JP4356640B2 - Document management device - Google Patents

Description

  The present invention relates to a document management apparatus, and more particularly, to a document management apparatus that manages a document set that is classified and hierarchized.

Typical conventional techniques for classifying documents in a hierarchical manner are described in Patent Document 1 and Patent Document 2. In the invention of Patent Document 1, a means for referring to classification information generated in the past and a means for reusing are given to the user. The classification information used is a classification label, for example, a classification definition that is a Boolean expression of words, a classified document. Classification and structuring are basically performed by the user. Further, in the invention of Patent Document 1, category structure information is generated based on the classification result, the generated category structure information is stored, and is generated in the past that is stored during the subsequent generation work of the category structure information. The category structuring work can be performed efficiently by referring to and modifying the category structure information and performing the category structuring work. Further, in the invention of Patent Document 2, hierarchical classification is realized by recursively calling a document classification apparatus based on the vector space method. Classification is not a user task, it is automated.
JP 2003-216622 A JP 2003-141129 A

However, in the method in which a user classifies documents by presenting information that helps classification in Patent Document 1, classification accuracy is high, but there is a problem that much labor is required for classification work. In order to maintain the classification accuracy, hierarchical classification must be performed by a limited user who is familiar with all areas to which the document belongs, and the obtained hierarchical classification is the hierarchical classification considered by the entire user. There is a possibility of deviation. Further, in Patent Document 1, although it is possible to efficiently perform category structuring work by referring to or correcting category structure information generated in the past, a history of search routes searched by all users according to the classification hierarchy Based on the above, it is impossible to automatically add a classification to a predetermined classification hierarchy and register a document. Further, the method of fully automating the hierarchical classification of Patent Document 2 does not require the labor of classification work, but has a problem that the accuracy of classification is lowered.
An object of the present invention is to provide an apparatus for performing hierarchical classification of documents with little effort and high accuracy. Here, high accuracy means that it is closest to the average concept hierarchy of all users.

The document management apparatus of the present invention is a document management apparatus for managing a hierarchically classified document group, a document storage means for storing a classification hierarchy and a document group associated with the classification hierarchy, and a user Interface means for interactively searching for a document along the classification hierarchy, and when the user searches for a document along the classification hierarchy by the interface means, the final classification reached by the user is used as a history. A history storage means for storing and a registration candidate for storing a document that the user wishes to newly register as a registration candidate document in association with the history when the user cannot find the document intended by the search registration climate to the document storage means, and the ones that are under the same final classification of the registered candidate document to determine from the history, the new classification of high similarity group of them Characterized in that it comprises a document classification means.
Further, the registration candidate document classification means generates a new classification by classifying the registration candidate document for each final classification reached by the user, and the number of the new classification exceeds a predetermined constant value. For example, the new classification can be added under the final classification.
Further, the registration candidate document classification means generates the new classification by classifying the registration candidate document for each final classification reached by the user, and the number of the new classification does not exceed a predetermined constant value In addition, the registration candidate document that best represents the new classification can be added as a new document.
Further, the registration candidate document classification unit generates a new classification by classifying the registration candidate document for each classification higher than the final classification reached by the user, and a constant value in which the number of the new classification is determined in advance. The new classification can be added below the superordinate class only when the number exceeds .
The document management method of the present invention is a document management method in a document management apparatus that manages a hierarchically classified document group, and stores a classification hierarchy and a document group associated with the classification hierarchy. And an interface step for allowing a user to search for a document interactively along the classification hierarchy , and a final result reached by the user when the user searches for a document along the classification hierarchy in the interface step. A history storing step for storing the classification as a history, and when the user cannot find a document intended by the search, a document that the user desires to newly register is associated with the history as a registration candidate document. a registration candidate document storage step of storing, what is under the same final classification of the registered candidate document to determine from the history, among them Characterized in that it comprises a registration candidate document classification step of the high group of similarity score as a new classification.
In the registration candidate document classification step, the registration candidate document is classified for each final classification reached by the user to generate a new classification, and the number of the new classifications exceeds a predetermined constant value. Only the new classification can be added below the final classification.
Further, in the registration candidate document classification step, the registration candidate document is classified for each final classification reached by the user to generate a new classification, and the number of the new classifications does not exceed a predetermined constant value. In this case, the registration candidate document that best represents the new classification can be added as a new document.
Further, in the registration candidate document classification step, the registration candidate document is classified for each classification higher than the final classification reached by the user to generate a new classification, and the number of the new classifications is determined in advance. Only when the numerical value is exceeded, the new classification can be added under the superordinate classification.
The document management program of the present invention is a document management program for controlling a document management apparatus that manages a hierarchically classified document group, and stores a classification hierarchy and a document group associated with the classification hierarchy. A storage step; an interface step for allowing a user to search for a document interactively along the classification hierarchy; and the user reaching when the user searches for a document along the classification hierarchy in the interface step. A history storage step for storing the final classification as a history, and when the user cannot find a document intended by the search, a document that the user desires to newly register corresponds to the history as a registration candidate document. a registration candidate document storage step of storing in association, those under same final classification of the registered candidate document to determine from the history Among them a registration candidate document classification step of the high similarity group and a new classification, characterized in that to be executed by the document management apparatus.
In the registration candidate document classification step, the registration candidate document is classified for each final classification reached by the user to generate a new classification, and the number of the new classifications exceeds a predetermined constant value. Only the new classification can be added below the final classification.
Further, in the registration candidate document classification step, the registration candidate document is classified for each final classification reached by the user to generate a new classification, and the number of the new classifications does not exceed a predetermined constant value. In this case, the registration candidate document that best represents the new classification can be added as a new document.
Further, in the registration candidate document classification step, the registration candidate document is classified for each classification higher than the final classification reached by the user to generate a new classification, and the number of the new classifications is determined in advance. Only when the numerical value is exceeded, the new classification can be added under the superordinate classification.

  According to the document management apparatus of the present invention, when the user cannot find the intended document by the search performed according to the classification hierarchy, the registration candidate document that the user desires to newly register is stored. The following effects can be obtained. The first effect is that the labor of hierarchical classification can be reduced. All that is required is that each user searches for the desired document according to the classification hierarchy and registers the document when it is not found. Adding new classifications to the classification hierarchy is done automatically. The second effect is that the accuracy of hierarchical classification can be maintained. This is because it is determined by all users who are related to the position of the new classification hierarchy in the existing classification hierarchy.

  The present invention relates to the field of managing a document set that is classified into hierarchies. For example, when a frequently asked question (FAQ) is searched according to a classification hierarchy and a desired question and answer cannot be found, the present invention is newly added. Including adding additional questions and answers

  Next, the best mode for carrying out the invention will be described in detail with reference to the drawings.

  FIG. 1 is a diagram for explaining the operation of the embodiment of the present invention. The document storage unit 101 stores a classification hierarchy and a document. The interface unit 102 causes the user to search for a document interactively along the classification hierarchy, and outputs a search history. If an appropriate document is not found, the interface unit 102 causes the user to input a registration candidate document and outputs the document. The history storage unit 103 stores the user search history output by the interface unit. The registration candidate document storage unit 104 stores the registration candidate document output by the interface unit. The registration candidate document classification unit 105 classifies the registration candidate documents, and adds the obtained classification to the existing classification hierarchy position.

  Before describing each component of the embodiment shown in FIG. 1, the concept of the classification hierarchy and the document will be described with reference to FIG. The first level of classification includes two classifications, classification 1 and classification 2. There are three sub-classes of class 1, class 11, class 12, and class 13. There are three sub-classes of class 2: class 21, class 22, and class 12. There are two subclasses of class 13: class 131 and class 132. The document 11a, the document 11b, and the document 11c are associated with the classification 11.

  For example, consider a user who wants to know how to manage PC email software. The user selects category 1 “About mail” and then selects category 11 “PC mail”. However, since a document with the content intended by the user cannot be found, the problem is solved by another means, and the content is changed. I would like to register the described document for the next user. This document is called a registration candidate document and stored in the registration candidate document storage unit 104. At this time, history information “class 1 → class 11 → registration candidate document” is stored in the history storage unit 103. In this way, registration candidate documents are successively stored in the document management apparatus.

  A feature of the first embodiment of the present invention is that these accumulated registration candidate documents are classified, and a new classification is created under the classification 11. For example, as shown in FIG. 9, when there are a large number of registration candidate documents related to “mail software management” and the independence from other new documents is high, “mail software management” is classified as a category 111 in the classification hierarchy. It is additionally registered. The difference from the prior art is that not only the contents of the document but also the history information is used for creating the classification.

  FIG. 10 shows a conceptual diagram of the second embodiment. In FIG. 10, unlike FIG. 9, a new document is added instead of adding a new category under the category 11. It is appropriate to add a document that does not become a large chunk as a classification in this way.

  FIG. 11 shows a conceptual diagram of the third embodiment. In FIG. 11, unlike FIG. 9, a new classification is added at the same level as the classification 11 instead of adding a new classification under the classification 11. Depending on the level of abstraction of the classification, it is appropriate to add a new classification at the same level as the existing classification.

  Subsequently, each component in FIG. 1 will be described in detail. FIG. 2 is a diagram illustrating a configuration example of the document storage unit 101. Reference numeral 201 denotes a classification hierarchy. The classification hierarchy 201 is a table including a classification ID 203, a higher classification ID 204, and a classification name 205. The classification ID 203 is a numerical value that uniquely identifies the classification. The upper class ID 204 indicates a class ID located at the upper level of the class. There can be a plurality of classification IDs. For example, in the figure, classification 12 indicates that classification 1 and classification 2 are higher classifications. The classification name 205 is the name of the classification, and is used when displayed to the user.

  Reference numeral 202 denotes a document. The document 202 is a table including a document ID 206, a classification ID 207, a document name 208, and a document content 209. The document ID 206 is a numerical value that uniquely identifies the document. A classification ID 207 is a classification ID to which the document belongs. The document name 208 is the name of the document, and is used when displayed to the user. The document content 209 is a character string that is the substance of the document, and is used when displayed to the user.

  FIG. 3 is a diagram illustrating a configuration example of the interface unit 102. 301 shows an example of screen transition. On the screen 304, classification 1 and its subordinate classifications classification 11, classification 12, and classification 13 are displayed. When “PC mail” of category 11 is selected on the screen 304 using a pointing device such as a mouse, the document names of documents belonging to the category 11 are displayed as in the screen 305. When the “type of mail software” which is the document 11 a is selected using a pointing device such as a mouse on the screen 305, the document name and document content of the document 11 a are displayed as in the screen 306.

  The screen control unit 302 controls such screen transition and transmits the classification selected by the user and the document ID to the history acquisition unit 303. If there is no document requested by the user, the user selects a new button 305a on the screen 305 by using a pointing device such as a mouse, and the screen changes to a screen for inputting a registration candidate document. Registration candidate documents can be registered from the screen. The registration candidate document is transmitted to the registration candidate document acquisition unit 307 by the screen control unit 302.

  FIG. 4 is a diagram illustrating a configuration example of the history storage unit 103. The history storage unit 103 is a table including a history ID 402, a final classification 403, a history 404, and an arrival time 405. The history ID 402 is a numerical value that uniquely identifies the history. The final classification 403 is a classification ID of the classification finally reached by the user. The history 404 is a classification route to the final classification 403. For example, as shown in the third and fifth rows of Table 103, the path may be different even if the final classification is the same. The arrival time 405 is the time when the user reaches the final classification 403.

  FIG. 5 is a diagram illustrating a configuration example of the registration candidate document storage unit 104. The registration candidate document storage unit 104 is a table including registration candidate document ID 502, history ID 503, and document content 504. The registration candidate document ID 502 is a numerical value that uniquely identifies the registration candidate document. The history ID 503 is a history ID of a route taken by the user when the registration candidate document is newly described. The document content 503 is a character string that is the substance of the registration candidate document.

  FIG. 6 is a diagram illustrating a processing flow of the registered document candidate classification unit 105. In 601, a registration candidate document is acquired from the registration candidate document storage unit 104. In 602, the document acquired in 601 is classified for each final classification 403. Since the history candidate ID 503 is assigned to the registration candidate document, the final classification 403 is determined by collating with the history ID 402 of the history storage unit 103.

  Thereafter, processing for determining whether a new category should be added for each final category 403 is performed. Step 603 is a step of determining whether or not the processing has been performed for all final classifications. In step 604, the registration candidate documents are classified. A known method may be used as the classification method. For example, a method may be used in which a vector is created based on the frequency of words appearing in a document, a document distance is calculated by a vector cosine measure, and documents having a distance equal to or less than a predetermined distance are clustered.

  In step 605, the nature of the document set classified in step 604 is examined, and it is determined whether or not to adopt as a new classification. As a criterion for determination, for example, the number of clusters and the minimum number of documents belonging to one cluster can be considered. These criteria are defined in advance for the entire apparatus. For example, if the number of clusters is small, it means that the number of classifications is small, so there is no need to create a new classification.

  Further, when the number of documents belonging to one cluster is small, there is a risk that the number of classifications will increase more than necessary if they are added as classifications. If it is determined in step 605 that the new classification should be added, the new classification and the document are added to the classification hierarchy 201 and the document 202 of the document storage unit 101 in step 606. After the addition, the corresponding document is deleted from the registration candidate document storage unit 104.

  FIG. 7 is a processing flow of the registered document candidate classification unit 105 according to the second embodiment. Step 705 and step 706 are different from FIG. 6 of the first embodiment. In step 705, it is determined whether a new document should be registered instead of creating a new category under the final category 604. For example, in FIG. 9 which is a conceptual diagram showing the addition of the classification in the first embodiment, the classification 111, the classification 112, and the classification 113 indicate new classifications added by the apparatus of the first embodiment.

  On the other hand, in FIG. 10, which is a conceptual diagram showing document addition in the second embodiment, representative documents of classification 111, classification 112, and classification 113 are registered as document 11d, document 11e, and document 11f, respectively. . For those that do not reach the criterion for adding a new classification in step 605 in FIG. 6, a second criterion for the number of clusters and the number of documents belonging to one cluster is determined in advance. In step 705, the document is registered as a representative document.

  A known technique may be used to select the representative document. For example, a method of selecting a document closest to the center of gravity of the cluster can be considered. Furthermore, considering the contents of nearby documents, it may be possible to provide an editing means that can create a new document manually.

  FIG. 8 is a processing flow of the registered document candidate classification unit 105 according to the third embodiment. FIG. 11 is a conceptual diagram showing a state in which a new classification is added according to this embodiment. A classification 14 and a classification 15 are added under the classification 1. These two classifications are generated from the sum of the registration candidate documents of classification 11, classification 12, and classification 13, which are the final classification. In other words, the new classification is not added under the classification 11, the classification 12, and the classification 13, but is added in parallel.

  FIG. 8 differs from FIG. 6 of the first embodiment in steps 802, 803, and 806. In addition, step 809 is newly added. In step 802, the registration candidate documents are not classified for each final classification as in step 602, but the registration candidate documents are classified according to the classification of an arbitrary hierarchy. For example, in FIG. 11, registration candidate documents under classification 11 and classification 12 are also handled as registration candidate documents of classification 1 that is higher than classification 11 and classification 12. Step 809 is a process of sorting the classes in order of depth in order to make a determination in order from the class of the deep hierarchy when adding a new class.

  For example, in FIG. 11, it is determined whether classification 11, classification 12, and classification 13 have priority over classification 1, classification 131 and classification 132 have priority over classification 13, and whether a new classification can be added below. In step 803, it is determined whether or not the processing has been completed for all arbitrary classifications, not the processing for all final classifications as in step 603. Step 806 is different from step 606 in which a new classification is added under the final classification, and is added under the classification determined in step 803.

  It should be noted that the configuration and operation of the above-described embodiment are examples, and it goes without saying that they can be changed as appropriate without departing from the spirit of the present invention.

  The present invention can be applied to applications such as an information retrieval apparatus that efficiently retrieves a large amount of documents and a program for realizing the information retrieval apparatus in a computer.

It is a block diagram which shows the structural example of embodiment of this invention. 2 is a diagram illustrating a configuration example of a document storage unit 101. FIG. 3 is a diagram illustrating a configuration example of an interface unit 102. FIG. 3 is a diagram illustrating a configuration example of a history storage unit 103. FIG. 3 is a diagram illustrating a configuration example of a registration candidate document storage unit 104. FIG. It is a processing flow of the registration document candidate classification | category part 105 in 1st Embodiment. It is a processing flow of the registration document candidate classification | category part 105 in 2nd Embodiment. It is a processing flow of the registration document candidate classification | category part 105 in 3rd Embodiment. It is a conceptual diagram of 1st Embodiment. It is a conceptual diagram of 2nd Embodiment. It is a conceptual diagram of 3rd Embodiment.

Explanation of symbols

Reference Signs List 101 Document storage unit 102 Interface unit 103 History storage unit 104 Registered document candidate storage unit 105 Registered document candidate classification unit 201 Classification hierarchy 202 Document 203 Document ID
204 Upper classification ID
205 Classification name 206 Document ID
207 Classification ID
208 Document name 209 Document content 301 Screen transition of interface 302 Screen control unit 303 History acquisition unit 304 Screen showing classification and its lower classification 305 Screen showing classification and corresponding document 306 Screen showing document content 307 New document acquisition unit 402 History ID
403 Final classification 404 History 405 Arrival time 502 Registration candidate document ID
503 History ID
504 Document Contents

Claims (12)

A document management apparatus for managing a hierarchically classified document group,
Document storage means for storing a classification hierarchy and a document group associated with the classification hierarchy;
Interface means for allowing a user to interactively search for a document along the classification hierarchy;
When the user searches for a document along the classification hierarchy by the interface means, history storage means for storing the final classification reached by the user as history,
A registration candidate document storage unit that stores a document that the user desires to newly register as a registration candidate document in association with the history when the user cannot find the document that the user intended by the search;
Document management comprising: a registration candidate document classifying unit that determines, from the history, documents under the same final classification among the registration candidate documents , and sets a group having a high similarity among them as a new classification apparatus. The registration candidate document classification means classifies the registration candidate document for each final classification reached by the user, generates a new classification, and only when the number of the new classification exceeds a predetermined constant value. the document management apparatus according to claim 1, characterized in that adding the new classification under the final classification. The registration candidate document classification means classifies the registration candidate document for each final classification reached by the user to generate a new classification, and when the number of the new classification does not exceed a predetermined constant value, The document management apparatus according to claim 1, wherein the registration candidate document that best represents the new classification is added as a new document. The registration candidate document classification means classifies the registration candidate document for each classification higher than the final classification reached by the user to generate a new classification, and the number of the new classifications exceeds a predetermined constant value. 2. The document management apparatus according to claim 1, wherein the new classification is added below the upper classification only in a case . A document management method in a document management apparatus for managing a hierarchically classified document group,
A document storage step for storing a classification hierarchy and a document group associated with the classification hierarchy;
An interface step for allowing a user to interactively search for a document along the classification hierarchy;
When the user searches for a document along the classification hierarchy in the interface step, a history storage step for storing the final classification reached by the user as a history;
A registration candidate document storage step of storing a document that the user desires to newly register in association with the history as a registration candidate document when the user cannot find the document intended by the search;
Document registration comprising: a registration candidate document classification step for determining, from the history, documents under the same final classification among the registration candidate documents , and setting a group having a high similarity among them as a new classification. Method. In the registration candidate document classification step, the registration candidate document is classified for each final classification reached by the user to generate a new classification, and only when the number of the new classifications exceeds a predetermined constant value. Te, document management method according to claim 5, characterized in that adding the new classification under the final classification. In the registration candidate document classification step, the registration candidate document is classified for each final classification reached by the user to generate a new classification, and when the number of the new classification does not exceed a predetermined constant value the document management method according to claim 5, characterized in that adding the registration candidate documents that best represent the new classification as a new document. In the registration candidate document classification step, the registration candidate document is classified for each classification higher than the final classification reached by the user to generate a new classification, and a constant value in which the number of the new classifications is determined in advance. 6. The document management method according to claim 5, wherein the new classification is added below the higher classification only when exceeding the upper classification. A document management program for controlling a document management apparatus that manages a hierarchically classified document group,
A document storage step for storing a classification hierarchy and a document group associated with the classification hierarchy;
An interface step for allowing a user to interactively search for a document along the classification hierarchy;
When the user searches for a document along the classification hierarchy in the interface step, a history storage step for storing the final classification reached by the user as a history;
A registration candidate document storage step of storing a document that the user desires to newly register in association with the history as a registration candidate document when the user cannot find the document intended by the search;
Determining from the history those documents that are under the same final classification among the registration candidate documents, and causing the document management apparatus to execute a registration candidate document classification step in which a group having a high degree of similarity is a new classification. Document management program characterized by In the registration candidate document classification step, the registration candidate document is classified for each final classification reached by the user to generate a new classification, and only when the number of the new classifications exceeds a predetermined constant value. Te, document management program according to claim 9, characterized in that adding the new classification under the final classification. In the registration candidate document classification step, the registration candidate document is classified for each final classification reached by the user to generate a new classification, and when the number of the new classification does not exceed a predetermined constant value the document management program according to claim 9, characterized in that adding the registration candidate documents that best represent the new classification as a new document. In the registration candidate document classification step, the registration candidate document is classified for each classification higher than the final classification reached by the user to generate a new classification, and a constant value in which the number of the new classifications is determined in advance. 10. The document management program according to claim 9, wherein the new classification is added below the superordinate classification only when the number exceeds the upper classification.

Images