JP2003178055A - Document data relation extracting device and extracting program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a document data relation extracting device capable of easily acquiring the information on the relation between documents by various procedures, by accumulating the relation information between the documents. <P>SOLUTION: This device is provided with a document data managing function 103 for managing the registration of a document, the reference and the acquisition of the differential information of the document on the basis of the instruction from a terminal, a document database 110 for storing the document information specifying the registered document and the document data indicating the contents of the document in a state of being related to each other, a system definition information defining function 104 for defining the relation extracting condition for extracting the relation of the stored document, and the system definition information for defining the change and the elimination of the extracting condition, a relation information extracting function 105 extracting the information indicating the relation between the documents from the document stored in the document database on the basis of the system definition information, and a display data creating function 106 for displaying the information indicating the relation between the extracted documents. <P>COPYRIGHT: (C)2003,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a document data relevance extracting apparatus and the like, and more particularly to a document data relevance extracting apparatus and the like for extracting relevance information of document data accumulated in a database (DB).

[0002]

2. Description of the Related Art The research of one event or matter is often not completed unless a plurality of documents such as research papers, specifications, instruction manuals, and official documents are read. This is because there is a relationship between the documents, and maintaining the relationship is one of the issues in document production. When revising a document, make sure that there is no inconsistency in the related information in multiple documents.For example, if you delete one information, modify the same information in another document, or rewrite another information. Work such as deleting related information in other documents is required.

It is not uncommon for one event or matter to be distributed and written in a plurality of documents, and the relevance between documents is not always indicated only by reference instructions or citations. As a result, when investigating the effects of document revisions,
It is necessary to read through all the documents that are likely to be related and check them, or to extract them using a general-purpose full-text search function, which increases the work load.

Originally, a document is written or has some relation with other documents. For example, a reference instruction or a citation is used to clearly indicate the relevance, and link information corresponds to a tagged document such as an HTML document. Also, focusing on the fact that the same terms are used for related items, it can be assumed that the same terms imply that they are related.

As a technique focusing on the relation between documents,
For example, Japanese Patent Laid-Open No. 9-146968 or Japanese Patent Laid-Open No. 20
00-99543 and the like. JP-A-9-146968
The issue shows that the title of the reference is manually entered when the document is registered, compared with the title in the document management information, and if there is a matching document, the result is saved in the related document table. . Japanese Patent Laid-Open No. 2000-99543 discloses
After converting the document data into an XML-formatted tagged document and extracting the nodes on the document structure including the character strings indicating the references,
It is shown that the relevance of documents is grasped by using the method of extracting the title of the reference document from the extracted information.

All of these techniques are based on the premise that the document includes a character string that clearly indicates the relevance such as a reference document or a related document. That is, the related documents are searched by focusing on the document names and searching for the reference relationships between the documents in the order in which they are referenced.

[0007]

Each of the techniques disclosed in the above publications only deals with the specified relevance information such as reference instructions and citation information, and between documents that do not contain such information. It is difficult to understand the relevance.

In order to find all the relationships between documents without omission, as described above, the user can repeatedly enter the related terms manually by using the general-purpose full-text search function, or visually. Only the method of confirmation can deal with the reality.

The present invention has been made in view of these circumstances. By accumulating relevance information between documents without distinction between explicit and implicit, information according to relevance between documents can be easily obtained by various procedures. The present invention provides a relevance extraction device for document data that can be acquired.

[0010]

The present invention adopts the following means in order to solve the above problems.

A document data management function for managing registration and reference of a document and acquisition of difference information of the document based on an instruction from the terminal, and document information for specifying the registered document and content of the document. A document database that stores document data in association with each other, and a system definition information definition function that defines a relationship extraction condition for extracting the relationship between the stored documents and system definition information that defines change or deletion of the extraction condition And a function of extracting relevance information for extracting information indicating relevance between documents from documents stored in the document database based on the system definition information, and displaying information indicating relevance between extracted documents It has a function to generate display data.

[0012]

DETAILED DESCRIPTION OF THE INVENTION Embodiments of the present invention will be described below with reference to the accompanying drawings. FIG. 1 is a diagram showing a document data relevance extraction apparatus according to an embodiment of the present invention. In the figure, a terminal 101 is connected to a document data relevance extraction apparatus 102, and a user operates the document data relevance extraction apparatus 102 via this terminal 101. The constituent elements of the document data relevance extraction device 102 are a document data management function 103 and a system definition information definition function 10.
4. Relevance information extraction function 105, display data generation function 106, system definition information DB 107, document DB 11
0 and the relevance DB 113.

Document data management function 10 from the terminal 101
3, the document data management function 103 receives from the terminal 101 the document information 111 and the document data 112 indicating the document name, creator, etc. It is stored in the DB 110.
When deleting a document, the corresponding document data 112 and document information 111 are deleted from the document DB 110. Therefore, the document DB 110 stores the document information 111 and the document data 112 as a set. Document DB1
When an instruction to newly register or delete document data is received from the terminal 101, the document data management function 103 causes the relationship information extraction function 105 to start extraction of relationship information between documents. Instruct. Furthermore, terminal 1
01, the document information 111 and the document data 112 are taken out from the document DB 110 as the information necessary to realize the instruction, and the difference information between the old and new document data is acquired as necessary to the relevance information extraction function 105. hand over.

The information passed by the document data management function 103 to the relevance information extraction function 105 is the document information 111 and the document data 11 to be newly registered when the document is newly registered.
2, the difference information is obtained by comparing the old and new document data 112 at the time of revision and the old and new document information 111, and at the time of deletion, the document information 111 of the corresponding document data 112. Note that the document DB 110 only needs to have the functions of registering, deleting, and referring to the document information 111 and the document data 112 in accordance with the instructions of the document data management function 103. Is not specified.

When the document data management function 103 gives an instruction to start the extraction of the related information, the related information extraction function 105 stores the information received from the document data management function 103 and the system definition information 107. By referring to the existing sentence pattern dictionary 108, the information for identifying the document data of the other party is extracted from the character string indicating the reference or the citation. Further, referring to the term division designation 109, a term that matches the term division of interest is extracted. Next, the extracted information is stored in the relevance DB 113 as the relevance information 114. In addition, a process for revision is performed at the time of revision, and a process for deletion is performed at the time of deletion.

The text pattern dictionary 108 stored in the system definition information 107 is for extracting information for specifying the document data of the other party from a character string indicating reference or citation. The term category designation 109 designates the term category of interest in the character string used in the document data. By these designations, it is possible to determine what type of term is extracted from the document data.

The information of the system definition information 107 can be defined by using the system definition information definition function 104 when constructing the system. This definition includes what kind of document is registered in the document DB 110, what kind of character string is treated as a character string indicating reference or citation, and what term is used as a key to extract the relevance of the document. You can consider and decide. If the information in the sentence pattern dictionary 108 or the term classification designation 109 is changed after the document is registered in the document DB 110, the relevance information extraction function 105 causes the document DB 11 according to the changed definition.
The relevance extraction processing is performed again for all the documents registered in 0, and the relevance information 114 accumulated in the relevance DB 113 is collectively updated. This processing is performed because the information stored in the system definition information 107 is a condition for extracting the relevance between documents.

The relevance information 114 stored in the relevance DB 113 is also processing data of the display data generation function 106. The relevance information 114 includes information for identifying the other party document extracted by the relevance information extracting function 105, terms used in the document data of the other party document and their appearance position information, and the extracted information. The mapping information and the like are included.

The display data generation function 106 is used by the terminal 10
According to the instruction from 1, the relevance information is extracted from the relevance DB 113, the extracted information is converted into displayable data, and then displayed on the terminal 101.

FIG. 2 is a flowchart showing the processing performed by the relevance information extraction function 105. When the processing is started, first, distribution is performed according to an instruction from the terminal 101 ((step 201).

When the instruction from the terminal 101 is new registration, first, the document data management function 103 causes the document information 1
11 and the document data 112 (step 20)
2) Next, the sentence pattern dictionary 108 and the term classification designation 109 defined in the system definition information 107 are received (step 203). Next, a character string is extracted using the received information (step 204). In the extraction of the character string, the document data is analyzed according to the sentence pattern dictionary 108, and the information for identifying the document data of the reference destination is extracted from the character string indicating the reference or the citation. Next, according to the term classification designation 109, terms (character strings) used in the document data are extracted. The extracted information is once stored in the relevance DB 113 and then subjected to matching processing with existing information to add information indicating relevance between documents.

When the instruction from the terminal 101 is a revision, first, in addition to the document information 111 and the document data 112, the difference information between the old and new document data is received from the document data management function 103 (step 206). The difference in processing from the case of the new registration lies in the handling of the difference information. The difference information is information indicating the difference between the old version and the new version, and is used to correct the existing relevance information 114 stored in the relevance DB 113. Next, the sentence pattern dictionary 108 and the term classification designation 109 defined in the system definition information 107 are received (step 207). Extract the character string using the received information. To extract a character string, the document data is analyzed with reference to the difference information, and information for identifying a reference destination or a term used on the document data is extracted from the character string indicating reference or citation. (Step 208). Then, in the update 209 of the existing information, the relevance information 1 based on the extracted information.
Modify the information stored in 14 (step 20
9) Then, the change in the relevance between the documents due to the revision is reflected by the matching process with the existing information (step 21).
0).

When the instruction from the terminal 101 is deletion, first, the document information 111 is received from the document data management function 103 (step 211), and then all the information related to the corresponding document data 112 is deleted from the relevance DB 113. (Step 212).

By the above three processes, the relation DB1
The accuracy of the relevance information 114 stored in 13 can be secured. Note that the form of the information stored in the relevance information 114 and how it is updated will be described below.

FIG. 3 is a schematic diagram showing how the relevance information 114 is stored at the time of new registration.

The related information 114 includes the document data 112.
Management table 302 and reference / citation table 30 for each
3, term table 304, and related document information 305 are prepared. This is also the case when the document data is revised, and the related information of the old version and the new version is stored respectively.
As a result, the relationship between the old documents is maintained.

In FIG. 3, the relevance information 11 of the document data A is shown.
4 is shown as a data group 301 of the document data A.
The document name (document data A) is registered in the first column of the management table 302. A link indicating the correspondence with the document data stored in the document DB 110 is stored in the second column,
A link pointing to the old version of the relevance information 114 is stored in the third column, and a reference / citation table of the corresponding document data, a term table, and a link to the related document information are stored in the fourth column. From this management table 302, the relationship with the document DB 110, the relationship with the old version of the relationship information 114, and the range of information that constitutes the relationship information 114 for certain document data can be specified. Since FIG. 3 is information at the time of new registration, the link to the old version relevance information in the third column of the management table 302 has a null value.

The first column of the reference / quotation table 303 stores information for specifying the document data of the other party extracted from the character string indicating the reference or the citation based on the information of the sentence pattern dictionary 108, and the second column of the first column. The type of information indicated by the character string of is stored. In this case, the "document name" corresponds to it, and it is possible to match it with the information type of the document information 111. Regarding the relationship with the document information 111, FIG.
Will be described in detail in. The third column stores a link to the relevance information of another document found from the information in the first column. If the referenced document data cannot be found, the third column has a null value.

Further, in order to confirm whether or not the newly registered document is referenced, the reference / citation table of another document is checked, and if it is referenced, the third column of the reference / citation table of the other party is checked. Add a link to the relevance information of the newly registered document. This preserves explicit document relationships such as references and citations.

The first column of the term table 304 stores the terms extracted from the document data based on the information of the term classification designation 109, the second column stores the attributes indicating new, added, changed, and deleted, and the third column. The original term when the term is changed is stored in, and appearance position information indicating where the term stored in the first column is used in the document data is stored in the fourth column. As the appearance position information, information for identifying the position such as the number of characters up to the appearance position, page number, etc. can be used. As a result, it is possible to grasp at which position on the document data the term stored in the first column is used. In the case of new registration, the second column stores all “new” attributes, and the third column stores null values. The third column is used at the time of revision.

The related document information 305 includes a term table 3
The registered term is stored in the first column of 04, and zero or one or more links to the relevance information of other documents using the same term are stored in the second column. Specifically, the first column is generated based on the information in the term table, and the second column is brute-force checked for the terms in the first column and the first column of the related document information of another document by the relevance information extraction function 105. By doing so, if there is the same term, related document information of yourself and the other party 3
The second column of 05 is updated at the same time. Furthermore, paying attention only to the second column, it is checked whether or not a link relating to an unrelated term is registered, and it is corrected or deleted as necessary.

Specifically, by comparing the first column of the related document information 305 and the related document information 306, the document data A
And the term “Term A” commonly used in the document data B and the related document information 305 and the related document information 3 are found.
In the 1st line and 2nd column of 06, the relevance information 114 of each document data
A link to will be added. Further, if an inconsistent link is registered in the second column in the relation between the own document data and the other party's document data, it will be corrected. By this processing, the relationship between documents is maintained regardless of the difference in registration conditions such as new registration or revision, or the registration order of documents.

FIG. 4 is a schematic diagram showing how the relevance information 114 is modified upon revision. Document data A
Is registered, the existing relevance information 114 of the document data A is copied, the first column of the management table 402 is corrected to the document name of the revised document data, and the second column is linked to the document DB. Is corrected to a link indicating the new version of the document data A, and the third column is corrected to a link to the related information 114 of the old version. Further, the fourth column is corrected to the copied reference / citation table, term table, and link to related document information. As a result, the relationship between the new version document data A stored in the document DB 110 and the relevance information 114 and the relationship information 114 for the old version document data is retained.

The reference / citation table 403 is initialized once and the same processing as that at the time of new registration is performed.

Existing information in the term table 404 is modified based on the difference information received from the document data management function 103. If the terms are not rewritten, the second and third columns have null values, and if the appearance position information does not change, there is no correction. When the term has been changed, the “change” attribute is stored in the second column, the term before the change is stored in the third column, and the appearance position information is also corrected as necessary. When the term is deleted, the “delete” attribute is designated in the second column and the appearance position information is also deleted. When a term is added, the "addition" attribute is designated in the second column, and the appearance position information is also added.

A specific example of rewriting processing will be described using the difference 406. The difference 406 is information obtained by comparing a part of the old document data and the new document data. Among the character strings of the difference, the underlined part is the added character string, and the strikeout line part is the deleted character string. The double-strike-through character string is the modified term, and the underlined character string is the modified term. The display of the difference 406 does not define the format of the data, the method of taking the difference, or the like, but is a sample that illustrates the difference information.

The terms stored in the term table 404 are
It is extracted by the relevance information extraction function 105 according to the information of the term classification designation 109, and is stored together with the appearance position information in the order of appearance on the document data. Therefore, it indicates that the terms in the document of the old document data are registered in the order of appearance, and the term table 404 can be corrected using the information of the difference 406 as a key.

Since "Term A" and "Term C" in the term table 404 have not been corrected by the difference 406, the second and third columns of the term table 404 have null values, and if there is no page misalignment, etc., the appearance position. The information remains unchanged. The same applies to the processing regarding the appearance position.

Since the "Term C" on the third line is changed to "Term E" on the difference 406, the first column is changed to "Term E", the "change" attribute is set to the second column, and the third column is set to the "change" attribute. The original term “Term C” is registered. This will retain information about term changes.

Since there is no change in "Term A" on the 4th line, 1
Since the same processing as the line is performed and "Term D" on the 5th line has been deleted from the new document data, the "delete" attribute is registered in the 2nd column and the appearance position information in the 4th column is also corrected to a null value. Has been done.

Since it is shown that "Term F" in the sixth line is not in the old document data and is added on the difference 406, a new line is added and "Term F" is added in the first column. Two
The “addition” attribute is registered in the column, and the appearance position information is registered in the fourth column.

By the above processing, the term table 404
With respect to the above, all the effects of the revision of the document data have been understood.

The related document information 405 is modified in relation to the added, changed, or deleted terms. When the term having the "addition" attribute is not in the related document information 405, this term is added and a link to the relevance information of another document is registered,
If so, no action is taken.

In processing a term having a change attribute, first, a line for the changed term is added, a changed term is added in the first column, and an old version for the term before the change is added in the second column. In addition to copying the link to the stored relevance information of the other document, a "change" attribute is added to the first and second columns. This is because it is necessary to assume that the change of the target term is necessary for other document data.

The "change" attribute added to the first and second columns of the third line is held until all the links to the relevance information of other documents of the old version disappear or are corrected. The link to the relevance information of other documents in the old version is corrected because the reference document data is revised and "term C" is replaced with "term E" so that the change attribute is canceled or
It is either deleted by directly using the “term C” or the corresponding document data is deleted.

In the processing of the term having the "deletion" attribute, the processing of deleting all the links from the related document information included in the related information 114 of another document to the related information of the relevant document data, and "Delete" The term having the "attribute" is finally deleted from the term table 404, so that unnecessary information does not remain.

By the series of processes described above, all the influences of the revision are reflected in the data group 401 of the revised document data A.

FIG. 5 shows the system definition information definition function 104.
FIG. 6 is a diagram illustrating a process of defining the contents of system definition information 107 using.

When defining or modifying the contents of the system definition information 107, first, the system definition information definition screen 501 is displayed.
Is displayed on the terminal 101. On the definition screen 501 of the system definition information, two items of “designation of sentence pattern dictionary” and “designation of term category” are displayed as a menu of definition items.

When "Specify text pattern dictionary" is selected, the text pattern dictionary specification screen 502 is displayed, and the contents of the text pattern dictionary 505 of the system definition information 504 are referenced and displayed in an editable state.

The sentence pattern dictionary 505 stores a sentence pattern of a character string indicating a reference or a citation appearing in the sentence in the first column, and a character string indicating the type of the extracted information in the second column. The reason for paying attention to the sentence pattern of a character string that indicates a reference or a citation that appears in a sentence is that when a human reads the sentence, the character string indicates a reference or a citation, etc. This is because it is written based on the rules of expression. Therefore, by using the character string in the first column of the sentence pattern dictionary 505 to extract information for specifying the document data of the other party from the character string indicating reference, citation, etc., the relevance between documents can be specified. is there.

The form of the character string registered in the first column of the sentence pattern dictionary 505 will be described in detail.

In the case of "Refer to manual" (. *) "On the first line, it is a sentence pattern that matches a character string such as" Refer to manual "Introduction to ABC""in the document data.
Indicates that ". *" Matches any variable character string and extracts the character string within the range enclosed by the parentheses. Therefore, the character string "Introduction to ABC" can be extracted as the document name from the character string "Refer to manual" Introduction to ABC "".

If any document data can be specified from the issue date, "([0-9] * / [0-9] * / [0-9]
By specifying the character string "*)", "2001/0"
You can match the string "8/07".
This is because the part of "[0-9] *" can match the number of arbitrary digits.

It is also possible to deal with a character string having a plurality of formats such as a patent publication number. For example, "Japanese Patent Laid-Open No. 9-1
When the character strings "46968" and "JP 2000-99543" are matched in one sentence pattern, "(JP ?? [0-9] *-[0-9]" shown in the third line is displayed.
*) ”Is specified. The part "Hira?" Indicates that 0 or 1 of the characters "Hira" match, so this character string is used to indicate "Japanese Patent Laid-Open No. 9-146968".
And "Japanese Patent Laid-Open No. 2000-99543" can be matched.

Naturally, if the text pattern is not registered in the text pattern dictionary 505, even a character string indicating reference or citation cannot be extracted, but unnecessary information such as reference and citation can be eliminated, It is important to be able to control the references and citations that you want to pay attention to.

Next, when "Specify term category" is selected, the term category designation screen 503 is displayed, the content of the term category designation 506 of the system definition information 504 is referred to, and the items on the screen are displayed according to the existing definition content. Will be updated.

The first column of the designation 506 of term classification is the term classification or additional condition item of interest. In the state where "on" or "off" is specified in the second column and "on" is selected,
This indicates a state where "off" is not selected. The term classification designation screen 503 is a sample, and the items that can be selected on the screen are registered according to the type and purpose of the document data to be handled, the morphological analysis, or the functional requirements of a sentence analysis method such as syntactic analysis. For example, a term category designation screen 50
In 3, the classification in the national grammar such as “proper noun”, “general noun”, “verb”, and additional conditions for determining the handling of particles can be specified.

In the term segment designation screen 503, only the "proper noun" portion is selected, so only the second column of the first line of the term segment designation 506 is on and the other items are off. ing.

The reason for extracting the terms used in the document data and grasping the usage status of each term is that one information is subdivided into multiple documents and a plurality of documents are used according to the usage of the information. It is to understand the relationship between documents implied by the terms used in response to the conditions at the time of document production such as the situation where information is grouped.

Therefore, by using the sentence pattern dictionary 505 and the term classification designation 506, the relevance between documents can be comprehensively defined by the character strings that express reference or citation explicitly and the terms used in the sentence data. You will be able to handle it.

FIG. 6 is a diagram for explaining the process of specifying the referenced document using the information registered in the sentence pattern dictionary 108.

In the first line, first column of the text pattern dictionary 602 of the system definition information 601, "manual" (. *) "Is displayed.
The character string "refer to" is registered. When the document data registered in the document DB 603 is searched using this character string, refer to the manual “ABC System Guide” for details contained in the document data of the ABC Primer 604. The part of "Refer to the manual" Refer to "ABC system guide""in the line" matches and the "ABC system guide" is extracted as information for identifying the other party.

Furthermore, since the extracted information indicates that it is the document name in the second column of the sentence pattern dictionary 602, it is defined in the document information of another document using the character string "ABC system guide". The ABC system guide 605 can be specified as the other party by searching the document name that is displayed.

As described above, by using the sentence pattern dictionary 602, information for identifying the document data of the other party can be extracted, and the document data of the reference destination can also be identified.

FIG. 7 is a diagram for explaining a method of extracting a term according to designation 109 of a term category. When a term is extracted from the sentence sample 701 using the designation of the term classification 702 in which only “proper noun” is turned on, “ABC system”, “log file”, “[file (F)] −
The three terms “log (R) command” are extracted.

"Proper noun" and "include particle" no ""
When a term is extracted from the sentence sample 701 using the designation of the term classification 703 in which “ON” is turned on, “ABC system”, “Generate log file”, “[File (F)]-[Log (R)] Three terms of "command" are extracted.

The difference between the above two definitions of terms in terms of designation is whether or not "include" no "of particle" is turned on, and the extracted terms are "log file" in the former and "latter in the latter". Is "generation of log file", and it is shown that different terms are extracted by specifying the term classification.

As described above, by using the term classification designation 109, the terms used in the document data can be extracted in a form necessary for extracting the relevance between documents.

FIG. 8 is a diagram showing a sample screen of the user interface generated by the display data generating function 106.

As described above, the relevance DB 113 stores various information regarding the document data stored in the document DB 110. This information can be displayed on the display screen of the terminal 101 by using the display data generation function 106.

The display screen 801 shown in FIG.
14 is a sample user interface displayed on the terminal 101 for designating a condition for extracting information from 14. The display screen 801 has a search function and a display function. When using the search function, it indicates that a term search can be realized by entering a term and clicking the [Search] button.

In addition to this, functions such as "display of document related diagram", "display of term related diagram", "display of reference destination of specific document" and the like are also prepared. For example, if you select "Display Document Relationship Diagram", the Document Relationship Diagram will be displayed, and "Display Uncorrected Terms" will be displayed.
If you select, a list of documents containing unmodified terms is displayed.

All of these functions are related DB 113.
It can be realized by using the relevance information 114 stored in the document, and can be said to be a convenient function for grasping the relevance between documents.

Regarding a specific method of using the information on the relevance DB 113, a typical function is a "search" function,
A description will be given below using three functions of "display document relation diagram" function and "display uncorrected term" function.

FIG. 9 is a diagram for explaining the “search” function provided by the display data generation function 106. When “Term C” is input in the search term input area (see FIG. 8) of the relevance information display screen 901 displayed on the terminal 101 and the search execution button is pressed, the search result 902 is displayed. The search result 902 has a document name and a link for displaying details. When the document name link is selected, the document DB 11
The document data stored in 0 can be retrieved,
When the link of the appearance location is selected, the appearance position information of the relevant term is displayed in a list from the term table of the relevant document data.

When displaying the document name in the search result 902, the information to be searched is the related document information of the relevance information 114. For example, document data A, document data B,
The related document information of the document data C is the document data A, respectively.
Related document information 903 and related document information 9 of document data B
04, in the case of the related document information 905 of the document data C, if "Term C" is input as a search term, all three will match.

When "Term D" is input, the document data B is displayed, and when "Term E" is input, only the document data C is displayed as the search result.

FIG. 10 is a diagram for explaining the "display document relation diagram" function provided by the display data generation function 106.

When the display of the document relation diagram (see FIG. 8) on the relation information display screen 1001 displayed on the terminal 101 is selected and the display execution button is pressed, the document relation diagram 1002 is displayed. A document relation diagram 1002 illustrates reference / citation relations between document data. By selecting document data, the document data stored in the document DB 110 can be extracted, and the relations between documents can be extracted. You can visually grasp the sex.

Information to be searched when the document relation diagram 1002 is displayed is the reference / citation table of the relation information 114. For example, the reference / citation table of the document data A, the document data B, the document data C, the document data D, the document data E, and the document data F are the reference / citation table 1003 of the document data A and the reference / citation table of the document data B, respectively. Citation table 1004, document data C reference / citation table 1005, document data D reference / citation table 100
6, the reference / quotation table 1007 of the document data E and the reference / citation table 1008 of the document data F are displayed as a document relation diagram 1002.

Specifically, "document data A" includes a reference to "document data C", and "document data B" includes a reference to "document data C" and "document data G" that does not exist. Indicates that Since the “document data G” does not exist, it can be confirmed that the “document data G” is displayed in the shape of an ellipse on the document relation diagram 1002.

Since "document data E" includes references to "document data C" and "document data D" and is also referred to by "document data D", it is a document data having a cross-reference relationship. I can confirm. If the document data is not referenced by any document, then "document data F"
Even if the document data does not include a reference to another document, it is displayed as a referenced document such as “document data C” if the document data is referenced by another document.

FIG. 11 is a diagram for explaining the “display uncorrected term” function provided by the display data generation function 106.

When the display of uncorrected terms (see FIG. 8) on the related information display screen 1101 displayed on the terminal 101 is selected and the display execution button is pressed, the uncorrected term display screen 1 is displayed.
102 is displayed. Uncorrected term display screen 1102
Is used to grasp the range of influence when a specific term is changed.To display the changed term, the document candidate that needs to be changed, and the appearance position for each term selected in the term list. A list of links is displayed. If the document data displayed in the document candidates that need to be changed is selected, the document data stored in the document DB 110 can be retrieved, and if a link (appearance position) for displaying the appearance position is selected, A list of appearance points in each document data is displayed.

Information to be searched when displaying the uncorrected term display screen 1102 is the term table of the relevance information 114 and the related document information. For example, the relevance information of the document data A and the relevance information of the document data B are the relevance information 1103 of the document data A and the relevance information 1104 of the document data B, respectively.
If “Term C” is selected in the term list, the uncorrected term display screen 1102 is displayed as shown in the figure.

Specifically, the relevance information of the document data in which the "change" attribute is designated in the second column of the term table and the "term C" is designated in the third column, and then designated in the first row and the first column. The related document information corresponding to the selected term will be referred to.

In this case, the relevance information 11 of the document data A
03, the second column of the related document information stored in the relevance information 1104 of the document data B is treated as a document candidate that needs to be changed. Among the links stored in the second column of the related document information, Information that does not have a "change" attribute is excluded. As described above, the condition for deleting the “change” attribute in the second column of the related document information is, for example, when the referenced document is revised, and the relevance information 1 of the document data A
The related document information 103 indicates that “document data D” has been revised, and “term C” included in “document data D” has been changed to “term E”. As a matter of course, since the “term C” is changed to the “term E” due to the revision of the “document data D”, the “document data” existing in the second column of the relevant document information of the relevance information 1104 of the document data B. The link to "D" has been deleted.

As a result, the uncorrected term display screen 1102 is displayed.
"Terms E" and "F" as "terms after change"
However, "document data C" and "document data D" are displayed as the document candidates that need to be changed.

As described above, according to the present embodiment, information indicating the relevance between documents is extracted without distinction between the explicitness / implication of the relevance, and the extracted information is stored separately from the document data and the like. After that, the information according to the relevance between the documents is acquired using the extracted information, so that the information according to the relevance between the documents can be acquired without omission by a simple operation.

[0091]

As described above, according to the present invention, the information according to the relation between documents can be easily obtained by various procedures by accumulating the relation information between documents without distinction between explicit and implicit. It becomes possible to do.

[Brief description of drawings]

FIG. 1 is a diagram showing a document data relevance extraction apparatus according to an embodiment of the present invention.

FIG. 2 is a flowchart showing a process performed by a relevance information extractor.

FIG. 3 is a schematic diagram showing how relevance information is stored during new registration.

FIG. 4 is a schematic diagram showing how the relevance information is modified upon revision.

FIG. 5 is a diagram illustrating a process of defining the contents of system definition information using a system definition function.

FIG. 6 is a diagram illustrating a process of identifying a referenced document using information registered in a text pattern dictionary.

FIG. 7 is a diagram illustrating a method of extracting a term according to designation of a term classification.

FIG. 8 is a diagram showing a sample screen of a user interface generated by a display data generation function.

FIG. 9 is a diagram illustrating a “search” function provided by a display data generation function.

FIG. 10 is a diagram illustrating a “display document related diagram” function provided by a display data generation function.

FIG. 11 is a diagram illustrating a “display uncorrected term” function provided by a display data generation function.

[Explanation of symbols]

101 terminal 102 Document Data Relationship Extraction Device 103 Document data management function 104 System definition information definition function 105 Relevance information extraction function 106 Display data generation function 107 system definition information 108 sentence pattern dictionary 109 Specifying term classification 110 document DB 111 Document information 112 document data 113 Relationship DB 114 Related information

Claims (5)

[Claims] 1. A document data management function for managing registration and reference of a document and acquisition of difference information of the document based on an instruction from a terminal, document information for specifying the registered document, and contents of the document. A document database that stores document data that represents the above, and system definition information that defines a relationship extraction condition for extracting the relationship between the stored documents and system definition information that defines change or deletion of the extraction condition Definition function, a function of extracting relevance information that extracts information indicating relevance between documents from documents stored in the document database based on the system definition information, and information indicating relevance between extracted documents A document data relevance extraction device having a display data generation function for displaying. 2. The system definition information defining function according to claim 1, wherein the system definition information definition function corrects the system definition information, and based on the modification, the extraction unit re-extracts information indicating the relationship between the documents. Characteristic document data relation extraction device. 3. The display data generation function according to claim 1, further comprising a user interface for designating a display form of the relationship between documents displayed on the terminal, and the extracted document inter-documents are designated according to the designation of the interface. A document data relevance extraction device characterized by displaying information indicating relevance of the. 4. A step of managing registration and reference of a document and difference information of the document based on an instruction from a terminal, and document information specifying the registered document and document data representing the content of the document are mutually exchanged. Storing them in a document database in association with each other; defining relationship extraction conditions for extracting the relationships of the stored documents and defining system definition information for changing or deleting the extraction conditions; A step of extracting information indicating the relationship between the documents from the documents stored in the document database based on the above, and a step of displaying the information indicating the relationship between the extracted documents. Document data relevance extraction program 5. The method according to claim 4, wherein the step of defining the system definition information includes a step of modifying the system definition information, and the step of extracting information indicating the relevance between the documents is based on the modification. A document data relationship extracting program comprising the step of re-extracting information indicating the relationship between documents.