JP5060601B2 - Document analysis apparatus and program - Google Patents

Description

  Embodiments described herein relate generally to a document analysis apparatus and program for presenting a tendency of a category to which a document important for a user belongs to the user.

  With the advancement of information systems in recent years, for example, documents such as patent documents, newspaper articles, web pages and books can be digitized and stored (saved) in large quantities.

  Therefore, it is required to effectively use such a large amount of documents for daily work.

  For example, by classifying a huge amount of newspaper articles in the past, it is arranged so that it can be used easily by many people, or the trend of patent groups of other companies related to the technology being researched and developed is analyzed. It is conceivable to consider a new R & D field.

  However, relevant documents are extracted from a large number of document groups, the document groups are classified into similar documents (that is, categories), or the tendency of the contents of the large number of document groups is analyzed. This work takes a lot of labor and cost.

  Therefore, various technologies such as document retrieval, document classification, and document analysis techniques have been developed to support these operations.

Japanese Patent No. 4240280

“Information Retrieval and Language Processing” Language and Calculation 5 The University of Tokyo Press pp39-43 (1999) Takenobu Tokunaga, Junichi Sakurai

  Here, from among a large number of document groups classified into a plurality of categories, for example, the user is still informed based on a document that is already known to the user and important for the user (hereinafter referred to as a known important document). Assume that a document important for the user who is not known (hereinafter referred to as an unknown important document) is searched.

  In this case, when an unknown important document is searched from all document groups regardless of the category etc., for example, a word appearing in a known important document even though the document is not actually important. There is a case where a document in which “” appears accidentally is retrieved as an important document.

  By the way, for example, when a user selects (specifies) an important document (an important document for the user) from a group of documents classified into a plurality of categories, the user views the document by paying attention to only one category. By doing so, it is often determined whether the document is the important document. That is, it is rare that the user knows what other category the document (important document) determined to be important belongs to.

  Therefore, for example, when an unknown important document is searched from a category designated by the user (a category that the user thinks contains an important document), a category not designated by the user (that is, an important document is An unknown important document cannot be searched from a category that is not noticed by the user if it is included.

  Therefore, if it is possible to grasp the tendency such as what categories the important documents for the user belong to, the trend is used together with the discovery of the category that the user is not aware that the important documents are included. Thus, it is considered possible to search (that is, narrow down) unknown important documents.

  Specifically, when you want to search for important patents of other companies that threaten the product based on important patents (documents) that describe the technology that is the basis of your product, If you can find the tendency of “Many patents to solve the B problem of Company A” or “Many applications from Company C in 1995”, narrow down focusing on such categories and reduce the labor of important patents of other companies. It will be possible to search (find) with.

  Therefore, an object of the present invention is to provide a document analysis apparatus and program capable of presenting to a user a category tendency to which a document important for the user belongs.

  The document analysis apparatus according to the embodiment includes, for each category into which a plurality of documents are classified, storage means for storing category identification information for identifying the category and documents belonging to the category in association with each other.

  The document analysis apparatus according to the embodiment includes an important document designation unit that designates the first document stored in the storage unit in response to a user operation.

  The document analysis apparatus according to the embodiment includes a first specifying unit that specifies category identification information stored in the storage unit in association with the designated first document.

  The document analysis apparatus according to the embodiment includes a second specifying unit that specifies the second document stored in the storage unit in association with the category identification information specified by the first specifying unit.

  The document analyzing apparatus according to the embodiment associates the similarity between the designated first document and the second document specified by the second specifying unit with the first document, and stores the storage unit. And similarity calculation means for calculating based on the category identification information stored in the storage means and the category identification information stored in the storage means in association with the second document.

  The document analysis apparatus according to the embodiment determines an important document candidate determining unit that determines the second document specified by the second specifying unit as an important document candidate based on the similarity calculated by the similarity calculating unit. It comprises.

  The document analysis apparatus according to the embodiment corresponds to the category identification information, out of the similarities calculated by the similarity calculation means, the importance of the category identified by the category identification information stored in the storage means In addition, there is provided importance calculation means for calculating based on the similarity between the second document determined as the important document candidate stored in the storage means and the designated first document.

  The document analysis apparatus according to the embodiment determines the category identified by the category identification information stored in the storage unit as the important category candidate based on the importance calculated by the importance calculating unit. A determination means is provided.

  The document analysis apparatus according to the embodiment includes a presentation unit that presents the category determined as the important category candidate.

1 is a block diagram showing a hardware configuration of a document analysis apparatus according to a first embodiment. The block diagram which mainly shows a function structure of the document analyzer 30 shown in FIG. The figure which shows an example of the data structure of the document memorize | stored in the document memory | storage part 22 shown in FIG. The figure which shows an example of the data structure of the information regarding the category of the root | route in the hierarchy structure of a category. The figure which shows an example of the data structure of the information regarding the child category of the category shown in FIG. 4 in the hierarchical structure of a category. The figure which shows an example of the data structure of the information regarding the child category of the category shown in FIG. 5 in the hierarchical structure of a category. The figure which shows an example of the data structure of the information regarding the child category of the category shown in FIG. 4 in the hierarchical structure of a category. The figure which shows an example of the data structure of the information regarding the child category of the category shown in FIG. 7 in the hierarchy structure of a category. The figure which shows an example of the data structure of the information regarding the child category of the category shown in FIG. 8 in the hierarchy structure of a category. 6 is a flowchart showing a processing procedure of the document analysis apparatus 30 according to the present embodiment. The figure which shows an example of the important document designation | designated screen displayed by the category display operation part 311. FIG. The figure which shows an example of the display screen after operation which designates an important document was performed in the important document designation | designated screen 100 shown in FIG. 12 is a flowchart showing a processing procedure of important document candidate extraction processing by an important document candidate extraction unit 321. The flowchart which shows the process sequence of the category vector generation process performed in an important document candidate extraction process. The flowchart which shows the process sequence of the important category candidate extraction process by the important category candidate extraction part 322. FIG. The figure which shows an example of a display screen when an important category candidate is displayed by the category display operation part 311. FIG. The flowchart which shows the process sequence of the related important category candidate extraction process by the related important category candidate extraction part 323. FIG. The figure which shows an example of a display screen when a related important category candidate is displayed by the category display operation part 311. FIG. The figure which shows an example of the display screen when the confirmation object category is selected on the display screen 300 shown in FIG. The block diagram which mainly shows the function structure of the document analysis instrumentation which concerns on 2nd Embodiment. 5 is a flowchart showing a processing procedure of the document analysis apparatus 40 according to the present embodiment. The figure which shows an example of the total result of the important document by the important document total part 411. FIG.

  Hereinafter, each embodiment will be described with reference to the drawings.

(First embodiment)
FIG. 1 is a block diagram illustrating a hardware configuration of the document analysis apparatus according to the first embodiment. As shown in FIG. 1, the computer 10 is connected to an external storage device 20 such as a hard disk drive (HDD). The external storage device 20 stores a program 21 executed by the computer 10. The computer 10 and the external storage device 20 constitute a document analysis device 30.

  FIG. 2 is a block diagram mainly showing a functional configuration of the document analysis apparatus 30 shown in FIG. As shown in FIG. 2, the document analysis apparatus 30 includes a user interface unit 31 and an important document processing unit 32. In the present embodiment, these units 31 and 32 are realized by the computer 10 illustrated in FIG. 1 executing the program 21 stored in the external storage device 20. This program 21 can be stored in advance in a computer-readable storage medium and distributed. Further, this program 21 may be downloaded to the computer 10 via, for example, a network.

  Further, the document analysis device 30 includes a document storage unit 22 and a category storage unit 23. In the present embodiment, the document storage unit 22 and the category storage unit 23 are stored in, for example, the external storage device 20.

  The document storage unit 22 stores a plurality of documents to be analyzed by the document analysis device 30. Each of the documents stored in the document storage unit 22 includes, for example, a document number (document identification information) for identifying the document and a text.

  The category storage unit 23 stores information related to a plurality of categories into which a plurality of documents stored in the document storage unit 22 are classified. The category storage unit 23 stores, for each category, a category number (category identification information) for identifying the category and a document number for identifying a document classified in the category (a document belonging to the category). Has been. A plurality of categories into which a plurality of documents stored in the document storage unit 22 are classified constitutes a hierarchical structure, for example.

  That is, in a storage unit (not shown) including the document storage unit 22 and the category storage unit 23, for each category into which a plurality of documents are classified, category identification information for identifying the category and documents belonging to the category are stored. It is stored in association.

  The document storage unit 22 and the category storage unit 23 may be realized using a file system or a database, for example.

  The user interface unit 31 includes a category display operation unit 311, an important document designation unit 312, and a category selection unit 313.

  The category display operation unit 311 refers to the category storage unit 23 and presents (displays) a category in which a plurality of documents stored in the document storage unit 22 are classified and a document belonging to the category to the user. Further, the category display operation unit 311 has a function of accepting the user's operation on the category and the document presented to the user. Thereby, the user can select and specify the category and the document presented by the category display operation unit 311.

  The category display operation unit 311 may be realized by a technique such as a graphical user interface (GUI).

  The important document designating unit 312 designates the document (first document) stored in the document storage unit 22 in accordance with the user operation received by the category display operation unit 311. The important document specifying unit 312 may specify a plurality of documents. The document specified by the important document specifying unit 312 is, for example, a document important for the user specified by the user (hereinafter referred to as an important document).

  The category selection unit 313 is a category (for example, presented by the category display operation unit 311) identified by the category number stored in the category storage unit 23 in accordance with a user operation received by the category display operation unit 311. (Category) is selected.

  The important document processing unit 32 includes an important document candidate extraction unit 321, an important category candidate extraction unit 322, and a related important category candidate extraction unit 323.

  The important document candidate extraction unit 321 uses the document (important document) designated by the important document designation unit 312 and selects the important document candidate (hereinafter, important document) from among a plurality of documents stored in the document storage unit 22. A function of extracting a document candidate).

  The important document candidate extraction unit 321 specifies the category to which the important document designated by the important document designation unit 312 belongs. In this case, the important document candidate extraction unit 321 specifies the category identified by the category number stored in the category storage unit 23 in association with the document number for identifying the important document.

  Further, the important document candidate extraction unit 321 specifies a document (second document) belonging to the specified category. In this case, the important document candidate extraction unit 321 specifies the document identified by the document number stored in the category storage unit 23 in association with the category number for identifying the identified category.

  The important document candidate extraction unit 321 associates the similarity between the important document and the identified document with the category number stored in the category storage unit 23 in association with the document number for identifying the important document and the identified document. It is calculated based on the category number stored in the category storage unit 23 in association with the document number for identifying the document. In other words, the important document candidate extraction unit 321 calculates the similarity between the important document and the identified document based on the category to which each document belongs.

  Note that the important document candidate extraction unit 321 uses a category vector (described later) representing a category to which the important document and the specified document belong when calculating the similarity between the important document and the specified document.

  The important document candidate extraction unit 321 determines the identified document as an important document candidate based on the similarity between the calculated important document and the identified document.

  The important category candidate extraction unit 322 extracts a category candidate to which a document important for the user belongs (hereinafter referred to as an important category candidate) from the categories identified by the category number stored in the category storage unit 23. It has a function.

  Of the similarities calculated by the important document candidate extraction unit 321, the important category candidate extraction unit 322 and a document determined as an important document candidate by the important document candidate extraction unit 321 (hereinafter simply referred to as an important document candidate) and an important Based on the similarity to the document, the importance of the category identified by the category number stored in the category storage unit 23 is calculated. The importance level of a category is calculated based on the similarity level between the important document candidate belonging to the category and the important document.

  The important category candidate extraction unit 322 determines the category as an important category candidate based on the calculated importance of the category.

  Note that a category determined as an important category candidate by the important category candidate extraction unit 322 (hereinafter simply referred to as an important category candidate) is presented to the user via the category display operation unit 311.

  The related important category candidate extraction unit 323 is a category related to the category selected by, for example, the category selection unit 313 from the important category candidates extracted (determined) by the important category candidate extraction unit 322 (hereinafter referred to as related important category candidate). Is extracted). In this case, the related important category candidate extraction unit 322 is based on the important document candidate belonging to the category selected by the category selection unit 313 and the important document candidate belonging to the important category candidate, and the relevant important category candidate is a related important category candidate. It is determined whether or not.

  The related category candidates extracted by the related important category candidate extraction unit 323 are presented to the user by the category display operation unit 311.

  FIG. 3 shows an example of the data structure of a document stored in the document storage unit 22 shown in FIG. 3 is a document (patent document) relating to a patent application, for example.

  As shown in FIG. 3, the document stored in the document storage unit 22 includes a document number, a document name, a body text, an applicant, an application date, and an importance (information indicating). The document number is document identification information for uniquely identifying a document. The document name and the text are text data indicating the name of the document (document name) identified by the document number and the text of the document. The applicant indicates the applicant of the patent application filed by the document (patent document) identified by the document number. The filing date indicates the filing date of the patent application filed by the document (patent document) identified by the document number. The importance level indicates whether or not the document identified by the document number is designated (set) by the user as an important document, for example. The applicant and the application date included in the document are attribute data of the document (patent document).

  In the example illustrated in FIG. 3, a document 221 is stored in the document storage unit 22. The document 221 includes a document number “d0001”, a document name and body of the document 221, an applicant “Company A”, an application date “2006/01/25”, and an importance “important”.

  According to this, the applicant of the patent application filed by the document (patent document) 221 identified by the document number “d0001” is Company A, and the filing date of the patent application filed by the document 221 is January 2006. Shown to be 25 days. Further, it is indicated that the document 221 is designated by the user as an important document.

  In the example illustrated in FIG. 3, the document 221 has been described as including the importance “important”, but the importance included in the document includes “unnecessary” and “none” in addition to “important”. It is. The importance level “unnecessary” indicates that, for example, the document identified by the document number is designated by the user as an unnecessary document (unnecessary document). The importance level “none” indicates that the user has not designated (ie, has not been set) whether the document is an important document or an unnecessary document.

  Here, the importance has been described as taking the values of “important”, “unnecessary”, and “none”. However, the importance may be indicated by, for example, a numerical value in five stages according to the degree of importance. Absent.

  4 to 9 show an example of the data structure of categories (information relating to) constituting the hierarchical structure stored in the category storage unit 23 shown in FIG. As shown in FIGS. 4 to 9, the category storage unit 23 stores a category number, a parent category number, a category name, and a document number in association with each category.

  The category number is category identification information for uniquely identifying a category. The parent category number is category identification information for identifying a category (that is, a parent category) located at a higher level of the category identified by the category number in the hierarchical structure of the category. The category name is text data indicating the name of the category identified by the category number. The document number is document identification information for identifying a document belonging to (categorized) the category identified by the category number.

  FIG. 4 shows the root category in the category hierarchy. In the category illustrated in FIG. 4, the category number “c01” is stored in the category storage unit 23. Since the category shown in FIG. 4 is a root category, it does not have a parent category (that is, no parent category number is stored in association with the category number “c01”). In the example shown in FIG. 4, since the document number is not stored, the document is not classified into the category (root category) identified by the category number “c01” (that is, documents belonging to the category are not classified). Does not exist).

  FIG. 5 shows child categories of the category shown in FIG. 4 in the category hierarchical structure (categories positioned below the category shown in FIG. 4 in the category hierarchical structure). In the category shown in FIG. 5, category number “c02”, parent category number “c01”, and category name “by applicant” are stored in association with category storage unit 23.

  According to this, the parent category positioned above the category identified by the category number “c02” in the hierarchical structure of the category is the category identified by the parent category number “c01” (that is, the category shown in FIG. 4). It is shown that there is. Further, the category name (category name) identified by the category number “c02” is “by applicant”. In the example shown in FIG. 5, since the document number is not stored, the document is not classified in the category identified by the category number “c02” (that is, there is no document belonging to the category). Indicated.

  FIG. 6 shows child categories of the category shown in FIG. 5 in the category hierarchy. In the category shown in FIG. 6, category number “c04”, parent category number “c02”, category name “T company”, and document numbers “d15, d23, d36,...” Are stored in association with category storage unit 23. ing.

  According to this, the parent category positioned above the category identified by the category number “c04” in the hierarchical structure of the category is the category identified by the parent category number “c02” (that is, the category shown in FIG. 5). It is shown that there is. Further, the category name (category name) identified by the category number “c04” is “T company”. Further, it is indicated that the document identified by the document numbers “d15”, “d23”, and “d36” belongs (classified) to the category identified by the category number “c04”.

  As shown in FIG. 6, the category storage unit 23 may store conditions that should be satisfied by a document classified into a category. In the example shown in FIG. 6, the condition “applicant =“ Company T ”” is stored in the category storage unit 23. According to this, as described above, it is indicated that the document is not classified into the category identified by the category number “c04” unless the applicant included in the document is T company. That is, for example, the document 221 shown in FIG. 3 is not classified into the category identified by the category number “c04” because the applicant included in the document 221 is company A.

  In the example illustrated in FIG. 6, the document number and the condition are described as being associated with each other and stored in the category storage unit 23. However, for example, the document number may not be explicitly listed but only the condition may be stored. Absent.

  FIG. 7 shows child categories of the category shown in FIG. 4 in the category hierarchy. In the category shown in FIG. 7, category number “c03”, parent category number “c01”, and category name “by technology” are stored in association with category storage unit 23.

  According to this, the parent category positioned higher than the category identified by the category number “c03” in the category hierarchical structure is the category identified by the parent category number “c01” (that is, the category shown in FIG. 4). It is shown that there is. In addition, it is indicated that the category name (category name) identified by the category number “c03” is “by technology”. In the example shown in FIG. 7, since the document number is not stored, the document is not classified in the category identified by the category number “c03” (that is, there is no document belonging to the category). Is shown.

  FIG. 8 shows child categories of the categories shown in FIG. 7 in the category hierarchy. In the category shown in FIG. 8, the category number “c31”, the parent category number “c03”, the category name “dialogue classification”, and the document numbers “d07, d23, d58,...” Are stored in association with the category storage unit 23. ing.

  According to this, the parent category positioned higher than the category identified by the category number “c31” in the category hierarchical structure is the category identified by the parent category number “c03” (that is, the category shown in FIG. 7). It is shown that there is. In addition, the category name (category name) identified by the category number “c31” is “interactive classification”. Further, it is indicated that the document identified by the document numbers “d07”, “d23”, and “d58” belongs (classified) to the category identified by the category number “c31”.

  FIG. 9 shows child categories of the category shown in FIG. 8 in the category hierarchy. In the category shown in FIG. 9, category number “c43”, parent category number “c31”, category name “supervised classification”, and document numbers “d15, d32, d69,...” Are stored in association with category storage unit 23. Has been.

  According to this, the parent category positioned above the category identified by the category number “c43” in the category hierarchical structure is the category identified by the parent category number “c31” (that is, the category shown in FIG. 8). It is shown that there is. In addition, the category name (category name) identified by the category number “c43” is “supervised classification”. Furthermore, it is indicated that the document identified by the document numbers “d15”, “d32”, and “d69” belongs (classified) to the category identified by the category number “c43”.

  Note that the same document may belong to a plurality of categories as shown in the categories shown in FIGS. For example, the document identified by the document number “d15” belongs to both the category identified by the category number “c04” shown in FIG. 6 and the category identified by the category number “c43” shown in FIG. Have been). For example, the document identified by the document number “d23” belongs to both the category identified by the category number “c04” shown in FIG. 6 and the category identified by the category number “c31” shown in FIG. (Classified).

  Next, a processing procedure of the document analysis apparatus 30 according to the present embodiment will be described with reference to the flowchart of FIG.

  First, the category display operation unit 311 included in the user interface unit 31 displays a screen for designating (selecting) an important document (important document) for the user (hereinafter referred to as an important document designation screen) (step S1). ).

  The important document designation screen includes, for example, a category identified by a category number stored in the category storage unit 23 and a document identified by a document number stored in the category storage unit 23 in association with the category number ( The document name and text included in the

  Here, FIG. 11 shows an example of an important document designation screen displayed by the category display operation unit 311.

  In the important document designation screen 100 shown in FIG. 11, areas 101 to 103 are provided.

  In the area 101, the category identified by the category number stored in the category storage unit 23 is displayed in a hierarchical structure together with the name of the category (that is, the category name).

  In the area 102, the document names of documents belonging to the category designated by the user among the categories displayed in the area 101 are displayed. In other words, the area 102 displays a list of document names included in each of the documents identified by the document number stored in the category storage unit 23 in association with the category number identifying the category designated by the user. The The document name is included in each document stored in the document storage unit 22. In the important document designation screen 100 shown in FIG. 11, document names of documents belonging to the category “dialog classification” among the categories displayed in the area 101 are displayed.

  In the area 103, the text included in the document specified by the user among the documents (document names) displayed in the area 102 is displayed. Note that the text is included in the document stored in the document storage unit 22. In the important document designation screen 100 shown in FIG. 11, the text of a document whose document name is “document classification method and apparatus” in the list of document names displayed in the area 102 is displayed.

  When the important document designation screen 100 as described above is displayed, the user refers to the important document designation screen 100 to check the category (classification result) and the content of the document (document name, text, etc.). An operation for designating an important document (document important for the user) can be performed on the document analysis apparatus 30.

  When an operation for designating an important document is performed on the document analysis apparatus 30 by the user, the user's operation is received by the category display operation unit 311.

  The important document designation unit 312 designates an important document from the documents stored in the document storage unit 22 in accordance with the user operation received by the category display operation unit 311 (step S2).

  When an important document is designated by the important document designation unit 312, the designation result is stored in the document storage unit 22. Specifically, the importance included in the important document designated by the important document designation unit 312 stored in the document storage unit 22 is changed to “important”. As a result, it is set that the document (important document) designated by the important document designation unit 312 is an important document for the user.

  Here, FIG. 12 shows an example of a display screen after an operation of specifying an important document is performed on the important document specifying screen 100 shown in FIG.

  When the user performs an operation for designating, for example, a document (its document name) displayed in the area 102 on the important document designation screen 100 as an important document, as shown in FIG. A mark 104 indicating that the document has been designated (set) as an important document is displayed.

  In the example shown in FIG. 12, for example, a document whose document name is “document classification method and apparatus” and “text automatic classification method” is designated as an important document.

  Next, the important document candidate extraction unit 321 included in the important document processing unit 32 selects an important document from the documents stored in the document storage unit 22 based on the important document designated by the important document designation unit 312. Processing for extracting candidates (important document candidates) (hereinafter referred to as important document candidate extraction processing) is executed (step S3).

  In the important document candidate extraction process, a similarity between a vector representing a category to which the important document designated by the important document designating unit 312 belongs and a vector representing a category to which a document (to be described later) targeted for the important document candidate extraction process belongs. The degree is calculated, and important document candidates are determined based on the degree of similarity. Details of the important document candidate extraction process will be described later.

  The important category candidate extraction unit 322 is based on the important document candidate extracted by the important document candidate extraction unit 321 and the similarity (similarity between the important document candidate and the important document) calculated by the important document candidate extraction unit 321. Processing for extracting a category candidate (important category candidate) to which a document important for the user belongs from the categories identified by the category number stored in the category storage unit 23 (hereinafter referred to as important category candidate extraction processing) Is executed (step S4).

  In the important category candidate extraction process, the importance of each category identified by the category number stored in the category storage unit 23 is the similarity between the important document candidate calculated by the important document candidate extraction unit 321 and the important document. The important category candidate is determined based on the importance.

  The importance of a category is calculated as the ratio of the similarity between the important document candidate belonging to the category and the important document with respect to the sum (total value) of the similarity between the important document candidate belonging to each category and the important document. The details of the important category candidate extraction process will be described later.

  Next, the category display operation unit 311 presents (displays) the important category candidates extracted by the important category candidate extraction unit 322 (step S5). In this case, the category display operation unit 311 displays the important category candidate according to the importance of the category (important category candidate) calculated in the important category candidate extraction process. A specific example when the important category candidate is displayed will be described later.

  Here, the user can perform an operation for selecting one of the important category candidates presented by the category display operation unit 311, for example, on the document analysis apparatus 30. In this case, for example, the user selects a category (important category candidate) for which it is desired to check whether an important document is included.

  When an operation for selecting one of the important category candidates is performed on the document analysis apparatus 30, the user's operation is received by the category display operation unit 311.

  The category selection unit 313 selects one of the important category candidates extracted by the important category candidate extraction unit 322 in accordance with the user operation accepted by the category display operation unit 311 (step S6). Hereinafter, the important category candidate selected by the category selection unit 313 is referred to as a selected important category candidate.

  Next, the related important category candidate extraction unit 323 is related to the selected important category candidate from the important category candidates (important category candidates other than the selected important category candidate) extracted by the important category candidate extraction unit 322 (related) The process of extracting important category candidates (related important category candidates) (hereinafter referred to as related important category candidate extraction process) is executed (step S7).

  In the related important category candidate extraction process, based on the important document candidate belonging to the selected important category candidate and the important document candidate belonging to each of the important category candidates other than the selected important category candidate, the relevant important category candidate is the relevant important category candidate. It is determined whether or not there is. Details of the related important category candidate extraction process will be described later.

  The category display operation unit 311 presents (displays) the related important category candidates extracted by the related important category candidate extraction unit 323 (step S8). In this case, the category display operation unit 311 displays the number of important document candidates (and important documents) belonging to the related important category candidate together with the related important category candidates. The number of important document candidates and important documents belonging to the related important category candidates is determined based on the category number for identifying the relevant important category candidates and the document number for identifying the relevant important document candidates and important documents. It is specified by referring to the storage unit 23.

  Here, for example, the document analysis apparatus 30 performs an operation of selecting a category in which, for example, whether or not an important document exists is selected from the important category candidates and the related important category candidates presented by the category display operation unit 311. Can be done against.

  When an operation for selecting a category for which it is desired to confirm whether or not an important document exists is performed on the document analysis apparatus 30, the operation of the user is accepted by the category display operation unit 311.

  Here, the category display operation unit 311 determines whether an operation for selecting a category for which it is desired to confirm whether or not an important document exists has been received, that is, whether or not the category has been selected by the user (step S9). .

  When it is determined that an operation for selecting a category for which it is desired to confirm whether or not an important document exists is accepted (YES in step S9), the category selection unit 313 responds to the operation accepted by the category display operation unit 311. Then, a category (a category in which the user wants to confirm whether or not an important document exists) is selected from the important category candidates and the related important category candidates.

  Next, the category display operation unit 311 presents (displays) a document belonging to the category selected by the category selection unit 313 (step S10). In this case, the category display operation unit 311 is included in the document identified by the document number stored in the category storage unit 23 in association with the category number for identifying the category selected by the category selection unit 313. Document names are acquired from the document storage unit 22 and a list of the document names is displayed.

  When a document belonging to the category selected by the category selection unit 313 is displayed (for example, an important document specified by the important document specification unit 312 and an important document extracted by the important document candidate extraction unit 321). Candidates are displayed at the top. The important document candidates extracted by the important document candidate extraction unit 321 are displayed in descending order of similarity to the important document (similarity calculated by the important document candidate extraction unit 321).

  The user may designate an important document from among the documents by confirming the document displayed by the category display operation unit 311 (the document belonging to the category for which the user wants to confirm whether or not the important document exists). it can. If there is an important document in the document (YES in step S11), the user designates the important document, and the document analysis apparatus 30 repeats the processing from step S3. If there is no important document (NO in step S11), the processing of the document analyzer 30 is terminated. In this way, the user designates an important document based on the important document designated by the important document designation unit 312, and the category display operation unit 311 again presents document candidates belonging to the category. By repeating such processing, the user gradually becomes clear the tendency of important documents.

  On the other hand, when it is determined that an operation for selecting a category for which it is desired to confirm whether or not an important document exists is not accepted (NO in step S9), the processing of the document analysis apparatus 30 is ended.

  Next, a processing procedure of important document candidate extraction processing (processing in step S3 shown in FIG. 10) by the important document candidate extraction unit 321 will be described with reference to the flowchart of FIG.

  First, in the processing from step S21 to step S25, a vector (category vector) representing the category (frequency) to which the important document designated by the important document designation unit 311 described above belongs is generated.

  The important document candidate extraction unit 321 empties a category vector (hereinafter referred to as a category vector vdd ′) of a set of important documents (hereinafter referred to as an important document set Dd ′) designated by the important document designation unit 311 ( Step S21).

  Next, the important document candidate extraction unit 321 executes the following steps S22 and S23 for each important document in the important document set Dd ′. Hereinafter, an important document to be processed is referred to as an important document d ′.

  The important document candidate extraction unit 321 executes processing for generating a category vector vd ′ representing the category to which the important document d ′ belongs (hereinafter referred to as category vector generation processing) (step S22).

  In this category vector generation process, for example, a vector having each dimension identified by the category number stored in the category storage unit 23 as a dimension is generated. Details of the category vector generation process will be described later.

  Next, the important document candidate extraction unit 321 adds the category vector vd ′ of the important document d ′ generated in the category vector generation process to the category vector vdd ′ (step S23).

  The important document candidate extraction unit 321 determines whether or not the processing in steps S22 and S23 described above has been executed for all important documents in the important document set Dd ′ (step S24).

  If it is determined that the processing has not been executed for all the important documents in the important document set Dd ′ (NO in step S24), the process returns to the above step S22 and is repeated. In this case, the process is executed with the important document in the important document set Dd ′ for which the processes of steps S22 and S23 have not been executed as the important document d ′.

  In this way, the processing of steps S22 and S23 is repeated for each important document in the important document set Dd ′, thereby generating the category vector vdd ′ of the important document set Dd ′.

  When it is determined in step S24 that the processing has been executed for all the important documents in the important document set Dd ′, the important document candidate extraction unit 321 uses the value of each dimension of the generated category vector vdd ′ as the norm | vdd ′. By dividing by |, the category vector vdd ′ is normalized (step S25). As a result, the category vector vdd ′ is a vector having a norm of 1.

  Next, in the process of step S26, a set (hereinafter referred to as a target document set Dd) of documents to be processed in the following steps S27 to S31 (documents to be subjected to the above-described important document candidate extraction process) is specified. Is done.

  The important document candidate extraction unit 321 specifies a category to which each important document in the important document set Dd ′ belongs. In this case, the important document candidate extraction unit 321 uses the category number (category identified by) stored in the category storage unit 23 in association with the document number for identifying each important document in the important document set Dd ′. Identify.

  The important document candidate extraction unit 321 specifies a set of documents belonging to at least one of the categories identified by the specified category number as the target document set Dd (step S26). In this case, the important document candidate extraction unit 321 sets a document set identified by the document number stored in the category storage unit 23 in association with the identified category number as the target document set Dd.

  That is, the target document set Dd is a set of documents belonging to at least one of the categories to which each important document in the important document set Dd ′ belongs.

  Next, a document that is an important document candidate is extracted from the target document set Dd by the processes of steps S27 to S32.

  The important document candidate extraction unit 321 performs the following steps S27 to S31 for each of the documents in the target document set Dd. Hereinafter, a document to be processed is referred to as a target document d.

  The important document candidate extraction unit 321 executes a process (category vector generation process) for generating a category vector vd representing the category to which the target document d belongs (step S27). In step S27, processing similar to the category vector generation processing in step S22 described above is executed for the target document d. That is, the category vector generation process generates a vector having each category identified by the category number stored in the category storage unit 23 as a dimension.

  The important document candidate extraction unit 321 normalizes the category vector vd by dividing the value of each dimension of the generated category vector vd by the norm | vd | (step S28). As a result, the category vector vd is a vector whose norm is 1.

  Next, the important document candidate extraction unit 321 calculates the similarity between the category vector vdd ′ normalized in step S25 and the category vector vd normalized in step S28 (hereinafter referred to as similarity s). (Step S29). The similarity s is, for example, a cosine value of the category vector vdd ′ and the category vector vd.

  The important document candidate extraction unit 321 determines whether or not the calculated similarity s is greater than or equal to a predetermined value (hereinafter referred to as a threshold) (step S30).

  When it is determined that the similarity s is greater than or equal to the threshold (YES in step S30), the important document candidate extraction unit 321 determines the target document d as an important document candidate.

  In this case, the important document candidate extraction unit 321 associates the target document d (document number for identifying) with the similarity s calculated in step S29 and stores it in the list (variable list) (step S31).

  On the other hand, when it is determined that the similarity s is not greater than or equal to the threshold (NO in step S30), the following process in step S31 is not executed.

  The important document candidate extraction unit 321 determines whether or not the processing in steps S27 to S31 described above has been executed for all the documents in the target document set Dd (step S32).

  When it is determined that the processing has not been executed for all the documents in the target document set Dd (NO in step S32), the process returns to the above step S27 and is repeated. In this case, the processing is executed with the document in the target document set Dd for which the processing of steps S27 to S31 has not been executed as the target document d.

  On the other hand, when it is determined that the processing has been executed for all the documents in the target document set Dd (NO in step S32), the important document candidate extraction processing is ended.

  Note that each of the documents stored in the list in step S31 described above is an important document candidate extracted in the important document candidate extraction process. The list in which the important document candidate and the similarity are stored in step S31 (hereinafter referred to as an important document candidate list) is used in the important category candidate extraction process described below.

  Next, a processing procedure of the category vector generation process (the processes in steps S22 and S27 shown in FIG. 13) executed in the above-described important document candidate extraction process will be described with reference to the flowchart in FIG.

  Hereinafter, a document that is a target of category vector generation processing will be described as a document d. Note that the document that is the target of the category vector generation process in step S22 shown in FIG. 13 is the important document d ′. Further, the document that is the target of the category vector generation process in step S27 shown in FIG. 13 is the target document d.

  First, the important document candidate extraction unit 321 empties a category vector (hereinafter referred to as a category vector vd) representing a category to which the document d belongs (step S41).

  Next, the important document candidate extraction unit 321 includes the category to which the document d belongs (that is, the category identified by the category identification number stored in the category storage unit 23 in association with the document number for identifying the document d). For each of the above, the following steps S42 and S43 are executed. Hereinafter, the category to be processed is referred to as category c.

  The important document candidate extraction unit 321 refers to the document storage unit 22 and the category storage unit 23 to determine whether the category c is an unnecessary category (hereinafter referred to as an unnecessary category) (step S42).

  Here, the unnecessary category refers to a category in which importance “unnecessary” is set for all documents belonging to the category. That is, the processing in step S42 specifies the document number stored in the category storage unit 22 in association with the category number for identifying the category c, and is identified by the document number stored in the document storage unit 22. It is executed by referring to the importance included in the document to be executed.

  When it is determined that the category c is not an unnecessary category (NO in step S42), the important document candidate extraction unit 321 sets the category c as one dimension in the category vector vd and sets the value of the dimension as 1 (step S43). .

  Next, the important document candidate extraction unit 321 determines whether or not the processing in steps S42 and S43 described above has been executed for all categories to which the document d belongs (step S44).

  If it is determined that processing has not been performed for all categories to which the document d belongs (NO in step S44), the process returns to step S42 described above and is repeated. In this case, the process is executed with the category to which the document d, for which the processes of steps S42 and S43 have not been executed, as the category c.

  On the other hand, if it is determined that the process has been executed for all categories to which the document d belongs (YES in step S44), the category vector generation process ends.

  In addition, when it determines with the category c being an unnecessary category in above-mentioned step S42, the process of step S43 is not performed but the process of step S44 is performed.

  The category vector of the document d (category vector representing the category to which the document d belongs) vd generated by the above-described category vector generation processing has each of the categories to which the document d belongs and not an unnecessary category as one dimension. , A vector whose value of each dimension is 1.

  Here, the important document candidate extraction process described above will be specifically described. Here, it is assumed that the first to fourth important documents are designated by the important document designation unit 311. Further, it is assumed that the first important document belongs to categories A, B, C, D, and E. It is assumed that the second important document belongs to categories A, C, F, G, and H. It is assumed that the third important document belongs to categories B, C, E, G, and I. It is assumed that the fourth important document belongs to categories A, C, E, G, and I.

  In this case, the category vector vdd ′ of the important document set Dd ′ (that is, the category vector representing the frequency of the category to which each of the first to fourth important documents belongs) is “Category A: 3, Category B: 2, Category C: 4, Category D: 1, Category E: 3, Category F: 1, Category G: 3, Category H: 1, Category I: 2 ". The category vector vdd ′ of the important document set Dd ′ is the sum of the category vectors (vd ′) of the first to fourth important documents.

  In the important document candidate extraction process, each of the documents belonging to at least one of the categories A to I to which each of the first to fourth important documents belongs is set as a target document, and the category vector (vd) of the target document is set. Generated. Next, the similarity s between the category vector vd of the target document and the category vector vdd ′ of the important document set Dd ′ is calculated, and when the similarity s is equal to or greater than a threshold, the target document is determined as an important document candidate ( Extracted).

  Here, it is assumed that the first to third important document candidates are extracted in the important document candidate extraction process.

  The first important document candidate is a document belonging to the categories A, C, G, and E, and the similarity between the category vector vd of the first important document candidate and the category vector vdd ′ of the important document set Dd ′ is For example, it is 0.86 or more at 0.86. Accordingly, the first important document candidate is extracted in the important document candidate extraction process.

  The second important document candidate is a document belonging to the categories A, C, F, G, and E, and the similarity between the category vector vd of the second important document candidate and the category vector vdd ′ of the important document set Dd ′. The degree is, for example, 0.83, which is equal to or greater than the threshold value. Therefore, the second important document candidate is extracted in the important document candidate extraction process.

  The third important document candidate is a document belonging to categories B, C, D, and H, and the similarity between the category vector vd of the third important document candidate and the category vector vdd ′ of the important document set Dd ′. The degree is, for example, 0.64, which is equal to or greater than a threshold value. Therefore, the third important document candidate is extracted in the important document candidate extraction process.

  That is, in the important document candidate extraction process, documents belonging to many of the same categories as the category to which the important document belongs, that is, documents belonging to the same viewpoint as the important document are extracted as important document candidates.

  Therefore, a document that belongs to many categories other than the categories A to I to which each of the first to fourth important documents belongs is determined by the category vector vd of the document and the category vector vdd ′ of the important document set Dd ′. The similarity is low and is not extracted as an important document candidate.

  Next, the processing procedure of the important category candidate extraction process (the process of step S4 shown in FIG. 10) by the important category candidate extraction unit 322 will be described with reference to the flowchart of FIG.

  In the important category candidate extraction process, as will be described later, the importance of each category stored in the category storage unit 23 is calculated, and based on the importance, the category candidate to which the important document belongs (from the category) ( Important category candidates) are extracted.

  In the important category candidate extraction process, the important document candidate list described above is used. As described above, the important document candidate list includes the important document candidate (document number for identifying) and the similarity between the category vector of the important document candidate and the category vector of the important document set (hereinafter simply referred to as the important document list). (Similarity and notation of document candidates) are stored in association with each other.

  First, the important category candidate extraction unit 322 obtains scores (hereinafter referred to as scores sdc) of a set of all categories identified by category numbers stored in the category storage unit 23 (hereinafter referred to as category set Dc). Empty (step S51). The score sdc of the category set Dc is used for calculating the importance of the category as will be described later.

  Next, the important category candidate extraction unit 322 performs the following processes of steps S52 to S56 for each category in the category set Dc. Hereinafter, the category to be processed is referred to as category c.

  The important category candidate extraction unit 322 empties the score of the category c (hereinafter referred to as a score sc) (step S52).

  Next, the important category candidate extraction unit 322 executes the following processes of steps S53 and S54 for each of the documents belonging to the category c. A document belonging to category c is a document identified by a document number stored in category storage unit 23 in association with a category number for identifying category c. Hereinafter, a document to be processed is referred to as a document d.

  The important category candidate extraction unit 322 determines whether or not the document d is stored in the important document candidate list, that is, whether or not the document d is an important document candidate (step S53).

  If it is determined that the document d is an important document candidate (YES in step S53), the important category candidate extraction unit 322 associates the document d (important document candidate) with the similarity stored in the important document candidate list. The degree (that is, the similarity of the important document candidate) is added to the score sc of category c (step S54).

  On the other hand, when it is determined that the document d is not an important document candidate (NO in step S53), the process in step S54 is not executed.

  The important category candidate extraction unit 322 determines whether or not the processes in steps S53 and S54 described above have been executed for all documents belonging to the category c (step S55).

  If it is determined that processing has not been performed for all documents belonging to category c (NO in step S55), the process returns to step S53 described above and is repeated. In this case, the processing is executed with the document belonging to the category c for which the processing of steps S53 and S54 has not been executed as the document d.

  In this way, the process of steps S53 and S54 is repeated for each of the documents belonging to category c, thereby calculating the score sc of the category c. That is, the score sc of the category c is the total similarity of the important document candidates belonging to the category c.

  When it is determined in step S55 that the processing has been executed for all documents belonging to category c, the important category candidate extraction unit 322 adds the score sc of category c to the score sdc of category set Dc (step S56).

  Next, the important category candidate extraction unit 322 determines whether or not the processing in steps S52 to S56 described above has been executed for all categories in the category set Dc (step S57).

  If it is determined that processing has not been performed for all categories in the category set Dc (NO in step S57), the process returns to step S52 described above and is repeated. In this case, the process is executed with the category for which the processes of steps S52 to S56 have not been executed as the category c.

  In this way, the process of steps S52 to S56 is repeated for each of the categories in the category set Dc, whereby the score sdc of the category set Dc is calculated. That is, the score sdc of the category set Dc is the total score (sc) of each category in the category set Dc.

  If it is determined in step S57 that the processing has been executed for all categories in the category set Dc, the important category candidate extraction unit 322 performs the following steps S58 to S61 for each category in the category set Dc. Execute. Hereinafter, the category to be processed is referred to as category c.

  The important category candidate extraction unit 322 calculates the importance of the category c using the score sdc of the category set Dc and the score sc of the category c (step S58). Here, the importance of category c is calculated by sc / sdc. That is, the importance of the category c is the ratio of the score sc of the category c to the score sdc of the category set Dc (the total value of the scores of the categories in the category set Dc).

  Next, the important category candidate extraction unit 322 determines whether or not the calculated importance level of the category c is equal to or greater than a predetermined value (hereinafter referred to as a threshold value) (step S59). The threshold value used in the process of step S59 is, for example, 0.4 (40%).

  When it is determined that the importance level of the category c is not equal to or higher than the threshold value (NO in step S59), the following process in step S62 is executed.

  On the other hand, when it is determined that the importance level of the category c is equal to or higher than the threshold (YES in step S59), the important category candidate extraction unit 322 determines the category c as an important category candidate (step S60).

  Next, the important category candidate extraction unit 322 determines a background color when presenting the category c (the category c determined as the important category candidate) to the user according to the calculated importance of the category c. (Step S61). Note that a plurality of types of background colors determined by the important category candidate extraction unit 322 are prepared according to the importance of the category c.

  For example, when the importance (ratio) of the category c is 0.8 (80%) or more, the important category candidate extraction unit 322 has the importance of 0.6 or more and less than 0.8 (60% or more and less than 80%). In this case, the background color is determined in three stages when the degree of importance is 0.4 or more and less than 0.6 (40% or more and less than 60%). Note that the number of types of background colors and the range of importance of the category c can be appropriately changed.

  As will be described later, the degree (degree) of the important document belonging to the category c is presented to the user based on the background color determined by the important category candidate extraction unit 322.

  Next, the important category candidate extraction unit determines whether or not the above-described processing in steps S58 to S61 has been executed for all categories in the category set Dc (step S62).

  When it is determined that the processing has not been executed for all categories in the category set Dc (NO in step S62), the process returns to the above-described step S58 and is repeated. In this case, the process is executed with the category for which the processes of steps S58 and S59 have not been executed as category c.

  As described above, the processes of steps S58 to S61 are repeated for each of the categories in the category set Dc, whereby an important category candidate is determined (extracted) from the category set Dc.

  On the other hand, when it is determined that the process has been executed for all categories in the category set Dc (YES in step S62), the important category candidate extraction process is terminated.

  Here, when the important category candidate extraction process ends as described above, the category display operation unit 311 presents (displays) the important category candidates extracted in the important category candidate extraction process to the user.

  FIG. 16 shows an example of a display screen when an important category candidate is displayed by the category display operation unit 311.

  On the display screen 200 shown in FIG. 16, the category identified by the category number stored in the category storage unit 23 is displayed in a hierarchical structure together with the category name.

  Among the categories displayed on the display screen 200, categories 201 to 207 having a background color are extracted as important category candidates. Note that categories other than these categories 201 to 207 are categories that are not extracted as important category candidates.

  In the example illustrated in FIG. 16, the categories 201 to 207 that are important category candidates have different background colors (for example, density) in three stages as described above. This background color is assumed to be darkly displayed in proportion to the importance of the categories 201 to 207, for example.

Here, the category “operation” 205 is a category having high importance because the background color is the darkest among the categories 201 to 207. That is, the category “operation” 205 is a category having a high degree (degree) of important documents in the categories 201 to 207, while the category “T company” 201, the category “N company” 203, and the category “1996”. 207 is a category with low importance because the background color is the lightest among the categories 201 to 207. That is, the category “Company T” 201, the category “Company N” 203, and the category “1996” 207 are categories with a low degree (degree) of important documents belonging to the categories 201 to 207.

  Here, on the screen on which the important category candidates are displayed as described above (that is, the display screen 200 shown in FIG. 16), for example, the user makes the important category candidates (here, the category 201) to the document analysis device 30. To 207) can be performed.

  When such an operation is performed by the user, the category display operation unit 311 accepts the operation, and the category selection unit 313 selects one of the important category candidates according to the operation accepted by the category display operation unit 311. Select one category.

  When one important category candidate is selected by the category selection unit 313, the related important category candidate extraction unit 323 executes the related important category candidate extraction process (the process of step S7 shown in FIG. 10) as described above.

  Next, a processing procedure of related important category candidate extraction processing by the related important category candidate extraction unit 323 will be described with reference to a flowchart of FIG.

  In the following description, the important category candidate selected by the category selection unit 313 (one of the important category candidates displayed by the category display operation unit 311) is set as the selected important category candidate cc.

  The related important category candidate extraction unit 323 is other than the selected important category candidate cc among the important category candidates extracted by the important category candidate extraction unit 322 (that is, the category whose background color displayed by the category display operation unit 311 is not colorless). The following steps S71 to S73 are executed for each of the important category candidates. Hereinafter, an important category candidate to be processed is referred to as an important category candidate c.

  The related important category candidate extraction unit 323 includes the number of important candidate documents and important documents belonging to both the important category candidate c and the selected important category candidate cc, and the number of important candidate documents and important documents belonging to the important category candidate c. Using the important candidate documents and the number of important documents belonging to the selected important category candidate cc, it is determined whether or not the important category candidate c is statistically significant (step S71).

  The related important category candidate extraction unit 323 tests whether the important category candidate c is statistically significant by, for example, a chi-square test.

According to the chi-square test in this case, the chi-square statistic (χ) has a chi-square distribution (3.84) with a significance level of 5% and a chi-square distribution (3.84) with a significance level of 2 degrees ( If it is smaller than 6.63), it becomes statistically significant. On the other hand, when the chi-square statistic (χ) is larger than the chi-square distribution (3.84) with a significance level of 5% and the chi-square distribution (6.63) with a significance level of 1% and a significance level of 1%. Is not statistically significant. The χ square statistic (χ) is calculated by the following formula.

  In this equation, x11 is the number of important document candidates and important documents that belong to both the important category candidate c and the selected important category candidate cc. a1 is the number of important document candidates and important documents belonging to the important category candidate c. b1 is the number of important document candidates and important documents belonging to the selected important category candidate cc. n is the number of all important document candidates and important documents. X12 is a1-x11, x21 is b1-x11, and x22 is n-a1-x21.

  When it is determined by the chi-square test that the important category candidate c is statistically significant (YES in step S71), the related important category candidate extraction unit 323 selects the related important category candidate c as the related important category candidate (selected important category). The category is related to the candidate cc) (step S72).

  Next, the related important category candidate extraction unit 323 refers to the category storage unit 23, the important category candidate c determined as the related important category candidate, and the important document candidate and the important document belonging to the important category candidate c (that is, The number of important document candidates and important documents identified by the document number stored in the category storage unit 23 in association with the category number for identifying the important category candidate c is returned as a return value list (hereinafter referred to as related important items). Stored in the category candidate list) (step S73).

  Note that the important category candidates stored in the related important category candidate list in step S73 are related important category candidates extracted by the related important category candidate extraction process.

  On the other hand, when it is determined in step S71 described above that the important category candidate c is not statistically significant by the χ square test, the processes in steps S72 and S73 are not executed.

  The related important category candidate extraction unit 323 determines whether or not the processes in steps S71 to S73 described above have been executed for all important category candidates other than the selected important category candidate cc (step S74).

  When it is determined that the process is not executed for all the important category candidates other than the selected important category candidate cc (NO in step S74), the process returns to the above-described step S71 and is repeated. In this case, the process is executed with an important category candidate other than the selected important category candidate cc for which the processes of steps S71 to S73 are not executed as the important category candidate c.

  On the other hand, when it is determined that the process has been executed for all the important category candidates other than the selected important category candidate cc (YES in step S74), the related important category candidate extraction process ends.

  When the related important category candidate extraction process ends as described above, the important category candidates (related important category candidates extracted by the related important category candidate extraction process) stored in the related important category candidate list and the relevant The important document candidates belonging to the important category candidates and the number of important documents are presented (displayed) to the user by the category display operation unit 311.

  FIG. 18 shows an example of a display screen when related important category candidates are displayed by the category display operation unit 311. The display screen 300 shown in FIG. 18 is subjected to the related important category candidate extraction process after the user performs an operation of selecting the category “operability” 205 on the display screen 200 shown in FIG. It is a screen when the related important category extracted in the related important category candidate extraction process is displayed.

  In the display screen 300 shown in FIG. 18, related important category candidates are displayed in an area 301 near the category “operability” 205 which is a selected important category candidate.

  In the example illustrated in FIG. 18, the category “Company H” 202, the category “interaction classification” 204, and the category “1995” 206 are displayed as related important category candidates related to the category “operability” 205.

  In addition, the number of important document candidates and important documents belonging to the category are displayed in the vicinity of the category “Company H” 202, the category “Dialogue classification” 204, and the category “1995” 206, which are related important category candidates. Has been. For example, at the top of the category “Company H” 202, “30” is displayed as the number of important document candidates and important documents belonging to the category “Company H” 202.

  Here, the user has a category (here, “Company H” 202, category “Dialogue classification” 204, and category “1995” 206) related to the category “operability” 205 displayed on the display screen 300. ), It is possible to perform an operation of selecting a category (hereinafter referred to as a confirmation target category) for which it is desired to confirm whether or not an important document exists. Thereby, for example, the user searches for an important document only from the important category candidate (here, the category “operability” 205) selected on the display screen 200 shown in FIG. 16 or the category “operability” 205. It is also possible to select whether to search for important documents from the categories related to (here, “Company H” 202, category “dialog classification” 204, and category “1995” 206).

  When the operation for selecting the confirmation target category is performed by the user, the document (document name) belonging to the category is displayed.

  Here, FIG. 19 shows an example of the display screen when the category “operability” 205 and the category “Company H” 202 are selected as the confirmation target categories on the display screen 300 shown in FIG.

  In the display screen 300 shown in FIG. 19, for example, the document names of the documents belonging to the category “operability” 205 and the category “Company H” 202 selected as the confirmation target category in the area 302 provided in the upper right of the display screen 300. A list of will be displayed.

  In this case, in the area 302, the important documents among the documents belonging to the category “operability” 205 and the category “Company H” 202 are displayed at the top, the important document candidates are displayed next, and finally the other documents. (In other words, important documents and documents other than important document candidates) are displayed. The important document candidates are displayed in the order of the above-described similarity (similarity with the important document).

  The user can perform an operation of designating the important document candidate as an important document with respect to the document analysis apparatus 30 by referring to the important document candidate displayed on the display screen 300 shown in FIG. When an operation for designating an important document candidate is performed on the document analysis apparatus 30 by the user, the importance included in the designated important document candidate is changed to “important” in the document storage unit 22 as described above. Thus, the important document candidate is set as an important document.

  As described above, in the present embodiment, the similarity between the important document specified in accordance with the user's operation and the document that is the target of the important document candidate extraction process is calculated, and the important document candidate is determined based on the similarity. Extracted (determined). In the present embodiment, the importance of the category is calculated based on the similarity between the important document candidate belonging to the category and the important document among the calculated similarities, and extracted based on the importance of the category. The (determined) important category candidates are presented to the user.

  Thereby, in this embodiment, since the tendency of the category to which the document important for the user belongs can be presented to the user, the user includes the tendency of the important document (that is, the important document is included). It becomes possible to notice such a category). That is, in the present embodiment, the user can search for important documents efficiently with little leakage, and the labor for finding the important documents can be reduced.

  In the present embodiment, the category is presented (displayed) to the user by changing the background color of the category according to the importance of the category (important category candidate). Thereby, in this embodiment, the user can know the category to which the important document belongs by the degree (degree) of the important document belonging to the category, and can find the important document more efficiently. It becomes possible.

  In the present embodiment, important category candidates (related important category candidates) related to the important category candidates selected according to the user's operation are extracted (determined), and the relevant important category candidates are extracted from the user. Can be presented. Thereby, in this embodiment, not only the important category candidate selected according to the user's operation but also the important category candidate related to the important category candidate is presented to the user. It is possible to discover important documents without omission.

  In the present embodiment, the description has been made on the assumption that an important document is designated according to a user operation. However, for example, when a category is designated according to the user operation, all documents belonging to the category are important. The configuration may be specified as a document.

(Second Embodiment)
Next, a second embodiment will be described with reference to FIG. FIG. 20 is a block diagram mainly showing a functional configuration of the document analysis apparatus according to the present embodiment. The same parts as those in FIG. 2 described above are denoted by the same reference numerals, and detailed description thereof is omitted. Here, parts different from FIG. 2 will be mainly described.

  The hardware configuration of the document analysis apparatus according to this embodiment is the same as that of the first embodiment described above, and will be described with reference to FIG. 1 as appropriate.

  As shown in FIG. 20, the document analysis apparatus 40 according to the present embodiment includes an important document processing unit 41. In the present embodiment, the important document processing unit 41 is realized by the computer 10 illustrated in FIG. 1 executing the program 21 stored in the external storage device 20.

  The important document processing unit 41 includes an important document totaling unit 411. The important document totaling unit 411 performs cross tabulation on the important documents among the documents stored in the document storage unit 22 using a plurality of classification axes specified by the user. The important document processing unit 411 counts the number of important documents belonging to (all) a plurality of categories corresponding to each of the plurality of classification axes designated by the user.

  The result of counting by the important document totaling unit 411 is presented to the user via the category display operation unit 311.

  Next, a processing procedure of the document analysis apparatus 40 according to the present embodiment will be described with reference to the flowchart of FIG.

  First, steps S81 to S90 corresponding to the steps S1 to S10 described above are executed.

  Note that the user operates the document analysis device 40, for example, an important document (third document) from among the documents presented in step S90 (that is, important document candidates belonging to the important category candidate and the related important category candidate). ) Can be specified. In this case, the importance included in the document designated as the important document in the document storage unit 22 is changed to “important”.

  Here, for example, the user designates an important document from the documents presented in step S90, and after all the important documents are confirmed, an operation for instructing the aggregation of the important documents is performed on the document analysis apparatus 40. Can be done. In the operation for instructing the aggregation of important documents, the user designates (selects), for example, two classification axes.

  The category display operation unit 311 accepts an operation when an operation for instructing the aggregation of important documents is performed by the user.

  The category display operation unit 311 determines whether an operation for instructing the aggregation of important documents has been accepted, that is, whether there is an instruction to aggregate the important documents (hereinafter referred to as an aggregation instruction) from the user. (Step S91).

  When it is determined that there is a totaling instruction from the user (YES in step S91), the important document totaling unit 411 selects an important document (a document whose importance is “important”) stored in the document storage unit 22, Cross tabulation is performed on the two classification axes designated by the user in the tabulation instruction (step S92).

  In this case, the important document totaling unit 411 refers to the category storage unit 23, and each of the categories corresponding to one of the two classification axes (the viewpoint indicated by) designated by the user and the other classification axis. The number of important documents belonging to both of the categories corresponding to the classification axis is tabulated for each category combination.

  Here, the classification axis indicates a viewpoint common to a plurality of categories. For example, if the document (important document) stored in the document storage unit 22 is a patent document, for example, “by applicant” and “by application year” are designated as the classification axis. The category corresponding to the classification axis “by applicant” includes, for example, the categories “Company T” and “Company H”. The category corresponding to the classification axis “by application year” includes, for example, the categories “1995” and “1996”.

  Specifically, the user can select a category having a child category (that is, a parent category) as a classification axis from among a plurality of categories constituting the hierarchical structure. The category corresponding to the classification axis in this case is a child category positioned below the category (parent category) selected by the user.

  Here, FIG. 22 shows an example of the result of counting important documents by the important document totaling unit 411. In FIG. 22, “by applicant” (vertical axis) and “by application year (horizontal axis)” are designated as classification axes.

  In the example illustrated in FIG. 22, categories corresponding to the classification axis “by applicant” include the categories “Company T”, “Company H”, “Company N”, and “Company M”. On the other hand, the classification axis “by application year” includes categories “1995”, “1996”, “1997”, “1998”, and the like.

  In the tabulated result of important documents shown in FIG. 22, for example, it is shown that the number of important documents belonging to both the category “T company” and the category “1995” is ten. Further, for example, the number of important documents belonging to both categories in the category “Company H” and the category “1996” is 30. Further, since there is no important document among the documents belonging to the category “M company”, each of the category “M company” and the categories “1995”, “1996”, “1997”, and “1998”. It is shown that the number of important documents belonging to both categories is zero.

  Returning to FIG. 21 again, the category display operation unit 311 presents (displays) the totaled result by the important document totaling unit 411 to the user (step S93).

  By presenting the summary results of the important documents as shown in FIG. 22 to the user, the user can easily grasp the tendency of the important documents.

  Specifically, according to the result of tabulating important documents shown in FIG. 22, for example, important documents belonging to both the category “Company H” and the category “1996” (that is, important documents filed in 1996 by Company H). It is easy to grasp that there are the most patent documents).

  Note that when the result of aggregation of important documents shown in FIG. 22 is presented (displayed) to the user, the background color may be changed according to the number of important documents. This makes it possible to easily grasp the tendency of important documents.

  Returning to FIG. 21 again, if it is determined in step S89 that an operation for selecting a category for which it is desired to confirm whether or not an important document exists is not accepted, the processing in step S91 described above is executed. If it is determined in step S91 that there is no aggregation instruction from the user, the process ends.

  As described above, in the present embodiment, since the aggregation result of important documents belonging to two classification axes (a plurality of categories corresponding to each) specified by the user can be presented to the user, the user Makes it possible to easily grasp the tendency of important documents (category to which the document belongs).

  In the present embodiment, it has been described that the result of aggregation (cross tabulation) of only important documents among documents stored in the document storage unit 22 is displayed on, for example, a two-axis map. In addition, the total result of the important document candidates extracted by the important document candidate extraction unit 322 may be displayed, and further, the total result of documents other than the important document and the important document candidate may be displayed.

  According to the first and second embodiments described above, it is possible to provide a document analysis apparatus and a program capable of presenting a tendency of a category to which a document important for a user belongs to the user.

  Note that the present invention is not limited to the above-described embodiments as they are, and can be embodied by modifying the constituent elements without departing from the scope of the invention in the implementation stage. Further, various inventions can be formed by appropriately combining a plurality of constituent elements disclosed in the above embodiments. For example, some components may be deleted from all the components shown in each embodiment. Furthermore, constituent elements over different embodiments may be appropriately combined.

  DESCRIPTION OF SYMBOLS 10 ... Computer, 20 ... External storage device, 22 ... Document storage part, 23 ... Category storage part, 30, 40 ... Document analysis apparatus, 31 ... User interface part, 32, 41 ... Important document processing part, 311 ... Category display operation 312 ... important document designation part, 313 ... category selection part, 321 ... important document candidate extraction part, 322 ... important category candidate extraction part, 323 ... related important category candidate extraction part, 411 ... important document aggregation part.

Claims (5)

Storage means for associating and storing category identification information for identifying the category and documents belonging to the category for each category into which a plurality of documents are classified;
An important document designating unit for designating a first document stored in the storage unit in response to a user operation;
First specifying means for specifying category identification information stored in the storage means in association with the designated first document;
Second specifying means for specifying a second document stored in the storage means in association with the category identification information specified by the first specifying means;
Category identification information stored in the storage means in association with the first document, the similarity between the designated first document and the second document specified by the second specifying means, and Similarity calculation means for calculating based on category identification information stored in the storage means in association with the second document;
An important document candidate determining means for determining the second document specified by the second specifying means as an important document candidate based on the similarity calculated by the similarity calculating means;
The importance of the category identified by the category identification information stored in the storage means is stored in the storage means in association with the category identification information among the similarities calculated by the similarity calculation means. Importance calculating means for calculating based on the similarity between the second document determined as the important document candidate and the designated first document;
Based on the importance calculated by the importance calculating means, an important category candidate determining means for determining a category identified by the category identification information stored in the storage means as an important category candidate;
Presenting means for presenting a category determined as the important category candidate. The importance calculation means includes:
For each category identification information stored in the storage means, the second document determined as the important document candidate stored in the storage means in association with the category identification information and the designated first First calculating means for calculating a similarity to a document;
Second calculation means for calculating a total value of similarities calculated for each category identification information by the first calculation means;
The ratio of the similarity calculated for each category identification information by the first calculation unit to the total similarity calculated by the second calculation unit is expressed as the importance of the category identified by the category identification information. And a third calculating means for calculating as
The document analysis apparatus according to claim 1, wherein the presenting unit presents the category determined as the important category candidate according to the ratio calculated by the third calculating unit as the importance. Selection means for selecting a category determined as the important category candidate in accordance with a user operation;
The selected based on the second document determined as the important document candidate stored in the storage means in association with each of the category identification information for identifying the category determined as the important category candidate Determination means for determining whether the category determined as the important category candidate other than the selected category is a category related to the category selected by the user;
The presenting means sets a category determined as the important category candidate other than the selected category determined to be a category related to the selected category as a category related to the selected category. The document analysis apparatus according to claim 1, further presenting. A further counting means,
The important document designating unit designates a third document stored in the storage unit in association with category identification information for identifying the presented category according to the operation of the user,
The counting means refers to the storage means and counts the number of the designated first and third documents belonging to a plurality of categories designated by the user,
The document analysis apparatus according to claim 1, wherein the presenting unit further presents the total result. For each category into which a plurality of documents are classified, an external storage device having storage means for storing category identification information for identifying the category and documents belonging to the category in association with each other, and a computer using the external storage device A program executed by the computer in a document analysis device comprising:
In the computer,
Designating a first document stored in the storage means in response to a user operation;
Identifying category identification information stored in the storage means in association with the designated first document;
Identifying a second document stored in the storage means in association with the identified category identification information;
The category identification information and the second document stored in the storage means are associated with the similarity between the designated first document and the identified second document in association with the first document. Calculating based on category identification information stored in the storage means in association with each other;
Determining the identified second document as an important document candidate based on the calculated similarity;
The important document candidate stored in the storage means in association with the category identification information among the calculated similarities, the importance of the category identified by the category identification information stored in the storage means Calculating based on the similarity between the second document determined as and the designated first document;
Determining a category identified by category identification information stored in the storage means as an important category candidate based on the calculated importance;
Presenting the category determined as the important category candidate.

Images