JP2009075777A - Document processing system and method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a graph structure in which a node showing one document and a node showing the other document can be bidirectionally traced even when one document is not referenced from the other document although the other document is referenced from the one document. <P>SOLUTION: Each of a plurality of documents including reference information to another document includes a central node corresponding to the document and peripheral nodes corresponding to the other document linked to the document by the reference information, and a graph structure in which the central node has a node name showing the document and the peripheral nodes have a node name showing the other document is generated, and a plurality of graph structures corresponding to each of the documents are searched, and second peripheral nodes having the node name showing the second document are added to a second graph structure in which a first document corresponding to the first peripheral nodes in the first graph structure among the plurality of graph structures is included as the central node, and a second document corresponding to the central node of the first graph structure is not included as the peripheral nodes. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a document processing system for generating and displaying a graph structure representing a reference relationship between documents.

  In recent years, electronic document data having a reference relationship with each other, such as the World Wide Web, Wikipedia (registered trademark), and Weblog, has been increasing.

  As a method for displaying the reference relationship between documents in an easy-to-understand manner for the user, there is known a method in which each document is illustrated by a node and the reference relationship between documents is illustrated by a graph expression.

  For example, in Patent Document 1, a graph structure composed of clickable nodes is presented to the user, and the node that the user pays attention to and clicked ("focus" entity) is rearranged in the center of the graph, and detailed information on the node is displayed. A technique for developing on a screen is disclosed.

  Non-Patent Document 1 extends the function of Wikipedia, an encyclopedia that anyone on the Internet can write. For example, the relationship between the text “London and the text England” is “is capital of”. A framework has been proposed that allows authors to explicitly describe the knowledge "linked by relationship" in the text of England, and an example of a graph display based on this is shown.

  To display a document with reference information as a graph with a small amount of computation, analyze a single document and generate a single graph structure with this document as the central node (the “focus” entity). It can be presented to the user. In other words, when the total number of documents (number of nodes) is N, the presence or absence of a reference relationship is checked for a combination of all nodes (N (N-1 / 2)), and a huge graph structure including all nodes is constructed. Rather, it is to analyze the document N times and prepare N disjoint graph structures.

  As this application, for example, it is conceivable to present a graph structure obtained by analyzing a document related to “actress E” to a user who has performed a search using the keyword “actress E”.

For example, if there is a reference to "male talent G" in a document about actress E, "actress E" will be the central node of this graph structure, "male talent G" will be the peripheral node, and the central node to the peripheral node An arrow indicating the reference relationship to is displayed. Here, if the user clicks on the node of “Men Talent G” next time, this time, a new graph centered on “Men Talent G” obtained by analyzing the document about “Men Talent G” The structure can be presented.
JP 2005-122703 A "Volkel, et al .: Semantic Wikipedia, Proceedings of ACM WWW 2006, Edinburgh, Scotland, 2006"

  In the clickable graph structure presentation interface described above, for example, when a document corresponding to a node clicked by a user is displayed in a separate window and an advertisement is presented in that window, there are as many companies as possible that the business is advertising revenue. The more you click on the node, the more you earn. Therefore, it is desirable to let the user click on as many nodes as possible by freely tracing the graph structure.

  However, in the approach of obtaining a single graph structure by analyzing a single document as described above, a problem occurs when the reference relationship is not bidirectional.

  Specifically, in the above-mentioned examples of “actress E” and “male talent G”, there may be a case where there is no reference to “actress E” in the document regarding “male talent G”. In particular, such a phenomenon occurs frequently when multiple authors write texts according to the most important policy like Wikipedia.

  In this case, the graph structure having “male talent G” as the central node does not include the node of “actress E”. Therefore, when the user clicks on the peripheral node “male talent G” in the graph structure with “actress E” as the central node and transitions to the graph structure with “male talent G” as the central node, the user is “actress E”. There is no means to return to this node, and this is difficult to understand intuitively, which may prevent the user from clicking.

  As described above, conventionally, when one document is referred to the other document, but the other document is not referred to the one document, the graph structure having the other document as a central node is used. Since the peripheral node representing the one document is missing, the graph structure having the other document as the central node cannot recognize the presence of the one document (referring to the other document) There was a problem that it was not possible to transition to a graph structure with the document as the central node.

  Therefore, in view of the above problems, the present invention refers to the other document from one document, but the node representing the one document even when the other document is not referenced from the other document. An object of the present invention is to provide a graph structure capable of bidirectionally tracing between a node representing a document and a node representing the other document.

(1) The document processing system
Document collection means for collecting a plurality of documents including reference information to other documents,
First storage means for storing the collected documents;
For each document stored in the first storage means, the central node includes a central node corresponding to the document and a peripheral node corresponding to another document linked to the document by the reference information. Has a node name representing the document, and the peripheral node generates a graph structure having a node name representing the other document;
Second storage means for storing a plurality of graph structures corresponding to each of the plurality of documents;
The first graph structure includes, as a central node, a first document corresponding to a first peripheral node in the first graph structure among the plurality of graph structures stored in the second storage unit. Graph structure complementing means for adding a second peripheral node having a node name representing the second document to a second graph structure that does not include the second document corresponding to the central node of the second node as a peripheral node;
First search means for searching for a graph structure from the second storage means;
First display means for displaying the retrieved graph structure;
including.

  (2) The graph complementing unit adds, to the second graph structure, a second peripheral node having reference information to the first graph structure and a node name representing the second document.

  (3) Each peripheral node in the graph structure has reference information to another graph structure having a document corresponding to the peripheral node as a central node.

  (4) The first search means is a document corresponding to the peripheral node based on the node name or the reference information of the peripheral node selected from the graph structure displayed by the first display means. Another graph structure having a central node as a central node is retrieved from the second storage means.

  Even if one document refers to the other document, but the other document does not refer to the one document, the node representing the one document and the node representing the other document A graph structure that can be traced in both directions can be provided.

  Embodiments of the present invention will be described below with reference to the drawings.

  FIG. 1 shows a configuration example of a document processing system 1 according to the present embodiment.

  In FIG. 1, the document processing system 1 roughly includes a graph structure generation unit 100, a search unit 101, a first storage unit 102, a second storage unit 103, and a third storage unit 104.

  The graph structure generation unit 100 includes a document collection unit 11, a document analysis unit 12, and a complement unit 13. The search unit 101 includes a search request processing unit 14, a first search unit 15, a second search unit 16, and a third search unit 101. The search part 17 and the search result presentation part 18 are included.

  Although FIG. 1 shows a case where the document processing system 1 is configured as a single server device connected to the network NW, the present invention is not limited to this case, and may be configured with a plurality of server devices.

  The document collection unit 11 includes, for example, a crawler used for a search engine, and references information (URI) to other documents such as a document on the World Wide Web (WWW) or a Wikipedia (registered trademark) document. Documents with embedded information (information for identifying information resources such as (Uniform Resource Identifier) and URL (Uniform resource Locator)) (documents with reference information) are collected from the network NW and stored in the first storage unit 102 To do. The document collection unit 11 receives a document with reference information periodically or as needed from a specific site on the network NW, receives it, and stores it in the first storage unit 102. Also good.

  The first storage unit 102 stores a plurality of collected documents together with identification information (for example, URI, URL, etc.) for uniquely identifying the documents.

  The document analysis unit 12 analyzes each document collected by the document collection unit 11 and stored in the first storage unit 102 to generate graph structure data representing a graph structure corresponding to the document, and the second Store in the storage unit 103.

  One graph structure data is generated for each document stored in the first storage unit 102, and each graph structure data is a central node representing the document and another document linked to the document by reference information. And a graph structure including an edge from the central node to the peripheral node is described using a structured document description language such as XML (Extensible Markup Language).

  For each document stored in the first storage unit 102, the document analysis unit 12 generates the graph structure data having the document as a central node. Each piece of graph structure data includes identification information (for example, URI, URL, etc.) for uniquely identifying the graph structure data, and a plurality of pieces stored in the first storage unit 102 (and the third storage unit 104). Are stored in the second storage 103 together with the identification information of the document corresponding to the graph structure data (the central node).

  Details of the document analysis processing of the document analysis unit 12 will be described later.

  For each graph structure data stored in the second storage unit 103, the complementing unit 13 receives another document that is not linked to the document that is the central node of the graph structure (from the other document, the document Add a description of the peripheral node that represents Details of the complementing process in the complementing unit 13 will be described later.

  The search request processing unit 14 receives a search request from the user terminal 2 connected to the network NW, and activates the first search unit 15, the second search unit 16, and the third search unit 17.

  The first search unit 15 searches for graph structure data from the second storage unit 103 based on keywords and the like.

  The second search unit 16 searches the first storage unit 102 for a document corresponding to the graph structure data (the central node in the graph).

  The third storage unit 104 stores all documents collected from the network NW and their indexes.

  The third search unit 17 searches for documents from the third storage unit 104 based on keywords and the like.

  In FIG. 1, the third storage unit 104 and the first storage unit 102 for storing the document for generating the graph structure data are separately shown, but the third storage unit 104 is not necessarily required. Instead, the first storage unit 102 may be provided. In this case, the third search unit 17 searches for a document from the first storage unit 102 based on the keyword, and the graph structure generation unit 100 selects all of the documents stored in the first storage unit 102. Alternatively, graph structure data is generated for some documents.

  In the case of the configuration shown in FIG. 1, the document collection unit 11 may collect only the documents for generating the graph structure data from the network NW, and the graph storage data is generated in the first storage unit 102. Only documents to be stored may be stored.

  The search result presentation unit 18 is a document that is a search result in the first search unit 15, graph structure data that is a search result in the second search unit 16, and a document that is a search result in the third search unit 17. Or the list thereof is returned to the user terminal 2 as the search request source. As a result, in the user terminal 2, the search result is displayed on the display of the user terminal 2 by the browser.

  Next, the processing operation of the document analysis unit 12 will be described with reference to the flowchart shown in FIG.

  The document analysis unit 12 takes out documents with reference information one by one from the first storage unit 102, and performs the following processing on all the documents to be processed stored in the first storage unit 102.

  First, the document analysis unit 12 initializes the variable i to “1” (step S1). Next, the document analysis unit 12 extracts the i (= 1) -th document among the plurality of processing target documents stored in the first storage unit 102, and the node of the central node is extracted from the i-th document. A character string used as a name is extracted (step S2). The node name of the central node is, for example, the title, heading, title, etc. of the i-th document.

  Next, the document analysis unit 12 detects another document linked to the document by reference information from the i-th document (step S3). When another document linked to the i-th document is detected, a character string or an image in which reference information for the other document is embedded is represented from the i-th document. Extracted as node names of peripheral nodes (step S4).

  Further, a character string used as a label to be added to the edge from the central node to the peripheral node is extracted from the i-th document (step S5).

  For example, context information such as a character string within a certain range centering on a portion where reference information to the other document in the i-th document is embedded (for example, to the other document in the i-th document Are extracted as labels to be added to the edges of the other documents. Alternatively, the title of a chapter or section including a character string or an image in which reference information for the other document is embedded in the i-th document is extracted as a label to be added to the edge of the other document.

  Then, the document analysis unit 12 obtains the node name of the central node representing the i-th document, the node names of the peripheral nodes representing other documents linked to the i-th document, the center obtained as described above. Graph structure data corresponding to the i-th document including the labels of the edges from the nodes to the peripheral nodes is generated. The graph structure data includes a central node having a node name representing the i-th document, a peripheral node having a node name representing another document linked to the i-th document, and an edge from the central node to the peripheral node. The graph structure including the label is described using a structured document description language such as XML (Extensible Markup Language). The document analysis unit 12 stores the generated graph structure data in the second storage unit 103 (step S7), and increments the variable i by 1 (step S8).

  The processes from step S2 to step S8 are repeated until the value of the variable i becomes larger than the total number of documents to be processed stored in the first storage unit 102 (step S9).

  FIG. 3 shows an example of a document to be processed (document stored in the first storage unit 102) in the document analysis processing of FIG. 2, and a document with reference information describing information on “actress E” It is.

  In the document shown in FIG. 3, other documents are linked to the character strings such as “Actress Y”, “Actor S”, “Male Actor I”, etc. (reference information is embedded). In FIG. 3, a character string to which other documents are linked is surrounded by “[[” and “]]”. Actually, Wikipedia (registered trademark) indicates a character string in which reference information is embedded in this format. As shown in FIG. 3, the character string or image in which the reference information is embedded is surrounded by tags and symbols such as “[[” and “]]”. It is easy to detect a document corresponding to and a character string used as its node name.

  From the document shown in FIG. 3, “actress E” corresponding to the title, heading, and title of the document is extracted as the node name of the central node, and “actress Y”, “actor S”, and “male talent G” are respectively Extracted as the node name of the peripheral node. In addition, a character string including “Actress Y” in the document is extracted as a label of an edge from the central node to the peripheral node “Actress Y”, and a character string including “Actor S” in the document is the central node. Is extracted as the label of the edge to the peripheral node “Actor S”, and the character string including “Men Talent G” in the document is extracted as the label of the edge from the central node to the peripheral node “Men Talent G” The

  As a result, a graph structure having a central node of “actress E” as shown in FIG. 4 is obtained from the document of FIG. In FIG. 4, “Actress E” is a central node, and “Actress Y”, “Actor S”, and “Male talent G” linked to the document are represented as peripheral nodes. In addition, at the edge between nodes, context information (such as a character string around a portion where reference information is embedded) extracted from the document of “actress E” by the document analysis unit 12 or the title of a chapter / section is displayed. It is added as a label.

  The document analysis unit 12 generates graph structure data representing the graph structure as shown in FIG. 4 from the document of FIG. By applying a style sheet or the like created in advance to the graph structure data, the graph structure shown in FIG. 4 is displayed on the display of the user terminal 2.

  In FIG. 3, one label is added to each edge. However, the present invention is not limited to this, and the document analysis unit 12 extracts a plurality of context information for each edge as labels from the document to be processed. However, a plurality of labels or one label including a plurality of context information may be added to the edge. Further, two types of labels, a context information label and a chapter / section title label, may be added to one edge.

  In addition, when many documents are linked to one document, a graph structure including many peripheral nodes is formed. If this is displayed on the user terminal 2, it may be difficult for the user to see. Considering such a case, the document analysis unit 12 may limit the number of peripheral nodes linked to one central node to a predetermined number. For example, among the documents linked to the document corresponding to the central node, up to a predetermined number of documents in the order of appearance from the top are employed as the peripheral nodes.

  Further, for example, when the graph structure shown in FIG. 4 is presented to many user terminals 2, when the document of “actress Y” is frequently referred to in the document related to “actress E” in FIG. Reflecting the information, the document analysis unit 12 displays the node of “actress Y” in the graph structure of FIG. 4 to be large, or increases the thickness of the edge connecting “actress E” and “actress Y”. In addition, the graph structure data may be modified.

  The label of each edge may not be normally displayed but may be generated as graph structure data that pops up when the user places the mouse pointer on the edge, for example.

  FIG. 5 shows an example of a document related to “male talent G” linked to “actress E” in FIG. From this document, a graph structure as shown in FIG. 6 including the central node “male talent G” and its peripheral nodes “male talent K”, “actor T”, and “band P” is generated from the document analysis process described above. Is done.

  In the document of FIG. 5, the character string “affiliation band” is surrounded by a tag “==”, which is a tag representing a chapter or section title. In fact, Wikipedia (registered trademark) clearly indicates the title of a chapter in this format. Since the chapter and section titles are surrounded by predetermined tags and symbols, it is easy to detect the chapter and section titles from the entire document.

  In the graph structure of FIG. 6, “male talent K”, “actor T”, and “band P” are obtained as peripheral nodes of the central node “male talent G”. Here, as the labels of edges to “male talent K” and “actor T”, as in FIG. 4, in the document of FIG. 5, the character string to which the document corresponding to the peripheral node is linked and before and after The character string is adopted. On the other hand, as the label of the edge to “band P”, the character string “belonging band” which is the title of the chapter / section extracted from the document of FIG. 5 is adopted.

  Therefore, in the graph structure of FIG. 6, it is determined whether the edge label toward each peripheral node is a character string around the location where the document corresponding to the peripheral node is linked or the title of a chapter / section. As can be distinguished, when the label is the title of a chapter / section, it is displayed surrounded by a thick line.

  FIG. 7 shows graph structure data which is source data of the graph structure of FIG. The graph structure data in FIG. 7 is an example of the graph structure data generated by the document analysis unit 12 described in XML and having “male talent G” as a central node.

  The graph structure data in FIG. 7 includes a description portion C1 of a central node and description portions C2 and C3 of a plurality of (for example, two here) peripheral nodes. In the description portion C1 of the central node, the node name of the central node is described with a tag name <entryname>. In the description part C1 and C2 of the peripheral node, the node name of the peripheral node is described with the tag name <entryname>. In addition, in order to link the graph structure data having the peripheral node as the central node to the peripheral node, reference information (here, reference information to the graph structure data having the peripheral node as the central node with a tag name <link>) (Identification information) is described. Further, the label of the edge from the central node to the peripheral node is described with a tag name <labeltext>, and information indicating whether this label is the context information or the chapter / section title type, It is described with a tag name <labeltype>. For example, if the label is context information, a value “TITLE” is described, and if the label is a chapter / section title, a value “CONTEXT” is described.

  In the description part C2 of the first peripheral node, there are two edge labels from the central node to the peripheral node, the first of which is the context information label, and the second is the chapter / section. It turns out that it is a title. It can be seen that the description portion C3 of the second peripheral node has one edge label from the central node to the peripheral node, which is a label of context information.

  Further, in the description part of the peripheral node, information indicating whether or not the description part is added by a complementing process described later is described with a tag name <added_node>. Here, since the description parts C1 and C2 of the two peripheral nodes are not added by the complementing process (because they exist in the graph structure data), the value “NO” is described. .

  When the graph structure shown in FIG. 4 or FIG. 6 is displayed on the user terminal 2, when the user selects a peripheral node in the graph structure (when clicked), the source of the graph individual structure Reference information (identification information) included in the <link> element in the description part of the selected peripheral node in the data (graph structure data) is passed from the search request processing unit 14 to the first search unit 15. Based on the identification information, the first search unit 16 searches the second storage unit 103 for graph structure data specified by the identification information, that is, graph structure data having the peripheral node as a central node. To do. The retrieved graph structure data is displayed on the display of the user terminal 2 by the search result presentation unit 17.

  For example, when the graph structure of FIG. 4 is displayed on the user terminal 2, when the user clicks “male talent G”, the display of the graph structure of FIG. 4 is switched to the display of the graph structure of FIG. That is, the graph structure of FIG. 4 is transitioned to the graph structure of FIG.

  In this case, in the graph structure of FIG. 4, “male talent G” exists as a peripheral node of the central node “actress E”, and the graph structure of FIG. 4 can transition to the graph structure of FIG. In the graph structure of FIG. 6, the central node “male talent G” does not have a peripheral node “actress E”. Therefore, when the graph structure of FIG. The action of returning to the graph structure of 4 cannot be selected. This is because the document “Actress E” as shown in FIG. 3 is not linked to the document corresponding to the central node of the graph structure of FIG. 6 (see FIG. 5).

  What solves this problem is a complementing process in the complementing unit 13. Next, the processing operation of the complementing unit 13 will be described with reference to the flowchart shown in FIG.

  The complementing unit 13 extracts the graph structure data one by one from the second storage unit 103, and performs the following processing for all the graph structures to be processed stored in the second storage unit 103.

  First, the complement unit 13 initializes the variable i to “1” (step S21). Next, the complementing unit 13 extracts the i (= 1) -th graph structure data (graph structure data (i)) from the plurality of graph structure data to be processed from the second storage unit 103 (step S22). ).

  The complement unit 13 initializes the variable j to “1” (step S23). Reference information (graph structure data (j) reference information) included in the description part of the j-th peripheral node from the beginning in the graph structure data (i), with the peripheral node as the central node Based on the identification information), the graph structure data (j) specified by the identification information is extracted from the second storage unit 19 (step S24).

  Here, a graph structure represented by the graph structure data (i) is a graph structure (i), and a graph structure represented by the graph structure data (j) is a graph structure (j).

  In step S25, it is checked whether the graph structure data (i) is linked to the graph structure data (j). This can be done by checking whether or not the graph structure data (j) includes a description part of a peripheral node including reference information to the graph structure data (i).

  Since the graph structure data (j) is linked to the graph structure data (i), if the graph structure data (i) is further linked to the graph structure data (j), the graph structure (i) to the graph Transition to the structure (j) is possible, and transition from the graph structure (j) to the graph structure (i) is also possible. On the other hand, when the graph structure data (i) is not linked to the graph structure data (j), the graph structure (i) can be transitioned to the graph structure (j). Transition to structure (i) is not possible.

  When the graph structure data (i) is not linked to the graph structure data (j), the process proceeds from step S5 to step S26, and the graph structure for linking the graph structure data (i) to the graph structure data (j). A description of a new peripheral node having the same node name as the central node name of data (i) is added (step S26).

  The description of the new peripheral node corresponds to the reference information (identification information of the graph structure data (i)) to the graph structure data (i) and the graph structure data (j) in the graph structure data (i). And edge labels included in the description part of the peripheral node.

  For example, the source data having the graph structure shown in FIG. 4 is the graph structure data (i), and the graph structure data (j) having the j-th peripheral node in the graph structure data (i) as the central node is shown in FIG. 7 is assumed. FIG. 9 shows a description part of the peripheral node “male talent G” in the graph structure data (i).

  In this case, in step S26, the identification information of the graph structure data (i), the node name of the central node in the graph structure data (i), the label contained in the description part of FIG. Based on the above, a description of a new peripheral node including these pieces of information is added to the graph structure data (j) in FIG.

  FIG. 10 shows graph structure data (j) to which a description of a new peripheral node is added. In FIG. 10, the node name “actress E” of the central node of the graph structure data (i) is described with the tag name <entryname> in the description portion D1 of the added new peripheral node. Further, in order to link the graph structure data (i) to the new peripheral node, the identification information of the graph structure data (i) is described with the tag name <link>, and the new peripheral node and the central node As the label of the edge between and the type thereof, the same label and type as described in the <labeltext> tag of FIG. 9 or the <labeltype> tag are described. Furthermore, a value “YES” is described in the <added_node> tag, which indicates that the description part of the new peripheral node has been added by the complementing process.

  A display example of the graph structure data shown in FIG. 10 is shown in FIG. In the graph structure shown in FIG. 11, a new peripheral node “actress E” corresponding to the description portion D1 of FIG. 10 is added to the graph structure shown in FIG. In addition, a label of “as a hero of a hero who is jealous of male talent G” is added to the edge between the central node “male talent G” and a new peripheral node “actress E”. As described above, the character string used is not extracted from the document “male talent G” in FIG. 5 but extracted from the document “actress E” in FIG. 3. , The label of the edge between the central node “Actress E” and the peripheral node “Male Talent G”.

  In the display example shown in FIG. 11, the peripheral node “actress E” added to the graph structure by the above-described complementing process has the value of the <added_node> tag “YES”, so that the central node of the graph structure is displayed. It is displayed in a display form different from normal peripheral nodes representing documents linked in the corresponding document. For example, in FIG. 11, the edge between the central node and the normal node is indicated by a solid line, but the edge between the central node and the added new peripheral node “actress E” is It is displayed with a dotted line to clearly indicate that it is a new added peripheral node. In addition to displaying the edge between the central node and the added new peripheral node with a dotted line, it may be displayed in a thick actual battle or in a color different from other edges.

  As described above, the display form of the peripheral node added by the above complement processing in the graph structure is the edge between the central node and the added peripheral node, and the edge between the central node and the normal node as described above. In addition to the display form different from the above display form, the added peripheral node may be displayed in a display form different from the color, shape, font, line, etc. of the normal peripheral node.

  In any case, when the graph structure is displayed, the added peripheral node (including its edge) is added to the normal peripheral node (its edge) so that the added peripheral node can be distinguished from the normal peripheral node. It is desirable that the display format is different from the display format of

  Returning to the description of FIG. 8, if the graph structure data (i) is linked to the graph structure data (j) in step S25, the process proceeds to step S27, and the variable j is incremented by one. In step S26, the process proceeds to step S27 even after a description of a new peripheral node for linking the graph structure data (i) to the graph structure data (j) is added.

  The processes in steps S24 to S27 are repeated until the value of the variable j becomes larger than the total number of peripheral nodes in the graph structure data (i) (step S28). When the value of the variable j is larger than the total number of neighboring nodes in the graph structure data (i), the process proceeds to step S29, where the variable i is incremented by 1, and the value of the variable i is stored in the second storage unit 103. The processing from step S22 to step S29 is repeated until it becomes larger than the total number of graph structure data to be processed (step S30).

  In the complementing process shown in FIG. 8, the graph structure (i) transits to the graph structure (j) through the jth peripheral node, but the graph structure (j) has a transition that can transit to the graph structure (i). When the node is not included (step S25), a new peripheral node for transition to the graph structure (i) is added to the graph structure (j) (step S26). When the graph structure (j) includes a peripheral node that can transition to the graph structure (i) (step S25), the processing is moved to the next peripheral node as it is. Not limited to this, the complementary processing may be performed as shown in FIG.

  In the complementing process shown in FIG. 12, if the graph structure (j) includes a peripheral node that can transition to the graph structure (i) in step S25 of the complementing process in FIG. After the processing, the process proceeds to step S27.

  In the processing of step S41 and step S42, in order to display the label of the edge from the j-th peripheral node to the central node together with the label of the edge from the central node to the j-th peripheral node in the graph structure (i) Necessary information is added to the graph structure data (i).

  First, in step S41, the edge label and its type are extracted from the description part of the peripheral node to which the graph structure data (i) is linked in the graph structure data (j). That is, the value of the <labeltext> element or the <labeltype> element is extracted.

  In step S42, a <labeltext> element or <labeltype> element having information such as the label obtained in step S41 and its type as a value is added to the description part of the j-th peripheral node of the graph structure data (i). To do.

  For example, when the graph structure in FIG. 6 is the graph structure (i), the graph structure data (j) linked to the peripheral node “male talent K” includes a peripheral node that transitions to the graph structure data (i). It is assumed that “male talent G” is included. In this case, by the complementing process of FIG. 12, the label of the edge from the central node “male talent K” of the graph structure (j) to the peripheral node “male talent G” is changed to the central node “male talent K” of the graph structure (i). ”And the peripheral node“ male talent G ”. As a result, as shown in FIG. 13, when the graph structure (i) is displayed, the label of the edge from the central node “male talent G” to the peripheral node “male talent K” and the peripheral node “male talent K” are displayed. ”To the center node“ male talent G ”.

  In FIG. 13, reference to “male talent K” in the document corresponding to the central node “male talent G” and reference to “male talent G” in the document corresponding to the peripheral node “male talent K”. Both are displayed. The user may select a node with reference to these pieces of information and go back and forth on the graph structure.

  The graph structure data subjected to the complementing process as shown in FIG. 8 or 12 in the complementing unit 13 is stored in the second storage unit 103.

  In the above description, the document with reference information to be processed is not limited to Japanese, and may be a document written in any language. The node names of the central node and the peripheral nodes are not limited to person names, and may be anything such as an organization name, a title of a paper, a title of a web page, etc., as long as it can identify a document with reference corresponding to the node. For example, a single graph structure may include a node having a node name in which a person name and an organization name are mixed, or a node having a Japanese node name and a node having an English node name may be mixed. Good.

  Next, the processing operation of the search unit 101 will be described with reference to the flowchart shown in FIG.

  The keyword input from the user terminal 2 is accepted by the search request processing unit 14, and the keyword is passed from the search request processing unit 14 to the first search unit 15 and the third search unit 17 (step S100).

  Based on the keyword input from the search request processing unit 14, the first search unit 15 searches the second storage unit 103 for graph structure data in which the keyword is included in the node name of the central node ( Step S101). The retrieved graph structure data is transmitted to the user terminal 2 by the search result presentation unit 18, and the user terminal 2 displays the graph structure described in the received graph structure data on the display (step S102).

  The third search unit 17 searches the third storage unit 104 for a document including the keyword based on the keyword input from the search request processing unit 14 (step S111). For example, a document including the keyword is searched from the third storage unit 104. The retrieved document list is transmitted to the user terminal 2 by the retrieval result presentation unit 18, and the user terminal 2 displays the received retrieved document list on the display (step S112).

  FIG. 15 shows a screen display example of a graph structure displayed on the display of the user terminal 2 and a document list of search results. Here, when the user inputs the keyword “actress E” from the user terminal 2, the graph structure having “actress E” as a central node, searched by the first search unit 15 (the same graph structure as FIG. 4). ) And a list of documents searched by the third search unit 17 from the third storage unit 104 using “actress E” as a keyword.

  The display screen shown in FIG. 15 is provided with an area A1 for inputting and displaying keywords, an area A2 for displaying a graph structure, and an area A3 for displaying a list of searched documents. .

  When the user clicks on the peripheral node “male talent G” on the graph structure on the display interface as shown in FIG. 15 (step S103), the search request processing unit 14 receives this and, for example, selects the selected node. The node name of the peripheral node “male talent G” is set as a new keyword (step S104). That is, “male talent G” is displayed in place of “actress E” in the area A1 on the display screen of FIG. Further, the search request processing unit 14 passes this new keyword “male talent G” to the first search unit 15 and the third search unit 17, and the processes of steps S101 to S102 and steps S111 to S112 are performed. Is called.

  That is, in step S101, the first search unit 15 searches for the graph structure data in which the new keyword “male talent G” is included in the node name of the central node, and the searched graph structure data is the user terminal 2. 11 is displayed in the area A2 on the display screen of the user terminal 2 (step S102).

  In step S101, the graph structure data corresponding to the reference information may be retrieved from the reference information (value of the <link> element in the graph structure data) possessed by the selected peripheral node “male talent G”.

  In step S111, the third search unit 17 searches the third storage unit 104 for a document including the keyword based on the new keyword “male talent G”, and the list of searched documents is a user terminal. 2 and a list of received search result documents is displayed in the area A3 on the display screen of the user terminal 2 (step S112).

  On the other hand, when the user clicks the central node “actress E” on the graph structure on the display interface as shown in FIG. 15 (step S103), the process proceeds to step S121. In step S <b> 122, the search request processing unit 14 passes the node name “actress E” of the central node to the second search unit 16. When the search request processing unit 14 can acquire the identification information of the document corresponding to the graph structure displayed on the display screen of the user terminal 2 (that is, the document corresponding to the central node of the graph structure). (For example, when document identification information is added to or included in the graph structure data) The identification of the document together with or instead of the node name “actress E” of the central node The information is passed to the second search unit 16.

  The second search unit 16 searches for a document from the first storage unit 102 based on the node name “actress” input from the search request acquisition processing unit 14 or the identification information of the document corresponding to the node name. . Here, a document including “actress” in the title, headline, and title is searched. Alternatively, a document having the input identification information is searched. The document searched in this way is transmitted to the user terminal 2 by the search result presentation unit 18, and the user terminal 2 displays a new window different from the display screen of FIG. The received search result document is displayed.

  Although not shown in FIG. 14, when one document is selected from the list of documents displayed in the area A3 of the display screen in FIG. 15, the selected document (for example, A new window different from the display screen of FIG. 15 is displayed and displayed in this new window, and the graph structure displayed in the area A2 of the display screen of FIG. You may make a transition to the graph structure.

  In this case, when one document is clicked and selected from the list of documents displayed in the area A3 of the display screen in FIG. 15, the search request processing unit 14 identifies the selected document. (For example, URI, URL, etc.) is passed to the first search unit 15 and the third search unit 17.

  The first search unit 15 searches the second storage unit 103 for graph structure data stored in association with the document identification information input from the search request processing unit 14. The retrieved graph structure data is transmitted to the user terminal 2 and displayed in the area A2 on the display screen of the user terminal 2. The third search unit 15 searches the third storage unit 104 for a document having the identification information input from the search request processing unit 14. The retrieved document is transmitted to the user terminal 2 and displayed in a new window.

  As described above, when a desired peripheral node is selected from the graph structure displayed in the area A2 on the display screen of FIG. 15, another graph structure having the selected peripheral node as the central node is displayed. The Since the other graph structure always includes the central node of the original graph structure as a peripheral node, the original graph structure can be displayed again by selecting the peripheral node.

  By selecting a peripheral node of the displayed graph structure, the user can change to a graph structure having the peripheral node as a central node, and can follow a node (document) having a reference relationship, and the original node You can freely return to (document).

  That is, even when there is a missing node due to an asymmetric reference relationship between documents, the user can freely trace between the nodes of the graph illustrating the reference relationship between documents.

  Therefore, the user can bidirectionally navigate between many different graph structures by selecting the peripheral nodes on the displayed graph structure. Since the user can see not only the target information but also a lot of information related to it, it is possible to realize a search for a graph structure that is intriguing to the user.

  The method of the present invention described in the embodiment of the present invention (processing operation of the graph structure generation unit 100 and the search unit 10) is a program that can be executed by a computer as a magnetic disk (flexible disk, hard disk, etc.), optical disk. It can also be stored and distributed in a recording medium such as a semiconductor memory (CD-ROM, DVD, etc.).

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

1 is a diagram showing a configuration example of a document processing system according to an embodiment of the present invention. 6 is a flowchart for explaining document analysis processing in a document analysis unit. The figure which shows an example of the document containing reference information. The figure which shows the example of a display of the graph structure data produced | generated from the document of FIG. The figure which is another example of the document containing reference information, and shows the document referred from the document of FIG. The figure which shows the example of a display of the graph structure data produced | generated from the document of FIG. The figure which shows the graph structure data which are the source data of the graph structure of FIG. The flowchart for demonstrating the complement process in a complement part. The figure which showed the example of description of peripheral node "male talent G" among the graph structure data which is the source data of the graph structure of FIG. The figure which showed the example of the graph structure data to which the description of the new surrounding node was added. The figure which shows the example of a display of the graph structure data of FIG. The flowchart for demonstrating the other complementation process in a complement part. The figure which shows the example of a display of the label added between the two nodes which have a reference relation mutually by the complement process shown in FIG. The flowchart for demonstrating the processing operation of a search part. The figure which shows the example of a screen display of the graph structure displayed on the display of a user terminal, and the document list of a search result.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Document processing system, 2 ... User terminal, 11 ... Document collection part, 12 ... Document analysis part, 13 ... Complement part, 14 ... Search request process part, 15 ... 1st search part, 16 ... 2nd search part , 17 ... third search unit, 18 ... search result presentation unit, 100 ... graph structure generation unit, 101 ... search unit, 102 ... first storage unit, 103 ... second storage unit, 104 ... third storage Part

Claims (11)

Document collection means for collecting a plurality of documents including reference information to other documents,
First storage means for storing the collected documents;
For each document stored in the first storage means, the central node includes a central node corresponding to the document and a peripheral node corresponding to another document linked to the document by the reference information. Has a node name representing the document, and the peripheral node generates a graph structure having a node name representing the other document;
Second storage means for storing a plurality of graph structures corresponding to each of the plurality of documents;
A first document corresponding to a first peripheral node in the first graph structure of the plurality of graph structures stored in the second storage means as a central node; Graph structure complementing means for adding a second peripheral node having a node name representing the second document to a second graph structure that does not include the second document corresponding to the central node as a peripheral node;
First search means for searching for a graph structure from the second storage means;
First display means for displaying the retrieved graph structure;
Including document processing system. The first search means searches the second storage means for another graph structure having a peripheral node selected from the graph structure displayed by the first display means as a central node,
The document processing system according to claim 1, wherein the first display unit displays the searched other graph structure. When the first peripheral node is selected from the first graph structure displayed by the first display means, the second graph structure searched by the first search means is the first graph structure. 1 is displayed on the display means,
When the second peripheral node is selected from the second graph structure displayed by the first display means, the first graph structure searched by the first search means is the first graph structure. 3. The document processing system according to claim 2, wherein the document processing system is displayed by one display means.   The generation means includes a character string having the reference information to the other document in the document and its surrounding character string, or a title of a chapter or section including the character string having the reference information as the central node. The document processing system according to claim 1, wherein the document structure is generated in the graph structure included as a label of an edge between the peripheral nodes.   The graph structure complementing means uses the central node in the first graph structure and the first as a label of an edge between the central node in the second graph structure and the second peripheral node. 5. The document processing system according to claim 4, wherein labels of edges between peripheral nodes are added to the second graph structure. The graph structure complementing means includes:
A third graph structure having a third document corresponding to a third peripheral node in the first graph structure as a central node, and the second document corresponding to the central node in the first graph structure being When included as a peripheral node, the edge between the central node and the third peripheral node in the first graph structure corresponds to the central node and the second document in the third graph structure The document processing system according to claim 4, wherein a label of an edge between the peripheral nodes is added.   The second peripheral node in the second graph structure displayed by the first display means is a document linked to the second document corresponding to the central node of the second graph structure 4. The document processing system according to claim 3, wherein the document processing system is displayed in a display form different from that of the peripheral node corresponding to. A second search unit that searches the first storage unit for a document that includes a keyword input as a search condition or a document that includes a node name of a central node of the graph structure displayed by the first display unit;
Second display means for displaying a document searched by the second search means or a list of searched documents;
Further including
2. The document processing system according to claim 1, wherein the first search unit searches the second storage unit for a graph structure including the keyword in the node name of the central node. First storage means for storing a plurality of documents including reference information to other documents;
For each document stored in the first storage means, the central node includes a central node corresponding to the document and a peripheral node corresponding to another document linked to the document by the reference information. Has a node name representing the document, and the peripheral node generates a graph structure having a node name representing the other document;
Second storage means for storing a plurality of graph structures corresponding to each of the plurality of documents;
Graph structure complementing means for complementing each graph structure stored in the second storage means;
Retrieval means for retrieving a graph structure from the second storage means;
Display means for displaying the retrieved graph structure;
A document processing method in a document processing system including:
The graph structure complementing means includes, as a central node, a first document corresponding to a first peripheral node in the first graph structure of the plurality of graph structures, and the central node of the first graph structure A second peripheral node having reference information to the first graph structure and a node name representing the second document is added to a second graph structure that does not include the corresponding second document as a peripheral node. A graph structure completion step,
A first search step in which the search means searches the second storage means for a graph structure including a keyword input as a search condition in a node name of the central node;
A first display step in which the display means displays the retrieved graph structure;
Document processing method. The second storage unit stores the second graph structure having the first peripheral node selected from the first graph structure displayed in the first display step as a central node. A second search step for searching from the means;
A second display step in which the display means displays the searched second graph structure;
The second memory stores the first graph structure having the second peripheral node selected from the second graph structure displayed in the second display step as a central node. A third search step for searching from the means;
A third display step in which the display means displays the searched first graph structure;
The document processing method according to claim 9, further comprising: Computer
First storage means for storing a plurality of documents including reference information to other documents;
For each document stored in the first storage means, the central node includes a central node corresponding to the document and a peripheral node corresponding to another document linked to the document by the reference information. Has a node name representing the document, and the peripheral node generates a graph structure having a node name representing the other document,
Second storage means for storing a plurality of graph structures corresponding to each of the plurality of documents;
The first graph structure includes, as a central node, a first document corresponding to a first peripheral node in the first graph structure among the plurality of graph structures stored in the second storage unit. Graph structure complementing means for adding a second peripheral node having a node name representing the second document to a second graph structure that does not include the second document corresponding to the central node of the second node as a peripheral node;
First retrieval means for retrieving a graph structure from the second storage means;
First display means for displaying the retrieved graph structure;
Program to function as.

Images