JP3746233B2 - Knowledge analysis system and knowledge analysis method - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a system and method for knowledge analysis used in a knowledge management system, and more particularly to a knowledge analysis system and a knowledge analysis method for supporting knowledge analysis by searching and clustering stored documents using important words.
[0002]
[Prior art]
In recent years, introduction of groupware for sharing information among a plurality of users has been promoted mainly by companies. As typical groupware, an e-mail system, a workflow system, and the like are known, but recently, a knowledge management system for supporting sharing of knowledge and information has begun to be developed.
[0003]
This knowledge management system is for accumulating and managing personal know-how as a knowledge database in addition to Web information and electronic file information. By combining it with a search function such as natural language search, knowledge and information Efficient use is possible.
[0004]
By the way, in such a knowledge management system, how to collect and accumulate knowledge such as individual know-how is an important point. Because knowledge such as personal know-how is so-called tacit knowledge and not formalized like Web information or electronic file information, it is difficult to automatically collect and store it. is there.
[0005]
Therefore, recently, there is a demand for the development of a knowledge management system having a knowledge accumulation support function. Utilize knowledge as tacit knowledge in the same way as formal knowledge such as Web information and electronic file information by realizing a mechanism for automatically collecting and accumulating knowledge such as personal know-how Is possible.
[0006]
In addition, a knowledge management system that easily searches for knowledge and information accumulated in this way is being developed in parallel. A typical example is a natural language search system having a knowledge search support function for searching for useful knowledge and information by inputting a natural language question sentence.
[0007]
[Problems to be solved by the invention]
By the way, in this kind of knowledge management system, knowledge and information can be easily organized or viewed, or even for users who are using it for the first time, it is easy to understand what kind of knowledge information can be searched. There is a strong demand for systematization of knowledge for effective use of data. Along with this, the accumulated knowledge is structured (classified) by category.
[0008]
The process of analyzing knowledge and structuring (classifying) it is called clustering. Conventionally, words that become the axis for clustering must be directly specified as important words by the user, so that important words There was a problem that it took time and effort to list. Also, in this case, knowledge analysis and classification for clustering is performed based on the words specified by the user as important words. Therefore, depending on the words specified as important words by the user, an accurate classification result is not necessarily obtained. It may not be obtained. Such a situation occurs not only when clustering is performed, but also when a user specifies a search keyword for information search.
[0009]
The present invention has been made in consideration of such circumstances, and automatically extracts and determines important words from a document, and uses them to perform search and clustering to allow a user to specify important words. It is an object of the present invention to provide a knowledge analysis system and a knowledge analysis method capable of automatically executing accurate knowledge analysis by searching or clustering stored documents without any problem.
[0010]
[Means for Solving the Problems]
In order to solve the above problems, the present invention provides: A knowledge analysis system configured to be connectable to a plurality of client terminals via a network and supporting classification and analysis of a plurality of document information stored in a server in response to a request from each client terminal, For a plurality of stored document information, based on emphasis tags embedded in each of the plurality of document information, from each of the plurality of document information Word extraction means for extracting words having an emphasis attribute, and types of emphasis attributes of each word extracted by the word extraction means And the frequency of appearance of the word, the importance is determined for each word, the process of accumulating the determination result of the importance over the plurality of document information is executed, and the value of the accumulated importance is high The upper predetermined number of words are determined as important words of the entire document information over the plurality of document information. Important word determination means,
Determined by the important word determining means By clustering the plurality of document information by matching the important words of the entire document information with the important words determined from the document information, a cluster database in which the plurality of document information is classified and organized is created. And knowledge analysis means.
[0014]
In this knowledge analysis system, For each of a plurality of pieces of document information already accumulated, for each word based on the word extraction process, the type of emphasis attribute of each word extracted from each document by the word extraction process, and the appearance frequency of the word A process of accumulating the determination results of the importance levels over a plurality of pieces of document information, and determining a top predetermined number of words having a high accumulated importance value. It is determined as an important word for the entire document information. The plurality of pieces of document information are clustered by matching the important words of the entire document information with the important words determined from the document information. As a result, the time required for analyzing the stored document can be shortened, and clustering can be performed using important words determined in consideration of the characteristics of the entire stored document.
[0015]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the drawings.
[0016]
FIG. 1 shows the configuration of a knowledge analysis system according to an embodiment of the present invention. This knowledge analysis system is realized by a server computer 12 accessible from a plurality of client terminals 11 via a computer network 13 such as a LAN. Although not shown, the server computer 12 and the client terminal 11 each have a CPU, a main memory, a magnetic disk device as a storage device, and an input / output device having an input unit such as a keyboard and a mouse and a display unit such as a display. Is provided.
[0017]
The knowledge analysis system realized on the server computer 12 has a document information storage function for each client terminal 11, a search function for searching for a target document from a plurality of stored document information, and a plurality of stored documents. Provides a clustering function that classifies and organizes data. The clustering function classifies and organizes the stored documents, and provides the client computer 11 with the contents from the server computer 12 as an analysis result of the stored documents, thereby knowing the tendency to read from various documents collected for a certain purpose. Can be provided to the user. Here, an overview of the clustering function will be described.
[0018]
The clustering function is a cluster database in which features of each of a large amount of document information collected in the server computer 12 are analyzed, sorted into groups having similar contents, and a large amount of document information is classified and organized according to the features. It is a function to create. When executing clustering, an analysis condition designation screen as shown in FIG. 2 is provided to the client terminal 11, and various analysis conditions are designated by the user on the analysis condition designation screen. FIG. 2 shows an example in which a database named “10,000 db” and a database named “XX newspaper article” are selected as analysis target databases, and includes the following designated items.
[0019]
Analysis result name: This is a field for inputting a name for saving the analysis result.
Analysis period: This is a field for designating in which period a document created or stored should be analyzed. If there is no input, all documents stored in the database to be analyzed are analyzed.
[0020]
Refinement keyword: This is a field for inputting a search keyword for narrowing down a document to be analyzed from an accumulated document by a search. If there is no input, all documents stored in the database to be analyzed are analyzed.
-Number of filters: This field specifies the number of documents to be analyzed.
[0021]
-Number of hierarchies: This field specifies the number of hierarchies of the classification system used in clustering. Select either “1 layer” or “Multi layer”. The default is “multi-hierarchy”, and the documents are classified in a hierarchical structure by category.
-Duplicate knowledge: a field for designating whether one document is permitted to be registered in a plurality of classifications (clusters). Select either “Yes” or “No”.
[0022]
Maximum number of top clusters: This field specifies the maximum number of clusters created in the top hierarchy. If there is no input, clustering is performed with no designation.
[0023]
-Important word: This is a field for inputting a word or phrase as an important word when performing clustering.
Normally, it is necessary for the user to directly specify the important words used in the clustering. However, in the present embodiment, a function for automatically generating the important words is provided. The generated important words can be used to analyze and classify stored documents. From what point of view, the contents of stored documents are analyzed and the classification system is classified based on the key words that generate the axis of clustering. It can be said that this is the starting point for analysis and classification.
[0024]
Next, returning to FIG. 1, each element constituting the knowledge analysis system will be described.
[0025]
In the client terminal 11, a Web browser 111 is operating. By specifying a URL (Uniform Resource Locator) indicating a resource for knowledge analysis constructed on the server computer 12 from the Web browser 111, the knowledge analysis processing by the server computer 12 can be used from each client terminal 11. .
[0026]
Further, an image scanner 114 is connected to the client terminal 11, and the user uses a character recognition tool 113 to capture a paper document. Further, a document editing tool 112 is operating, and the user creates an electronic document using this. The captured image document data and the created electronic document data are described in a markup language such as SGML (Standard Generalized Markup Language), HTML (Hyper Text Markup Language), or XML (eXtensible Markup Language). After being automatically converted into a document format, it is sent to the server computer 12 via the network 13 as a registered document or as an input document for determining important words. In document information in a document format described in a markup language, the type of emphasis attribute of a word or phrase in the document is expressed by an emphasis tag. Accordingly, document information in which an emphasis tag corresponding to the emphasis attribute of each word is embedded is sent from the client terminal 11 to the server computer 12. The server computer 12 performs processing for determining an important word based on the emphasis tag in the document information.
[0027]
The knowledge analysis function of the server computer 12 is mainly performed by software such as the Web server 121, the control module 1211, the knowledge server 122, the highlight tag setting module 1221, the registration module 1222, the search module 1223, the clustering module 1224, and the software. This is realized by management information and actual data used for knowledge analysis. The management information includes login management information 1212 for performing user authentication for each client terminal 11. The actual data includes a knowledge database 1226, an intermediate processing database 1228, an analysis result database (cluster DB) 1229, an emphasis tag table 1225, and an important word table 1227.
[0028]
The emphasis tag table 1225 is for defining a relationship between an emphasis tag corresponding to each type of emphasis attribute and a weighting value (point) for the emphasis tag, and the user uses the web browser 111 of the client terminal 11. By starting the highlight tag setting module 1221, the user can set the contents of the highlight tag table 1225.
[0029]
The important word table 1227 is for registering an input document input from the client terminal 11 or an important word extracted from the document information already stored in the knowledge database 1226. The input document or the stored document The important word table 1227 is created by extracting the words enclosed by the emphasis tag, analyzing them into morphemes and breaking them into words, and scoring the words according to the type and appearance frequency of the emphasis tag.
[0030]
The emphasis tag table 1225 and the important word table 1227 are used when the clustering module 1224 creates an analysis result database (cluster DB) 1229 from the intermediate database 1228.
[0031]
The control module 1211 operates on the Web server 121 and controls the entire operation related to knowledge analysis. The knowledge server 122 and the Web server 121 which are the core programs of this knowledge analysis system, and the client terminal 11 In addition to an intermediary function with the Web browser 111, each client terminal 11 has a user authentication function when logging in to the knowledge server 122. For this user authentication, the control module 1211 manages login management information 1212. The login management information 1212 stores user IDs and passwords of the users participating in the knowledge analysis system. By this user authentication, control of permission / prohibition of access to the knowledge server 122 performed for knowledge analysis from each client terminal 11 is performed.
[0032]
The knowledge server 122 is for managing and operating the knowledge database 1226 and the analysis result database (cluster DB) 1229 in response to requests from the plurality of client terminals 11, and the analysis conditions designated by each client terminal 11 By performing the knowledge analysis, the stored document registered in the knowledge database 1226 is supported for analysis.
[0033]
Next, referring to the flowchart of FIG. 3, when important words are automatically determined from the input document information input from the client terminal 11 and the knowledge database 1226 is searched and knowledge analysis is performed using the important words. An overview of the processing will be described.
[0034]
First, based on the emphasis tag included in the target input document information (XML, HTML, SGML, etc), a phrase surrounded by the emphasis tag is extracted from the input document information (step S101). Next, each extracted phrase is subjected to morphological analysis and divided into words, whereby a word having an emphasis attribute is extracted and registered in the important word table 1227 (steps S102 and S103). Then, for each extracted word, a point for the emphasis tag is acquired from the emphasis tag table 1225, and the point is added to the important word table 1227 as a score indicating the priority (importance) of the corresponding word. (Step S104). Thus, for each extracted word, the score value expressed by the product of the appearance frequency of the word and the point corresponding to the type of the emphasis tag at that time is registered in the important word table 1227. The word groups registered in the important word table 1227 are sorted in descending order of score, and several words having higher scores are preferentially determined as important words. Of course, all the words registered on the important word table 1227 may be determined as important words.
[0035]
When a document matching the key concept of the input document information is searched from the knowledge database 1226, a search keyword is generated from each word determined as an important word, and a search such as full-text search is performed using the search keyword. Processing is executed by the search module 1223 (step S105).
[0036]
On the other hand, when the accumulated documents in the knowledge database 1226 are clustered, document analysis and classification are performed on the accumulated documents, with each word determined as an important word as the axis of clustering. Are clustered (step S106).
[0037]
Actually, the document analysis and classification processing with the key word as the axis is performed on the narrowed down document group registered in the intermediate database 1228, and the classification result is registered in the cluster database 1229. It will be. The analysis may be performed with the word itself determined as the important word as an axis, but the analysis may be performed using synonyms, synonyms derived from the word determined as the important word. In the analysis process, the features of each document to be analyzed are analyzed using important words as the starting point, and those with the same features are collected into one cluster with a classification name that accurately represents the features. . Of course, the important words themselves can also be used as classification names.
[0038]
There are various methods for analyzing the characteristics of each document. For example, the keyword is determined for each accumulated document to be analyzed by the same method as the keyword determination process from the input document described above, and how It is possible to analyze the characteristics of each of the documents by examining each of the accumulated documents to be analyzed to determine whether or not the important words are included. In this case, for example, by matching important words determined from input documents with important words determined from each document to be analyzed, clustering around important words determined from input documents can be easily performed. It becomes.
[0039]
Further, instead of determining the important word from the input document information, for example, the important word is determined from the entire analysis target document group registered in the intermediate database 1228 by the same method as the important word determination processing from the input document described above. A method is also suitable in which an entire document group to be analyzed is clustered using the determined clustering axis as a clustering axis. In this case, first, word extraction processing in steps S101 to S103 is performed for the entire document group to be analyzed, and in step S104, the type of emphasis attribute of each word extracted from each document and the word Based on the appearance frequency, scoring is performed for each word, and processing for accumulating the score over the entire document group to be analyzed is executed. The upper predetermined number of words having a high accumulated score value are determined as important words.
[0040]
In step S106, analysis and classification of the document group registered in the intermediate database 1228 are executed using the upper predetermined number of words determined as important words as the axis of clustering. Even in this case, for example, by matching the important words determined from the entire stored document with the important words determined from each of the stored documents to be analyzed, the important words determined from the entire stored document are used as axes. Clustering can be easily performed.
[0041]
Next, a document format conversion process executed by the character recognition tool 113 and the document editing tool 112 provided in the client terminal 11 will be described with reference to FIGS. In the following, a case where XML is used as document information in which an emphasis tag is embedded will be described as an example.
[0042]
FIG. 4 shows the document format before XML conversion, and FIG. 5 shows the document format after XML conversion. As shown in FIG. 4, in the document before conversion, characters are emphasized in various places, and these are word emphasis attributes. For example, <large character>, <bold character>, <italic>, <underline>, <reverse>, <quotation mark>, <background color>, and the like.
[0043]
The conversion from the document format of FIG. 4 to the document format of FIG. 5 is performed by the character recognition tool 113 and the document editing tool 112 of the client terminal 11, and is passed to the registration module 1222 of the knowledge server 122 via the Web browser 111, Registered in the database 1226. When the character recognition tool 113 is used, the emphasis attribute of a phrase applied to a paper document is converted into an emphasis tag, and XML format document information is generated. An example of conversion is shown below.
[0044]
For example, in the document of FIG. 4, the phrase “analyze enormous amount of information and knowledge” is described in bold characters, and the background of each character is colored. In this case, since two kinds of emphasis attributes such as a large character and a background color are added, when this phrase is converted into the XML document of FIG.
<Background color><Bold> Accurate analysis of huge amount of information and knowledge </ Bold></ Background color>
It becomes.
[0045]
Also, the phrase “rapid decision making” on the document of FIG. 4 is underlined, and when this phrase is converted to the XML document of FIG.
<Underline> Rapid decision making </ Underline>
It becomes.
[0046]
Thus, when a plurality of emphasis tags are added to one word, it means that the word is important. In the above example, it is determined that the former is more important than the latter.
[0047]
FIG. 6 shows an example of the emphasis tag table 1225. The emphasis tag table 1225 includes an emphasis tag field 1225-1 for setting an emphasis tag and a point field 1225-2 for setting a point corresponding to the emphasis tag. The user activates the emphasis tag setting module 1221 via the Web browser 111 of the client terminal 11, and highlights the emphasis tag corresponding to the emphasis attribute of the word and phrase that appears in the document before conversion in FIG. By registering in the table 1225, the importance (point) corresponding to each type of emphasis attribute can be set.
[0048]
In the emphasis tag table 1225 of FIG. 6, for example, 10 points are given for the emphasis tag <large character>, and 9 points for the emphasis tag <italic>, and the emphasis tag <bold> 8 points, 7 points for the emphasis tag <color character>, and 6 points for the emphasis tag <background color>.
[0049]
As for the above-mentioned words “<background color><bold> accurately analyze a large amount of information and knowledge </ bold></ background color>”, each word in the word corresponds to an emphasis tag “<bold>” Eight points and six points for the emphasis tag <background color> are added to give a total of 14 points. In this way, even when a certain word is emphasized with an emphasis attribute that is a mixture of various emphasis attributes, when the word is converted into an XML document, the mixed emphasis attribute includes a plurality of different emphasis attributes. Since it is converted into a tag, it is possible to easily calculate the point corresponding to the emphasis attribute of the word simply by adding the point for each emphasis tag.
[0050]
The default setting contents of the emphasis tag table 1225 may be determined in advance in the system, and the importance may be calculated using this.
[0051]
7 to 9 show how the important word table 1227 is generated. 7 shows a state during analysis of one document, FIG. 8 shows a state after analysis of one document, and FIG. 9 shows a state after analysis of all documents.
[0052]
7 and 8 correspond to the case of analyzing the XML document of FIG. As shown in FIG. 7, words obtained by morphological analysis of words (for example, “analyze huge amount of information and knowledge”) surrounded by emphasis tags, that is, “large”, “ “Information”, “Knowledge”, “Accuracy”, “Analysis”,... Are registered in the important word field 1227-1 of the important word table 1227, and the corresponding score field 1227-2 corresponds to each time the word appears. The values of points to be added are sequentially added. As a result of the analysis of the XML document in FIG. 5, the important word table 1227 in FIG. 8 is generated.
[0053]
When clustering the stored documents using the important words determined from the XML document in FIG. 5 as the clustering axis, the words registered in the important word table 1227 in FIG. ”,“ Share ”,“ Creation ”, etc.) are sequentially determined as important words to be prioritized in clustering. In this case, the upper predetermined number of words (all words if the number of words registered in the important word table 1227 is less than or equal to the upper limit that can be used as important words) is used as the axis of the cluster. The number of important words used may be provided as a specified value by the system, or the user may give it from the Web browser 111 of the client terminal 11.
[0054]
FIG. 9 shows an example of the important word table 1227 obtained by analyzing all the document groups to be analyzed stored in the intermediate database 1228, and shows the result of sorting in descending order of score. When determining important words from the analysis target document group stored in the intermediate database 1228, a predetermined number of words (registered in the important word table 1227) among the words registered in the important word table 1227 of FIG. If the number of words is less than or equal to the upper limit that can be used as an important word, all words) are used as cluster axes.
[0055]
Next, the processing procedure of the knowledge analysis system will be described with reference to the flowchart of FIG. Here, XML documents input from each client terminal 11 are sequentially registered in the knowledge database 1226, and when an analysis request for the knowledge database 1226 is received from the client terminal 11, important words are extracted from all the documents to be analyzed. Assume that clustering is performed by automatically extracting.
[0056]
The user of each client terminal 11 activates the control module 1211 of the web server 121 from the web browser 111 and logs in to the knowledge server 122 (step A01). Then, the knowledge server 122 emphasis tag setting module 1221 is started from the Web browser 111, and emphasis tags and points (frequency of strength) for the emphasis attribute of characters appearing in the XML document are set on the emphasis tag table 1225 (step A02). .
[0057]
Next, when a document described in a form is read using the image scanner 114 connected to the client terminal 11 and the read character is recognized by the character recognition tool 113 and converted into an XML document, or a document editing tool When an XML document is created at 112, if an emphasis attribute is added to a word, the word is enclosed in an emphasis tag and converted to an XML-style emphasis (step A03). When the processing ends, the registration module 1222 of the knowledge server 122 is activated from the Web browser 111 and the XML document is passed to the registration module 1222. The registration module 1222 registers the received XML document in the knowledge database 1226 (step A04).
[0058]
Then, the search module 1223 of the knowledge server 122 is activated from the Web browser 111, and a selection of which knowledge database to be analyzed in the knowledge database 1226 is selected (step A05). The search module 1223 searches the selected knowledge database using the search keyword given by the user, thereby creating an intermediate database 1226 consisting only of the knowledge narrowed down by the search keyword given by the user from the selected knowledge database ( Step A06).
[0059]
The search module 1223 scans each XML document at the time of creating the intermediate database 1228, finds a phrase enclosed by the emphasis tag registered in the emphasis tag table 1225, performs morphological analysis, and determines the points of the word and the emphasis tag as important words. Register in the table 1227. Then, every time the word appears, the corresponding points are added to the word score, and further sorted in descending order of score to create an important word table 1227 (step A07). When the clustering module 1224 inputs the intermediate database 1228 to perform clustering, the important words corresponding to the system default number or the user-specified number of important words are arranged in descending order from the important word table 1227 in which important words are arranged in the order of score. The cluster database 1229 is created by classifying and organizing the document group of the intermediate database 1228 by taking out and using it for clustering (step A08).
[0060]
FIG. 11 is a flowchart illustrating in detail the processing performed in step A07 of FIG. 10, and shows a procedure for creating the keyword table 1227.
[0061]
When creating the intermediate database 1228, the search module 1223 first analyzes each document extracted from the knowledge database 1226 and creates an important word table 1227. The process will be described. First, the important word table 1227 is cleared (step B01), and positioned to the first XML document in the knowledge database 1226 (step B02). Then, a phrase enclosed by one or more emphasis tags is extracted from the XML document (step B03), the extracted phrase is subjected to morphological analysis, a noun is extracted, and is entered into the important word table 1227 (step B04). Next, the points of these emphasis tags are taken out from the emphasis tag table 1225 and added to the score field 1227-2 of the corresponding word on the important word table 1227 (step B05) until the analysis of the XML document is completed. Step B03 to step B05 are repeated (step B06).
[0062]
When the analysis of the first XML document is completed (YES in step B06), the next XML document is positioned (step B07), and steps B03 to B07 are repeated until the analysis of all XML documents is completed (NO in step B08). If completed (YES in step B08), the important word field 1227-1 of the important word table 1227 is sorted in descending order of the score 1227-2 and the result is reflected (step B09).
[0063]
FIG. 12 is a flowchart illustrating in detail the processing in step A08 of FIG. 10, and is a flowchart of processing in which the clustering module 1224 creates the cluster database 1229.
[0064]
In order for the user to receive clustering authority control, the Web browser 111 of the client terminal 11 makes a login request to the control unit 1211 of the server computer 12 (step C01). Then, the control unit 1211 of the server computer 12 accesses the login management information 1212 to check whether or not the user ID and password input by the user are registered (step C02), and whether or not to permit this login. User authentication is performed to determine (step C03). If the user ID and password are not registered in the login management information 1212 and the login fails (NO in Step C04), the control unit 1211 returns a login failure to the Web browser 111 of the client terminal 11 through the Web server 121. This process is terminated (step C05).
[0065]
On the other hand, when the user ID and password are registered in the login management information 1212 and the login is successful (YES in step C04), the knowledge database to be analyzed is selected (step C06), and the important word table 1227 is used. When a word is extracted (step C07), the clustering module 1224 executes clustering based on the extracted important word to create a cluster database 1229 (step C08).
[0066]
In the created cluster database 1229, each document is classified and arranged according to its features in a classification system determined based on important words. It is possible to provide the user with the analysis result as to whether the document is registered as knowledge. Further, the user can designate a specific classification item based on the analysis result, and can perform further information search for the classification item.
[0067]
In this way, in this knowledge analysis system, focusing on the fact that words with emphasis attributes are words that are key concepts in the document, important words are determined from the XML document based on the emphasis tags, Clustering processing based on this important word is made possible.
[0068]
Since all the functions of the knowledge analysis system of this embodiment are realized by a computer program, the computer program can be stored in a computer-readable storage medium, and the computer program can be connected to a computer network through the storage medium. The effect similar to that of the present embodiment can be obtained simply by installing and executing the program on a normal computer.
[0069]
Further, the present invention is not limited to the above-described embodiment, and various modifications can be made without departing from the scope of the invention in the implementation stage. Further, the above embodiments include inventions at various stages, and various inventions can be extracted by appropriately combining a plurality of disclosed constituent elements. For example, even if some constituent requirements are deleted from all the constituent requirements shown in the embodiment, the problem described in the column of the problem to be solved by the invention can be solved, and the effect described in the column of the effect of the invention Can be obtained as an invention.
[0070]
【The invention's effect】
As described above, according to the present invention, it is possible to perform accurate knowledge analysis by performing information retrieval and clustering with automatically determined important words.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a system configuration of a knowledge analysis system according to an embodiment of the present invention.
FIG. 2 is a view showing an example of an analysis condition designation screen used in the knowledge analysis system of the embodiment.
FIG. 3 is an exemplary flowchart illustrating an outline of a procedure of knowledge analysis processing in the knowledge analysis system according to the embodiment;
FIG. 4 is an exemplary view for explaining a document format before XML conversion in the knowledge analysis system of the embodiment;
FIG. 5 is an exemplary view for explaining a document format after XML conversion in the knowledge analysis system according to the embodiment;
FIG. 6 is an exemplary view for explaining a format of an emphasis tag table used in the knowledge analysis system of the embodiment.
FIG. 7 is an exemplary view showing a state of an important word table during one document analysis in the knowledge analysis system of the embodiment.
FIG. 8 is a view showing a state of an important word table after one document analysis in the knowledge analysis system of the embodiment.
FIG. 9 is an exemplary view showing a state of an important word table after all documents are analyzed in the knowledge analysis system according to the embodiment;
FIG. 10 is an exemplary flowchart illustrating a procedure of knowledge analysis processing by the knowledge analysis system according to the embodiment;
FIG. 11 is an exemplary flowchart illustrating a procedure of important word determination processing by the knowledge analysis system according to the embodiment;
FIG. 12 is an exemplary flowchart illustrating a procedure of clustering processing by the knowledge analysis system according to the embodiment;
[Explanation of symbols]
11 ... Client terminal
12 ... Server computer
111 ... Web browser
112 ... Document editing tool
113 ... Character recognition tool
114: Image scanner
121 ... Web server
122 ... Knowledge Server
1221 ... Emphasis tag setting module
1222 ... Registration module
1223 ... Search module
1224 ... Clustering module
1225 ... Emphasis tag table
1226 ... Knowledge Database
1227 ... Keyword table
1228 ... Intermediate database
1229 ... Cluster database

Claims (5)

A knowledge analysis system configured to be connectable to a plurality of client terminals via a network and supporting classification and analysis of a plurality of document information stored in a server in response to a request from each client terminal,
Word extraction means for extracting a word having an emphasis attribute from each of the plurality of document information based on the emphasis tag embedded in each of the plurality of document information for the plurality of stored document information; ,
The importance is determined for each word based on the type of emphasis attribute of each word extracted by the word extraction means and the appearance frequency of the word, and the determination result of the importance is accumulated over the plurality of document information. An important word determination means for executing processing and determining the upper predetermined number of words having a high accumulated importance value as the important words of the entire document information across the plurality of document information;
The plurality of document information is classified and clustered by clustering the plurality of document information by matching the important words of the entire document information determined by the important word determination means and the important words determined from the document information. A knowledge analysis system comprising knowledge analysis means for creating an organized cluster database. Further comprising means for holding definition information indicating the relationship between the type of emphasis attribute and the corresponding weighting value;
The important word determining means includes
For each word extracted by the word extraction means, including means for determining the importance by the product of the weighting value corresponding to the emphasis attribute of the word acquired from the definition information and the appearance frequency of the word The knowledge analysis system according to claim 1, wherein: In accordance with a request from each client terminal, a server is a knowledge analysis method that supports classification and analysis of a plurality of document information stored in the server,
The server extracts words having an emphasis attribute from each of the plurality of document information based on emphasis tags embedded in each of the plurality of document information for the plurality of stored document information. A word extraction step;
The server determines the importance for each word based on the type of emphasis attribute of each word extracted by the word extraction step and the appearance frequency of the word, and the determination result of the importance is the plurality of documents. An important word determination step of executing a process of accumulating over the information, and determining an upper predetermined number of words having a high accumulated importance value as important words of the entire document information over the plurality of document information;
The server clusters the plurality of document information by matching the important words of the whole document information determined in the important word determination step with the important words determined from the document information, thereby the plurality of documents. A knowledge analysis method comprising: a knowledge analysis step of creating a cluster database in which information is classified and organized. In the important word determination step, for each word extracted in the word extraction step , the server acquires the importance level from the definition information indicating the relationship between the type of the emphasis attribute and the corresponding weighting value. 4. The knowledge analysis method according to claim 3, wherein the knowledge analysis method is determined by a product of a weighting value corresponding to the emphasis attribute and an appearance frequency of the word. In accordance with a request from each client terminal, a program for causing the server to perform classification and analysis of a plurality of document information stored in the server,
A word extraction procedure for extracting a word having an emphasis attribute from each of the plurality of document information based on an emphasis tag embedded in each of the plurality of document information for the plurality of document information stored; ,
The importance is determined for each word based on the type of emphasis attribute of each word extracted by the word extraction procedure and the appearance frequency of the word, and the determination result of the importance is accumulated over the plurality of document information. An important word determination procedure for executing processing to determine the upper predetermined number of words having a high accumulated importance value as the important words of the entire document information across the plurality of document information;
Classifying the plurality of document information by clustering the plurality of document information by matching the important words of the whole document information determined by the important word determination procedure with the important words determined from the document information. A program for causing the server to execute a knowledge analysis procedure for creating an organized cluster database.

Images