is the WWW server 1
8 to the search machine 20.

The search machine 20 searches the search database 22 for documents that meet the search conditions, and notifies the WWW server 18 of the search results. WWW server 18 is search machine 2
The search result from 0 is sent to the browser 16 of the client 12 and displayed.

The user looks at the search result processed by the browser 16 and selects a link described in the search result, so that the search target document 25 desired by the user is selected from the selected documents via the WWW server 18. You can upload and view the contents.

FIG. 3 is a block diagram of a functional configuration in the search system of FIG. First, the WWW browser 16 on the user side is provided with a search condition specifying unit 26. The search condition specifying unit 26 according to the present invention directly specifies a document file obtained by the user via the Internet or e-mail as a search condition, and specifies the specified file content via the Internet / intranet 14 via the WWW.
Document search unit 30 of search machine 20 via server 18
Send to

The search condition designating section 26 includes a search condition for file designation newly provided in the present invention,
(1) Keyword search, (2) Detailed search to search by specifying keywords for each document title, creator, text, and (3) Document related to text content by inputting everyday words and sentences Sentence search to search, furthermore
(4) It is also possible to specify search conditions such as similar document search using an existing document registered in the search database 22 as a search condition.

The search machine 20 provided on the WWW server 18 is provided with a search database creation unit 28, a document search unit 30, and a document reference unit 32. The search database creation unit 28 creates and registers an index in the search database 22.

That is, the search database creation unit 28 extracts the key words described in the search target document 25 from each of the search target documents 25 collected and stored in the document database 24, and Create and save an index composed of a set of words. of course,
The index also records the document name and storage location of the search target document.

The document search unit 30 generates a keyword from the file content when a file is specified as the search condition transmitted from the search condition specification unit 26 of the WWW browser 16, and generates an important keyword included in the index of the search database 22. A search and collation with a set of words are performed, a document similar to the document of the file specified as the search condition is searched by the WWW browser 16, and the search result is sent from the WWW server 18 to the WWW server 18.
It is returned to the W browser 16 and displayed.

When the document reference unit 32 selects a document to be referenced from the document list as a search result sent from the WWW browser 16 and notifies the document reference unit 32 via the WWW server 18, the document reference unit 32 The requested reference document is extracted from the inside and returned to the WWW browser 16.

FIG. 4 shows details of the functional configuration of the document search unit 30 of the present invention provided in the search machine 20 of FIG.

In FIG. 4, the document search unit 30 includes a search designation file storage unit 34, a text extraction processing unit 36, a morphological analysis unit 38, a keyword creation unit 40, and a search execution unit 42. In the search database 22, an index 52 including a set of important words, a document name, a storage location, and the like of each of the search target documents 25 in the document database 24 created by the search database creation unit 28 of FIG. Is stored.

The file content transmitted by the file designation of the search condition designation unit 26 in the WWW browser 16 of FIG. 3 is stored in the search designation file storage unit 34 of the document search unit 30.

Here, the transfer of the file contents from the WWW browser 16 is performed by cutting out the first file portion of the document file designated as the search condition, for example, the first 1 KB, and transmitting it to the WWW server 18 together with the search request.

As described above, by setting the file capacity to be transmitted as the search condition to a fixed capacity, for example, 1 KB, the content of the document to the search machine 20 can be determined regardless of the size of the document file specified as the search condition. In this case, the transfer load on the search machine 20 is made constant, and the search processing by the file specifying unit search unit 30 in the search machine 20 is stabilized and speeded up.

The text extraction processing unit 36 extracts a text document from the contents of a file specified as a search condition stored in the search specification file storage unit 34. As a format of a document file specified as a search condition in the WWW browser 16, there are various file formats such as an e-mail text file, an HTML file on the Internet, and an Excel file of a summary list. In order to be able to provide a search function with respect to the differences, only the text documents are extracted from the document files of various formats by the text extraction processing unit 36 and used as search conditions.

Subsequently, a morphological analysis unit 38 extracts nouns included in the extracted text document by using morphological analysis. Nouns in the document content extracted by the morphological analysis unit 38 are sent to the keyword creation unit 40, and the keyword creation unit 40 extracts important nouns for keyword creation.

The keyword extraction unit 40 first extracts the important words by counting the number H of occurrences of each noun in the number N of documents registered in the index 52 of the search database 22. Do.

When the document appearance number H in the index 52 is obtained, a word having the appearance number H within a predetermined range, for example, (2N / 3) ≧ H ≧ 1 is selected. Among the words selected in this way, the top 10 words having a large number of appearances H are selected for keyword creation. And the selected important word 1
Create a query expression by connecting 0 items with a logical sum, and execute the search execution unit 4
2 provided.

The search execution section 42 searches the search database 22 based on the query formula given from the keyword creation section 40.
Of the search 52, the index satisfying a predetermined similarity is extracted as a search result, the search result is transmitted to the WWW browser 16 side by the WWW server 18, and the search result is referred to the user in the form of a document list. It can be so.

Further, the document search unit 30 can perform a document search using property information of a file specified as a search condition stored in the search specification file storage unit 34. For this reason, the search condition specifying unit 2 of the WWW browser 16
6 extracts the property information of the specified document file when the document file is specified as the search condition, and sends the property information to the search machine 20 together with the first file portion of the document specified as the search condition, for example, 1 KB of the first file portion. Send.

The document search unit 30 shown in FIG. 14 extracts text sentences from file contents, extracts nouns by morphological analysis, creates keywords by selecting important words for nouns, and stores them in a search designation file storage unit 34. For example, a creation date, a creator, a title, and the like are extracted from the property information added to the contents of the file, and the keyword creation unit 40 includes the property information in the keyword, and the search execution unit 42 searches the index 52 of the search database 22. I do.

FIG. 5 is an explanatory diagram of index creation processing by the search database creation unit 28 provided in the search machine 20 of FIG. This search database creation unit 28
, The robot 54 is connected to the external document database 4
Collect documents 66 from 6,50 and create temporary file 6
2 and a list of collected documents 66 is added to the collected document list file 64 at the same time.

Subsequently, the robot 54 sets the text extraction unit 56
The text extracting unit 56 extracts the collected document 66 from the collected document list file 64 and stores it in the extracted text file 68.

Next, the process is passed to an important word extraction unit 58, which extracts nouns from the corresponding text document of the extracted text file 68 by morphological analysis, counts the appearance frequency of each noun, and The top 10 words that appear frequently are extracted as important words and stored in the important word file 70.

Next, the process is passed to the index creation unit 60, and the index creation unit 60 extracts a set of, for example, the top 10 important words for the document from the keyword file 70, and further adds the document name and the storage location to the index. Is created and stored in the search database 22 as index information.

FIG. 6 is a flowchart of a browser process for specifying search conditions and displaying search results by the WWW browser 16 of FIG. When the user opens the search function of the WWW browser 16, a search screen is displayed in step S1, the search screen is displayed, and in step S2, a search condition specifying a document file is performed.

Subsequently, in step S3, the presence / absence of a search start is checked.
Check if it is a file specified search. If the search is a file designation search, the process proceeds to step S5, where a file designated by the user is read, and in step S6, the first 1 KB of the designated file is transmitted to the server together with a search request message.

If it is not a file designation search, step S
At 7, a search request message corresponding to another search, for example, a keyword search, is transmitted to the server. Step S6
When the head portion of the specified file is transmitted to the server in step, the process waits for reception of the search result in step S8.

When the search result is received from the server in step S8, the process proceeds to step S9, where the display result of the search operation is performed, and the user views the search content. Such processing of steps S1 to S9 is repeated until there is a search end instruction to close the search screen in step S10.

FIG. 7 shows a concrete procedure and a screen when a document file is specified as a search condition in the browser processing of FIG.

In FIG. 7, the user first obtains a document file 72 to be specified as a search condition from, for example, the Internet. Then, the user enters the document file 72
The contents of the document file 72 are saved in a file designated in advance, for example, a file “news.txt” in order to perform a document search similar to the document file 72 by looking at the contents of the document file 72.

Subsequently, the user opens the keyword input screen 74. The keyword input screen 74 has a keyword input section 7
6, a file designation section 78, a reference button 80, and a search execution button 82 are provided. Therefore, when the user presses the reference button 80 on the keyword input screen 74, a file selection dialog 84 is displayed.

Since the file selection dialog box 84 stores the document file 72 to be specified as a search condition, the file name is “news.txt”.
Is selected by clicking with the mouse, and the selected file name “ne” is displayed in the file specification section 78 of the keyword input screen 74.
ws. txt ”is set.

When the file designation by the file designation section 78 is completed in this way, by pressing the search execution button 82, the document file "n" designated as the search condition is pressed.
ews. The first 1 KB of the document content of “txt” is transmitted to the server together with the search request.

FIG. 8 is a flowchart of a server search process realized by the document search unit 30 of FIG. In this server search process, a document file specified as a search condition is read in step S1, and a text document is extracted from the document file in step S2. Next, in step S3, nouns are extracted from the contents of the extracted text document using morphological analysis. Next, in step S4, a process of counting the number of occurrences H of the number of documents in the index 52 provided in the search database 22 in which each word extracted as a noun appears in the search database 22 is performed.

If the number of appearances H in the index of each word has been counted, the number of appearances H is calculated as (2N /
3) First, words in a range of 1 or more are selected first, and among the selected words, the top 10 words having a large number of appearances H are selected as important words to be used as keywords. Subsequently, in step S6, a query expression in which the ten words selected as important words are connected by a logical sum is generated.

In step S7, the search of the index of the search database is performed using the query expression generated as the search keyword, and the contents of the index having a predetermined similarity to the generated keyword are summarized in a list as a search document. In S8, the search result is transmitted to the browser.

FIG. 9 shows details of the text extraction process in step S2 of FIG. In this text extraction process, the extension of the document file is decoded in step S1.
If it is recognized from the file extension in step S2 that the document is an HTML document, the process proceeds to step S3, in which the data in the body tag in the HTML document is extracted as the text data body, and the tag data is removed.

For example, taking an HTML file as shown in FIG. 10A as an example, the data in the body word sandwiched between <> is extracted as the text data body, and this tag data is removed. An extracted text document as shown in (B) is obtained.

Next, in step S4, the property information of the file managed by the OS is obtained. This property information includes, for example, the file owner and the document type.

FIG. 11 shows an example of property information of a document file obtained from the Internet. The property information includes a document title “about the document management system”, a creation date, a change date, and the like. Acquire data for keyword generation.

On the other hand, if the document is not an HTML document but an Excel document, for example, in step S2, the file is transferred to the document library in step S5 to acquire text data. Subsequently, in step S6, file property information, such as a creator and a document title, set for each document is obtained by a property information obtaining function.

FIG. 12 shows an Excel file as a file other than the HTML file designated as a search condition in the present invention. When the text file is obtained by passing the Excel file of FIG. 12 to the document library, a result obtained by extracting the text document written in the Excel document as shown in the extracted text document of FIG. 13 is obtained.

The H obtained by such a text extraction process
A text document from a TML document or an Excel document, and a text document obtained from property information are put together, and a noun is extracted using morphological analysis in step S3 of FIG. 8, and a database index is extracted in steps S4 and S5. , A query formula is created by selecting the top 10 important words as keywords, and a search result can be obtained by performing an index search of the database.

The property information acquisition in steps S4 and S6 in the text extraction processing in FIG. 9 can be selected by the user on the WWW browser 16 as to whether or not to use the property information. Whether to use it depends on the user's judgment on how narrow the search results should be.

The present invention also provides a computer-readable recording medium in which a document information search program for causing the search machine 20 of FIG. 4 to execute the processing function of the document search unit 30 is recorded. As an embodiment of this recording medium, a CD-R
It includes a removable portable recording medium such as an OM or a floppy disk, a storage device of a program provider that provides a program via a line, and a memory device such as a RAM or a hard disk of a processing device in which the program is installed.

A document information search program for realizing the function of the document search unit 30 shown in FIG. 4 provided by the recording medium, more specifically, a document information search program having steps for executing the processing in the flowcharts shown in FIGS. The program is
It is loaded into a processing device such as a server and executed on its main memory.

The document information retrieval program of the present invention loaded on the server side, upon receiving a service request from the client side, uploads a WWW browser function for designating a retrieval condition by specifying a file to the client 12 side. Enable the use of a search system.

Although the above embodiment has been described with reference to the server-client type search system as an example, the present invention is not limited to this, and may take the form of a host terminal or an appropriate system. In addition, the present invention is not limited to the above-described embodiments, and includes appropriate modifications that do not impair the objects and advantages thereof. Furthermore, the present invention is not limited by the numerical values shown in the above embodiments.

(Supplementary Note 1) In a document information search apparatus that searches for and responds to document information based on a search request via a network, a file is specified as a search condition as a search request source, and the content of the specified file is specified. A search condition specifying unit for transmitting the data via the network; and a search unit for generating a keyword from the file content transmitted from the search condition specifying unit and searching the database for a similar document. Characteristic document information search device.
(1)

(Supplementary Note 2) In the document information search device according to Supplementary Note 1, the search condition specifying unit transmits a first file portion of the specified file content.

(Supplementary Note 3) The document information search device according to Supplementary Note 1, wherein the search condition specifying unit includes an HTML file and an Excel file in files specified as search conditions.

(Supplementary Note 4) In the document information search device according to Supplementary Note 1, the database stores, for each document, index information that lists important words extracted from the search target document, and the document search unit of the server includes: A text extraction processing unit that extracts a text sentence from the file content received along with the search request; a morpheme analysis unit that extracts a noun by morphological analysis of the text sentence; Document information, comprising: a keyword generation unit that generates a keyword connected by a sum; and a search execution unit that searches for a similar document by searching a search database using the keyword and notifies a client of a search result. Search device. (2)

(Supplementary Note 5) In the document information search device according to Supplementary Note 4, the keyword generation unit may determine the number of occurrences of each document in the index of each search document stored in the document database. A document information search apparatus, which counts a number of words and selects a predetermined number of upper words having a number of occurrences within a predetermined range to generate a keyword. (3)

(Supplementary Note 6) In the document information search device according to Supplementary Note 5, when the number of documents in the index is N, the keyword generation unit may include an appearance number H in the range of 2N / 3 ≧ H ≧ 1 A document information search apparatus characterized by selecting top 10 words having a number and generating a keyword.
(4)

(Supplementary Note 7) In the document information search device according to Supplementary Note 5, the keyword generation unit may perform a search by including property information extracted from a file received along with the search request in the keyword. Document information retrieval device. (5)

(Supplementary Note 8) In the document information search device according to supplementary note 7, the property information is a creator of a file, a document title, and the like received in response to the search request. .

(Supplementary Note 9) In the document information search device according to Supplementary Note 1, the search condition specifying unit of the search request source is provided by a WWW browser of a client, and a file content specified on a search request screen of the WWW browser is read A document information retrieval apparatus, which transmits the document information to a retrieval machine of a WWW server via a network and delivers the document to the document retrieval unit.

(Supplementary Note 10) A database storing index information listing important words extracted from a search target document for each document, and a file content received by a search request from a network specifying a document file as a search condition. A text extraction processing unit for extracting a text sentence,
A morphological analysis unit that extracts a noun by morphological analysis of the text sentence, and a keyword generation unit that extracts an important word from the noun and generates a keyword connected by OR.
A document information search device, comprising: a search execution unit that searches for a similar document by searching the database using the keyword and notifies a search result to a request source.
(6)

(Supplementary note 11) In the document information search device according to supplementary note 10, the keyword generation unit may determine the number of occurrences of each document in the index of each search document stored in the document database. A document information search apparatus, which counts a number of words and selects a predetermined number of upper words having a number of occurrences within a predetermined range to generate a keyword.

(Supplementary note 12) In the document information search device according to supplementary note 10, the database stores the index information and the property information extracted from the search target document, and the keyword generation unit stores the file received in response to the search request. A document information search device for searching by including the property information extracted from the keyword in the keyword.
(7)

(Supplementary Note 13) In a document information retrieval method for retrieving and responding to document information based on a retrieval request via a network, index information listing key words extracted from a retrieval target document is stored in a database for each document. When a file is specified in the search conditions at the search request source, the specified file content is transmitted to the server via the network together with the search request, and the search side receives the file content from the file content received with the search request. A text sentence is extracted and a noun is extracted by morphological analysis of the text sentence. Then, an important word is extracted from the noun to generate a keyword connected by a logical sum, and a similar document is searched by searching the database using the keyword. A document information search method characterized by performing a search and responding a search result. (8)

(Supplementary Note 14) In the document information search method according to supplementary note 13, as the generation of the keyword, the number of appearances of each document in the index of each document stored in the database is counted. A document information search method, wherein a keyword is generated by selecting a predetermined number of upper words having the number of appearances in a predetermined range.

(Supplementary note 15) The document information search method according to supplementary note 14, characterized in that property information extracted from a file received in response to the search request is included in the keyword and searched. (9)

(Supplementary Note 16) A step of receiving a search request specifying a document file as a search condition, a step of extracting a text sentence from the content of the file received in response to the search request, and a step of extracting a noun by morphological analysis of the text sentence Performing a keyword extraction from the noun to generate a keyword connected by logical OR, and searching a database based on the keyword for a similar document and notifying the request source of the search result. A computer-readable recording medium storing a document information search program comprising: (10)

(Supplementary Note 17) In the recording medium according to Supplementary Note 16, the step of generating a keyword of the document information search program may include determining which document in the index of each document stored in the database for each noun. A recording medium characterized by counting the number of occurrences of a word, and selecting a predetermined number of upper words having a number of occurrences within a predetermined range to generate a keyword.

(Supplementary Note 18) In the recording medium according to Supplementary Note 14, the document information search program further includes a step of searching the property information extracted from the file received in response to the search request by including the property information in the keyword. A recording medium characterized by the above-mentioned.

(Supplementary Note 19) A step of receiving a search request in which a document file is specified as a search condition in a computer, a step of extracting a text sentence from the file content received along with the search request, and a morphological analysis of the text sentence A step of extracting a noun, a step of extracting a keyword from the noun to generate a keyword connected by a logical sum, and a step of searching for a similar document by searching the database using the keyword and notifying the request source of the search result. And a step of executing the document information search program. (11)

[0098]

As described above, according to the present invention, when a user obtains a document containing contents of interest via e-mail, the Internet, or the like, a document search similar to this document is performed in a document file. Is directly specified as a search condition, it is possible to easily and quickly search for documents with similar contents, eliminating the need to enter a keyword based on the time-consuming document contents, and making it extremely efficient for the user to search for similar documents. Can be realized.

In addition, in generating a keyword necessary for a document search by specifying a file, in addition to extracting an important word from the document content, an important word is also extracted from property information of the document file and included in the keyword. Thus, the similarity search of the existing document registered in the database can be narrowed down more appropriately, and the accuracy of the search can be improved.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating the principle of the present invention.

FIG. 2 is an explanatory diagram of a system configuration of the present invention.

FIG. 3 is a block diagram of a functional configuration of the present invention.

FIG. 4 is a block diagram of a document search unit according to the present invention.

FIG. 5 is an explanatory diagram of a process performed by a search database creation unit in FIG. 3;

FIG. 6 is a flowchart of a browser process of FIG. 3;

FIG. 7 is an explanatory diagram of a search request operation for designating a document file as a search condition according to the present invention.

FIG. 8 is a flowchart of a server search process according to the present invention.

FIG. 9 is a flowchart of a text extraction process in FIG. 8;

FIG. 10 is an explanatory diagram for extracting a text document from an HTML file by the processing of FIG. 8;

FIG. 11 is an explanatory diagram of property information provided in an HTML file used for a search according to the present invention.

12 is a diagram illustrating an example of a text extraction target E by the processing of FIG.
Illustration of xcel document

13 is an explanatory diagram of a text document extracted from the Excel document of FIG.

[Explanation of symbols]

 10: server 12: client 14: Internet / intranet 16: WWW browser 18: WWW server 20: search machine 22: search database 24, 46, 50: document database 25: search target document 26: search condition specification unit 28: search database Creation unit 30: Document search unit 32: Document reference unit 34: Search designation file storage unit 36: Text extraction processing unit 38: Morphological analysis unit 40: Keyword creation unit 42: Search execution unit 44, 48: Document management server 54: Robot 56: Text extraction unit 58: Key word extraction unit 60: Index creation unit 62: Temporary file 64: Collected document list file 66: Document 68: Extracted text file 70: Key word file

Claims (11)

[Claims] 1. A document information retrieval apparatus which retrieves and responds to document information based on a retrieval request via a network, wherein a file is designated as a retrieval condition to the retrieval request source.
A search condition specifying unit for transmitting the specified file content via the network is provided, and a search machine for generating a keyword from the file content transmitted from the search condition specifying unit and searching for a similar document from the database is provided on the search side. A document information search device characterized by being provided. 2. The document information search device according to claim 1, wherein the database stores, for each document, index information listing important words extracted from the search target document, and the search machine responds to the search request. A text extraction processing unit for extracting a text sentence from the file content received therewith; a morphological analysis unit for extracting a noun by morphological analysis of the text sentence; an important word extracted from the noun and connected by a logical sum A document information search device, comprising: a keyword generation unit that generates a keyword; and a search execution unit that searches for a similar document by searching the search database using the keyword and notifies a client of a search result. 3. The document information search device according to claim 2, wherein the keyword generation unit determines the number of occurrences of each noun in an index for each search document stored in the document database. A document information search device which counts and selects a predetermined number of upper words having a number of occurrences within a predetermined range to generate a keyword. 4. The document information search device according to claim 3, wherein the keyword generation unit sets the number of occurrences H in the range of 2N / 3 ≧ H ≧ 1 where N is the number of documents in the index. A document information search apparatus characterized in that a keyword is generated by selecting the top 10 words having the following. 5. The document information retrieval apparatus according to claim 3, wherein the keyword generation unit causes the keyword to include property information extracted from a file received in response to a search request and search the document. Information retrieval device. 6. A database storing index information listing key words extracted from a search target document for each document, and a search request from a network that specifies a document file not registered in the search database as a search condition. A text extraction processing unit for extracting a text sentence from the file content received by the above, a morphological analysis unit for extracting a noun by morphological analysis of the text sentence, and a keyword obtained by extracting important words from the noun and connecting them by logical OR And a search execution unit that searches for a similar document by searching the database using the keyword and notifies a search result to a request source. 7. The document information search apparatus according to claim 6, wherein said keyword generation unit searches for the property information extracted from the file received in response to the search request by including the property information in the keyword. Information retrieval device. 8. A document information retrieval method for retrieving and responding to document information based on a retrieval request via a network, wherein index information listing important words extracted from a retrieval target document is stored in a database for each document. When a file is specified in the search conditions at the search request source, the specified file content is transmitted to the search destination via the network together with the search request, and the search side converts the text from the file content received with the search request to text. Extract a sentence and extract a noun by morphological analysis of the text sentence, then extract key words from the noun, generate keywords connected by logical OR, and search for similar documents by searching the database using the keywords And responding a search result. 9. The document information search method according to claim 8, wherein property information extracted from a file received along with the search request is included in the keyword and searched. 10. A step of receiving a search request specifying a document file as a search condition, a step of extracting a text sentence from the file content received with the search request, and a step of extracting a noun by morphological analysis of the text sentence Extracting key words from the nouns to generate keywords connected by logical OR; searching a database based on the keywords for similar documents and notifying the requester of the search result;
A computer-readable recording medium storing a document information search program comprising: 11. A computer, comprising: a step of receiving a search request in which a document file is specified as a search condition; a step of extracting a text sentence from the content of the file received in accordance with the search request; Extracting, extracting key words from the nouns to generate keywords connected by logical OR, searching a database based on the keywords to search for similar documents and notifying the request source of the search result When,
A document information search program characterized by executing the following.