JP2004206468A - Document management system and document management program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a document management system capable of saving labor of input work of document type information, and of flexibly sorting in accordance with a trend of stored document information. <P>SOLUTION: Keywords are extracted from new document information converted into text data by an OCR 2 by a keyword extracting part 20, extracted keywords of existing document information are read from a document information DB in a storage 7, both are compared, and document type information of existing document information with the most number of matching keywords is set as a document type of the new document information. <P>COPYRIGHT: (C)2004,JPO&NCIPI

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a document management system and a document management program for organizing and storing digitized document information.
[0002]
[Prior art]
2. Description of the Related Art In recent years, with the progress of computerization in society, various documents have been exchanged in large quantities, and a document management system for efficiently organizing and storing these various documents has been demanded. In a document management system, digitized document information is stored through a network system, or a document output on paper is subjected to OCR processing and stored as digitized document information. The document information is stored as electronic information. Makes it possible to store a large number of documents. However, there is a problem how to efficiently extract necessary document information from a large amount of stored document information.
[0003]
Generally, bibliographic information such as a document name, a registration date, and a creator is added to the stored document information. However, such bibliographic information alone has large variations in information and cannot be searched efficiently. Therefore, if the document type for classifying the document information is determined in advance, and the document information is classified in accordance with the document type, and the document information is searched based on the classification at the time of the search, it is possible to narrow down more efficiently. . However, it is troublesome to input information on the document type one by one when storing the document information, and there is a possibility of an input error.
[0004]
Therefore, a method of automatically classifying the stored document information has been proposed. For example, Patent Document 1 describes that a character profile and an image attribute of an electronic document are analyzed to construct a document profile, and the document profile is classified and stored in an existing directory based on the document profile. I have. Patent Document 2 discloses that a document is converted into a formatted document by media conversion, and a content feature and a format feature are extracted from the formatted document. From these features, a content feature and a format feature of each category are extracted. It describes that features are determined and compared with content features and formal features extracted from a document to be classified to perform classification. Further, in Patent Document 3, one or more keywords are registered in advance, the target document is divided into unit articles, and whether or not the registered keywords are included in the unit articles is checked. Describes a document processing apparatus that organizes and saves in a predetermined article storage area provided for each keyword.
[0005]
[Patent Document 1]
JP 2000-112993 A [Patent Document 2]
Japanese Patent Application Laid-Open No. 2000-268040 [Patent Document 3]
JP 2001-109772 A
[Problems to be solved by the invention]
Although the method of automatically classifying document information as described above has an advantage in that manual input is not required, information for specifying the classification is determined in advance, and the document information is determined based on the specific information. Since the information is classified, how the specific information is determined first determines whether the subsequent search for the document information is performed efficiently. However, the search needs for document information change over time, and it is difficult to reflect that in the classification from the beginning. Therefore, it is necessary to review the classification. The prior art does not sufficiently address these problems.
[0007]
SUMMARY OF THE INVENTION It is an object of the present invention to provide a document management system capable of saving labor of inputting document type information and flexibly classifying according to the tendency of stored document information. is there.
[0008]
[Means for Solving the Problems]
A document management system according to the present invention includes an extracting unit that extracts document specifying information from input document information, a specifying information storage unit that stores the extracted document specifying information, and a document information generating unit that is generated from the stored existing document information. Discrimination information storage means for storing discrimination information for discriminating the document type, and document type information of the new document information based on the document identification information of the newly input new document information and the discrimination information read from the discrimination information storage means And document type determining means for setting Further, the extracting means extracts a keyword from the document information as the document specifying information, the discriminating information storing means stores an extracted keyword for each existing document information as the discriminating information, Document type information is set by comparing the extracted keyword of the document information with the extracted keyword for each existing document information. Further, the extracting means extracts important sentence data from the document information as document specifying information, and the discriminating information storage means stores document type specifying information for each document type generated from existing document information as discriminating information. Preferably, the document type determining means sets the document type information by comparing the important sentence data of the new document information with the document type identification information for each document type. Further, the extraction means extracts a keyword from the document information as document identification information, and the discrimination information storage means stores the discrimination information as word book data for each document type generated from existing document information, The document type determining means sets the document type information by comparing the extracted keyword of the new document information with the word book data for each document type. Further, the discrimination information storage means stores a plurality of types of discrimination information, and comprises means for setting a priority of discrimination information used when the document type discrimination means discriminates a document type. And Further, the apparatus further comprises means for setting the same additional information as the additional information relating to the existing document information of the same type based on the document type information set by the document type discriminating means for the new document information. .
[0009]
In addition, the document management program according to the present invention includes: a computer for managing document information; a unit for extracting a keyword from newly input new document information; a unit for reading a keyword from stored existing document information; And a means for setting the document type information of the existing document as the document type information of the new document information based on the comparison result.
[0010]
Further, another document management program includes a computer for managing document information, a means for extracting keywords and important sentence data from newly input new document information, a means for reading keywords from stored existing document information, Means for comparing the extracted keyword of the document information with the keyword of the existing document information, means for reading the stored document type identification information for each document type, means for comparing the important sentence data with the document type identification information, and comparison It is characterized by functioning as a means for setting the document type information of the existing document as the document type information of the new document information based on the result.
[0011]
Further, another document management program includes a computer for managing document information, a means for extracting keywords and important sentence data from newly input new document information, a means for reading keywords from stored existing document information, Means for comparing the extracted keyword of the document information with the keyword of the existing document information, means for reading the stored document type identification information for each document type, means for comparing the important sentence data with the document type identification information, Means for reading out the word book data for each document type, means for comparing the extracted keyword of new document information with the word book data for each document type, and the document type information of the existing document based on the comparison result as a document of the new document information. It is characterized by functioning as means for setting as type information. Further, the document management program sets the additional information of the existing document as the additional information of the new document information together with the document type information.
[0012]
Further, a recording medium according to the present invention comprises a computer-readable recording medium recording the document management program.
[0013]
With the above configuration, the document type information of the new document information is automatically determined according to how the document type information is set in the existing document information. Since the document type information is set based on the discrimination information generated from the existing document information stored one after another, the document type information is always set according to the latest tendency. When document information that does not fit in the existing document type is input, if the document type is manually set, the subsequent document information is stored as the document type information of the existing document information. Since the document type information is automatically set for the document information, the setting of the document type information can be performed with more flexibility.
[0014]
Then, by extracting a keyword from the new document information and using the extracted keyword of the existing document information as the discrimination information, the document type information of the existing document information including the same keyword as the new document information is used for the document type discrimination. Can be.
[0015]
Also, by extracting important sentence data from new document information and using document type identification information for each document type generated from existing document information as discrimination information as discrimination information, the document type discrimination can be performed more efficiently and accurately. Can be done.
[0016]
Further, by extracting a keyword from the new document information and using the word book data for each document type generated from the existing document information as the discrimination information, it can be performed more efficiently and accurately. In addition, when performing the document discrimination based on the above-described discrimination information, by setting the priority order of which discrimination information to use in order, it is possible to set the document type information according to the needs of the user. Can be.
[0017]
In addition, since additional information about existing document information of the same type of new document information can often be adopted as additional information of new document information, if it is set as it is as additional information of new document information, manual input work can be performed. Labor can be saved.
[0018]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments according to the present invention will be described in detail. FIG. 1 shows an outline of the configuration of a document management system according to the present invention. The system main body 1 is connected to a network via an interface 11, and transmits and receives information such as document information to and from a terminal that creates a document. Further, it is connected to peripheral devices such as the OCR 2 and the printer 3 via the interface 12. Text data is generated from the document from the OCR 2 and transmitted to the system main body 1.
[0019]
The system body 1 includes an input device 5 for inputting necessary data, a display 6 for displaying information, a storage device 7 for storing information such as document information, an information processing device 8 for performing information processing, and an information processing device 8. The system includes a RAM 9 for temporarily storing information to be processed and a ROM 10 for storing a program and the like for functioning as a document management system according to the present invention. The program stored in the ROM 10 is recorded on a computer-readable recording medium such as a CD-ROM, an FD, and an MO, and is installed in the computer as needed.
[0020]
For example, when text data is generated from a document in the OCR 2 and new document information is transmitted to the system main body 1, the information processing device 8 temporarily stores the new document information in the RAM 9, A keyword is extracted from the new document information by the method. Document information on existing documents is stored in the storage device 7, keywords are extracted from the existing document information by the keyword extraction unit 20, and the keywords of the new document information and the existing document information are extracted by the keyword comparison unit 22. Keywords are compared. As shown in FIG. 2, the comparison between the two keywords is performed by extracting keywords such as “network”, “technology”,... Extracted from the stored existing document information 1, 2,. The keywords such as “network”, “method”,... Are compared. Then, the number of matching of both keywords is checked by the document type discriminating unit 25, the document type information of the existing document information having a large number of matching is read out, and the document type information is set as the document type information of the new document information. I do. In this case, the level of the reference of the number of coincidences may be appropriately set according to the similarity of the stored document information. For example, if the similarity is high, the number of matching existing documents may be increased to reduce the number of matching existing documents. If the similarity is low, the number of matching existing documents may be reduced to reduce the number of matching existing documents to some extent. Make it hit.
[0021]
In this way, the existing document information whose keyword matches the new document information is searched, and when these document types are read out by the document type discrimination unit 25, if they are all one document type, they are set as the document type of the new document information as they are. Although it is possible, when two or more document types are read, it is necessary to perform weighting to narrow down to one document type. Means for performing weighting include the following.
[0022]
First of all, there is a method based on important sentence data. Important sentence data is a sentence corresponding to “headline”, “proper word / phrase”, assertion, question, etc., which is a part where the characteristics of the document type are noticeable in the document structure. Analyzing the data narrows down the document type. Specifically, the important sentence data relating to the new document information is extracted by the important sentence extracting unit 21 from the text data generated by the OCR2. FIG. 3 shows an example of the extraction. Here, new document information related to “information on network testing” is given as an example, but “information on network testing”, “date and time ...”, “location ...” Is extracted as an important sentence.
[0023]
In the case of this example, “test guide”, “newspaper article”, and “patent” have already been set as the document types, and for each document type, “specific heading”, “specific word / phrase” , "Semi-specific words / phrases", etc., data indicating the specific expressions are set, and data on the possibility of matching the document type from the frequency of use of assertions, questions, and estimated sentences (FIG. 3) Are shown in three stages). Information for specifying such a document type is stored in advance in the “document structure DB”, the “specific heading DB”, and the “unique word / phrase DB” in the storage unit 7. Based on such document identification information, for example, if there is a unique word “test”, it is judged that the possibility of the document type of “test guide” is high, and if there is a heading of “claims”, it is called “patent”. It can be determined that the possibility of the document type is high. The document type specifying information stored in these DBs may be set in advance based on the existing document information, or may be appropriately reviewed as the existing document information is accumulated.
[0024]
The important sentence data extracted from the new document information has its document structure analyzed by the information stored in the document structure DB, and “test” is used as a unique word / phrase, and “date / time” / “place” is used as a quasi-specific word / phrase. And there is no assertion, etc., so that the possibility of the document type of “test guide” increases.
[0025]
Next, there is a method based on wordbook data. The word book is generated by previously collecting keywords representing characteristics of the document type based on keywords extracted from existing document information of the same document type. The generated word book is stored in the word book DB 34. If necessary, a keyword can be manually input to the word book from the input device 5. FIG. 4 shows an example of the word book data. In this example, the company name is stored as a word book relating to the document type of “patent”, and the name of a newspaper company is stored as a word book relating to the document type of “newspaper article”.
[0026]
Then, the word book search unit 24 performs a search process for a word book in which a keyword that matches the keyword extracted from the new document information exists. In FIG. 4, the word book is searched for “Riko” as an extracted keyword, and a word book of the “patent” document type in which the same keyword exists is hit. Therefore, the possibility of the "patent" document type increases.
[0027]
In addition to the above-mentioned "important sentence" and "word book data", "document name" of new document information and "memo" created for the document can also be used. This is because the feature of the document is considered to appear in the “document name” and the “memo”. The keyword comparison unit 22 compares the keyword extracted from the “document name” of the new document information and the “memo” created for the document with the existing document information to determine the document type.
[0028]
The order in which weighting is performed by “important sentence”, “word book data”, “document name”, and “memo” may be appropriately set depending on what kind of document information is stored.
[0029]
FIG. 5 shows an example of the above processing flow. As a new document information, a paper document is read by the OCR (S100) and OCR-processed (S101) to generate text data. Then, a keyword is extracted by a known method (S102), and is compared with the extracted keyword of the existing document information to check for the presence of the same keyword (S103). As a result of the comparison, if there is existing document information including the same keyword, the document type information is checked (S104). If there are two or more document types, the next weighting process is performed in accordance with a preset weighting order (S105), and the document type information is checked again.
[0030]
When the weighting process is performed and the document type information is narrowed down to one, the document type information is added as the document type of the new document information (S106). In addition, necessary information such as a document name, a word book, and a memo added to the existing document information is added as new document information (S107). Further, it is checked whether or not the access right to the existing document information is set (S108). If the access right is set, the same access right is added (S109). Further, it is checked whether or not a related document related to the existing document information is stored (S110), and the same related document as the stored related document is added (S111).
[0031]
If the same keyword does not exist in S103, display processing to that effect is performed (S112), and the document type information is manually input (S113).
[0032]
Therefore, manual input is required in the initial stage when the amount of existing document information is small, but as the existing document information is accumulated, the system automatically sets the document type information as the existing document information is accumulated. Further, since the setting is made in accordance with the stored existing document information, the document type information is set by always reflecting the tendency of the stored document information.
[0033]
【The invention's effect】
As described above, according to the present invention, the document type information of the new document information is automatically determined depending on how the document type information is set in the existing document information. Since the document type information is set based on the discrimination information generated from the existing document information stored one after another, the document type information is always set according to the latest tendency. When document information that does not fit in the existing document type is input, if the document type is manually set, the subsequent document information is stored as the document type information of the existing document information, and the subsequent document information is stored. Since the document type information is automatically set for the document information, the setting of the document type information can be performed with more flexibility.
[0034]
Then, by extracting a keyword from the new document information and using the extracted keyword of the existing document information as the discrimination information, the document type information of the existing document information including the same keyword as the new document information is used for the document type discrimination. Can be.
[0035]
Also, by extracting important sentence data from new document information and using document type identification information for each document type generated from existing document information as discrimination information as discrimination information, the document type discrimination can be performed more efficiently and accurately. Can be done.
[0036]
Further, by extracting a keyword from the new document information and using the word book data for each document type generated from the existing document information as the discrimination information, it can be performed more efficiently and accurately. In addition, when performing the document discrimination based on the above-described discrimination information, by setting the priority order of which discrimination information to use in order, it is possible to set the document type information according to the needs of the user. Can be.
[0037]
In addition, since additional information about existing document information of the same type of new document information can often be adopted as additional information of new document information, if it is set as it is as additional information of new document information, manual input work can be performed. Labor can be saved.
[Brief description of the drawings]
FIG. 1 is a block diagram of a device configuration according to an embodiment of the present invention.
FIG. 2 is an explanatory diagram relating to keyword processing in the embodiment according to the present invention.
FIG. 3 is an explanatory diagram relating to important sentence processing in the embodiment according to the present invention.
FIG. 4 is an explanatory diagram related to a word book process in the embodiment according to the present invention.
FIG. 5 is a processing flowchart of the embodiment according to the present invention.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 ... System main body, 2 ... OCR, 3 ... Printer, 4 ... Terminal, 5 ... Input device, 6 ... Display, 7 ... Storage device, 8 ... Information Processing device, 9 RAM, 10 ROM, 11 interface, 12 interface.

Claims (11)

Extraction means for extracting document identification information from input document information, identification information storage means for storing the extracted document identification information, and identification information for identifying a document type generated from the stored existing document information Discriminating information storing means for storing document type information of new document information and document type discriminating means for setting document type information of new document information based on discriminating information read from the discriminating information storing means. A document management system, comprising: The extracting means extracts a keyword from document information as the document specifying information, the discrimination information storage means stores an extracted keyword for each existing document information as the discrimination information, and the document type discriminating means 2. The document management system according to claim 1, wherein the document type information is set by comparing the extracted keyword of the document information with the extracted keyword for each existing document information. The extraction unit extracts important sentence data from document information as the document identification information, and the discrimination information storage unit stores document type specification information for each document type generated from existing document information as the discrimination information. The document according to claim 1, wherein the document type determining unit sets the document type information by comparing the important sentence data of the new document information with the document type specifying information for each document type. Management system. The extraction unit extracts a keyword from document information as the document identification information, and the discrimination information storage unit stores the discrimination information as word book data for each document type generated from existing document information, 4. The document management system according to claim 1, wherein the document type determination unit sets the document type information by comparing the extracted keyword of the new document information with the word book data for each document type. . The discrimination information storage means stores a plurality of types of discrimination information, and includes means for setting a priority of discrimination information used when the document type discrimination means discriminates a document type. The document management system according to claim 1. 2. A method according to claim 1, further comprising the step of setting, based on the document type information set by said document type discriminating means, the same additional information as the additional information relating to the existing document information of the same type. 6. The document management system according to any one of 1 to 5. Computer to manage document information,
Means for extracting a keyword from newly input new document information,
Means for reading a keyword from the stored existing document information,
Means for comparing the extracted keyword of the new document information with the keyword of the existing document information, and means for setting the document type information of the existing document as the document type information of the new document information based on the comparison result;
Document management program to function as. Computer to manage document information,
Means for extracting keywords and important sentence data from newly inputted new document information,
Means for reading a keyword from the stored existing document information,
Means for comparing the extracted keyword of the new document information with the keyword of the existing document information,
Means for reading the stored document type identification information for each document type,
Means for comparing the important sentence data with the document type identification information, and means for setting document type information of an existing document as document type information of new document information based on the comparison result;
Document management program to function as. Computer to manage document information,
Means for extracting keywords and important sentence data from newly inputted new document information,
Means for reading a keyword from the stored existing document information,
Means for comparing the extracted keyword of the new document information with the keyword of the existing document information,
Means for reading the stored document type identification information for each document type,
Means for comparing the important sentence data with the document type identification information,
Means for reading the stored word book data for each document type,
Means for comparing the extracted keyword of the new document information with the word book data for each document type, and means for setting the document type information of the existing document as the document type information of the new document information based on the comparison result;
Document management program to function as. 10. The document management program according to claim 7, wherein the additional information of the existing document is set together with the document type information as the additional information of the new document information. A computer-readable recording medium recording the document management program according to claim 7.

Images