JP4395094B2 - Text mining system - Google Patents

Description

  The present invention relates to a text mining system.

To extract important information necessary for operations such as product planning and quality control from a large amount of accumulated text, such as questionnaires and repair requests, create a concept dictionary, analyze it, and improve the business In recent years, text mining systems that can be used in the future have increased.
JP 2001-184351 A JP 2003-141134 A Yasuhiro Takayama and three others, “Information search system InfoMAP based on word association”, Informatics Foundation No. 053-001, March 1999

  Such a conventional text mining system cannot be judged in advance by the user unless it is actually used. Uniform pricing such as sold out packaged software can be expensive for users who are interested but cannot determine the effect. On the other hand, it is not possible to collect a fee commensurate with the amount of use even from a user who has come to use it sufficiently because of the effect. The present invention is for solving the above-described problems, and an object of the present invention is to obtain a text mining system capable of performing pay-as-you-go billing suitable for text mining.

A text mining system according to the present invention includes a license ID for confirming license retention, a registration receiving unit that receives registration of a plurality of documents to be analyzed, a word included in the document, and a co-occurrence of the word A text analysis unit that extracts information; a concept dictionary creation unit that creates a concept dictionary by associating the word with a concept vector calculated from the co-occurrence information; and a concept dictionary storage unit that stores a plurality of the concept dictionaries An analysis receiving unit that receives the license ID, the specification of the document, and the specification of an analysis condition, an analysis unit that outputs a result of analyzing the concept dictionary related to the document according to the analysis condition, and the license For each ID, a predetermined storage type for storing a predetermined charging type and a current value, and a predetermined storage type according to the charging type corresponding to the license ID. A limit confirmation unit that adds the number of executions of the processed processing, the number of created products, or the amount of created products to the current value, a current value corresponding to the license ID stored in the limit storage unit, and the charge type And a charge calculation unit for calculating a charge from a predetermined charge unit price. The restriction confirmation unit has a charge type corresponding to the license ID within the concept dictionary business range, and creates the concept dictionary. When the unit creates a new concept dictionary, in the new concept dictionary, the ratio of overlapping words with the existing concept dictionary stored in the concept dictionary storage unit is any one of all the existing concept dictionaries. Is smaller than a predetermined value, 1 is added to the current value corresponding to the conceptual dictionary work range .

  The limit storage unit further stores a limit value for each license ID, and when the limit confirmation unit determines that the limit value is exceeded when added to the current value, the registration receiving unit or the analysis receiving unit The process is interrupted.

  In addition to being able to specify a value as the limit value, it is possible to specify unlimited.

  The restriction storage unit further stores, for each license ID, a charge calculation type having a value of sum, high, or low as a charge calculation type based on a plurality of charge types, and the restriction confirmation unit includes , According to a plurality of billing types corresponding to the license ID, and added to the current value corresponding to each of the billing types, the charge calculation unit, respectively according to the plurality of billing types corresponding to the license ID If the charge calculation method is a sum, the sum of the respective charges is calculated. If the charge calculation method is high, the highest charge is calculated as the charge calculation method. If the price is low, the cheapest fee is set as the fee for the license ID.

  The restriction storage unit further stores, for each license ID, a restriction combination type having all or one value as a restriction type based on a plurality of the charge types, and the restriction confirmation unit includes the restriction check unit When the type is all, the value is added to the current value corresponding to each of the charge types according to a plurality of charge types corresponding to the license ID, and the value after the addition exceeds the limit value for all the current values In this case, the registration receiving unit outputs an error, and when the restriction type is one, according to a plurality of charging types corresponding to the license ID, it adds to the current value corresponding to each of the charging types, When the current value exceeds the limit value, the registration receiving unit interrupts the processing.

  The analysis accepting unit accepts a user ID and makes an error if the user ID is not a user ID corresponding to a predetermined license ID.

The registration accepting unit, a plurality of clients terminal, receiving the document and the license ID, the text analysis unit for each of the license ID, extracts the said co-occurrence information of the word, the concept dictionary creation unit, for each of the license ID, to create the concept dictionary, the analysis reception unit, to the user ID your capital, receives and the user ID, and the designation of the document, and the designation of the analysis conditions The analysis unit transmits the analysis result according to the analysis condition to the client terminal for each user ID.

  According to the present invention, in the text mining, a pay-as-you-go system can be realized, so that the user who has not been able to judge the effect can easily make an initial introduction decision. Moreover, it is easy to be convinced to collect a lot of usage fees from users who have increased the amount of usage due to the effect.

Embodiment 1 FIG.
Hereinafter, the configuration of the text mining system according to the present embodiment will be described with reference to FIG. The text mining system according to the present embodiment includes a registration receiving unit 101 for registering a document to be analyzed, a text analysis unit 102, a concept dictionary creating unit 103, a document index creating unit 104, and an attribute information creating unit 105. A document storage unit 106 for storing information related to the registered document, a text analysis result storage unit 107, a concept dictionary storage unit 108, a document index storage unit 109, an attribute information storage unit 110, and a management information storage unit 111; , An analysis receiving unit 112 and an analysis unit 113 for analyzing a registered document, a registered charging type confirmation unit 114 and a registration restriction confirmation unit 115 for charging related to registration, and a charging related to registration A registration limit storage unit 116 for storing related information, an analysis billing type confirmation unit 117 for charging for analysis, and an analysis limit check Section 118, analysis restriction storage section 119 for storing information related to billing relating to analysis, charge calculation section 120 for calculating charges for a predetermined period, and information related to charge calculation Accounting information storage unit 121.

  The registration receiving unit 101 receives a license ID 122, a registration condition 123, and a document 124 from a user, controls the text analysis unit 102, the concept dictionary creation unit 103, the document index creation unit 104, and the attribute information creation unit 105, and By transmitting and receiving data related to the control, the document 124 is stored in the document storage unit 106 in accordance with the registration condition 123, information necessary for analysis is created from the document 124 in the analysis unit 113, and the text analysis result storage unit 107 is stored. And stored in the concept dictionary storage unit 108, the document index storage unit 109, the attribute information storage unit 110, and the management information storage unit 111.

  Further, the registration receiving unit 101 controls the registration charge type confirmation unit 114 and the registration restriction confirmation unit 115, and transmits / receives data related to the control, so that the charging related to the license ID 122 stored in the registration restriction storage unit 116 is performed. The type is confirmed, and a value necessary for performing metered charging is counted according to the charging type and the usage status of the registration process, and stored in the registration restriction storage unit 116.

  The license ID 122 is an ID for confirming license holding for using the text mining system. For example, the license ID 122 is a license ID assigned to a marketing department of a mail order company that concludes a use contract for the text mining system. The registration condition 123 is a condition for confirming whether to use an existing document or an existing concept dictionary, and will be described later with reference to FIG. The document 124 is a collection of documents to be analyzed, such as a collection of questionnaires about year-end gifts, a collection of repair request documents relating to specific products, and the like.

  The text analysis unit 102 extracts a word included in the document 124 and co-occurrence information of the word (a combination of two words that appear together in a predetermined range, such as the same document, the same paragraph, and the same sentence), and performs text analysis The result is stored in the result storage unit 107. The concept dictionary creation unit 103 creates a concept dictionary by associating the word extracted by the text analysis unit 102 with the concept vector calculated from the co-occurrence information, and stores the concept dictionary in the concept dictionary storage unit 108. The document index creation unit 104 creates document index information of the document 124 from the text analysis result storage unit 107 and stores it in the document index storage unit 109. The attribute information creation unit 105 extracts attributes assigned to each document included in the document 124, creates attribute information, and stores the attribute information in the attribute information storage unit 110.

  The registration receiving unit 101 associates the license ID 122, the document stored in the document storage unit 106, the concept dictionary stored in the concept dictionary storage unit 108, and the attribute information stored in the attribute information storage unit 110, and stores the management information storage. Store in the unit 111.

  The registered billing type confirmation unit 114 acquires the billing type related to the license ID 122 stored in the registration restriction storage unit 116. The registration restriction confirmation unit 115 counts a value necessary to perform metered charging according to the usage status of the registration processing and the charging type in the flow of the registration processing controlled by the registration receiving unit 101. And stored in the registration restriction storage unit 116.

  The analysis reception unit 112 receives the user ID 125 and the analysis condition 126 from the user, controls the analysis unit 113, and transmits / receives data related to the control, thereby analyzing the document specified by the analysis condition 126. And the analysis result 127 is output.

  Further, the analysis accepting unit 112 acquires the license ID 122 corresponding to the user ID 125 from the management information storage unit 111, controls the analysis charging type confirmation unit 117 and the analysis restriction confirmation unit 118, and transmits / receives data related to the control. Thus, the charge type related to the license ID 122 stored in the analysis restriction storage unit 119 is confirmed, and the value necessary for performing metered charge is counted according to the charge type and the usage status of the analysis process, and the analysis is performed. Store in the restriction storage unit 119.

  The user ID 126 is an ID for confirming an analyst assigned for each license ID 122. For example, the user ID 126 is a license ID assigned to Mr. A of a marketing department of a certain mail-order company that concludes a use contract for a text mining system. . The analysis condition 126 is a condition for designating information such as a document to be analyzed, an analysis technique, and an attribute or keyword that is an axis of analysis. As the analysis result 127, for example, a result of performing extraction of main keywords, trend analysis for each attribute, time-series trend analysis, and the like is output.

  The analysis reception unit 112 acquires a license ID corresponding to the user ID 125 received by the analysis reception unit, and requests the analysis unit 113 to perform analysis. The analysis unit 113 refers to the management information storage unit 111 to obtain the license ID received from the analysis unit 113 and the concept dictionary, document index information, and attribute information associated with the document specified by the analysis condition 126. Each of them is acquired from the concept dictionary storage unit 108, the document index storage unit 109, and the attribute information storage unit 110, analyzed using these, and the analysis result 127 is output via the analysis reception unit 112.

  The analysis billing type confirmation unit 117 acquires the billing type related to the license ID 122 stored in the analysis restriction storage unit 119. The analysis restriction confirmation unit 118 counts a value necessary for performing metered charge according to the charge type in the flow of analysis processing controlled by the analysis reception unit 112, and the analysis restriction storage unit 119. To store.

  The fee calculation unit 120 receives an instruction from the administrator of the text mining system, or automatically performs registration processing and analysis processing stored in the registration restriction storage unit 116 and the analysis restriction storage unit 119 on a predetermined date. A charge 128 for each license ID 122 is calculated based on the usage status and the charge unit price stored in the charge information storage unit 121.

  Hereinafter, the registration condition 123 of FIG. 1 will be described with reference to FIG. As described with reference to FIG. 1, the text mining system according to the present embodiment creates a concept dictionary, document index information, and attribute information from a document under the control of the registration receiving unit 101, and stores the concept dictionary in association with the document. Registration processing, analysis processing that analyzes using the concept dictionary, document index information, and attribute information under the control of the analysis reception unit 112, and fee calculation processing by the fee calculation unit 120. However, depending on the designation of registration conditions, existing data can be used instead of creating all data newly. The registration conditions 123 include, for example, new document registration and new concept dictionary creation, new document registration and existing concept dictionary diversion, and additional document registration and existing concept dictionary diversion.

  When the registration condition 123 is new document registration and creation of a new concept dictionary, the registration receiving unit 101 stores the received document 124 in the document storage unit 106, and the text analysis unit 102 analyzes the result of analyzing the document 124 as a text analysis result. The concept dictionary creation unit 103 creates a concept dictionary from the text analysis result stored in the text analysis result storage unit 107 and stores it in the concept dictionary storage unit 108. The document index creation unit 104 Document index information is created from the text analysis result stored in the text analysis result storage unit 107 and stored in the document index storage unit 109, and the attribute information storage unit 110 creates attribute information from the document 123 and generates attribute information storage unit. The management information storage unit 11 stores the relationship among the new document 123, the new concept dictionary, the new document index information, and the new attribute information. Stored, these data are used for analysis processing after. For example, as shown in the row 201 of FIG. 2, normally, a questionnaire 1 dictionary is created from a questionnaire 1 document, and the questionnaire 1 dictionary is used for analysis of the questionnaire 1 document in later analysis processing.

  When the registration condition 123 is new document registration and creation of an existing concept dictionary, the registration receiving unit 101 stores the received document 124 in the document storage unit 106, and the text analysis unit 102 analyzes the result of analyzing the document 124 as a text analysis result. Instead, the concept dictionary creation unit 103 does not create a concept dictionary. Instead, the registration accepting unit 101 accepts designation of an existing concept dictionary, and the document index creation unit 104 performs text analysis result storage unit 107. The document index information is created from the text analysis result stored in the document index and stored in the document index storage unit 109. The attribute information storage unit 110 creates attribute information from the document 123 and stores it in the attribute information storage unit 110. The relationship between the document 123, the existing concept dictionary, the new document index information, and the new attribute information is stored in the management information storage unit 111. It is used to analyze the process. For example, as shown in the row 202 of FIG. 2, when it is expected that the questionnaire 1 document and the questionnaire 2 document have relatively similar contents, the time or the storage capacity of the concept dictionary storage unit 108 is saved. Therefore, without creating a concept dictionary, the questionnaire 1 dictionary is used for analysis of the questionnaire 1 document in later analysis processing.

  When the registration condition 123 is additional document registration and creation of an existing concept dictionary, the registration receiving unit 101 stores a document in which the newly received document 124 is added to the specified existing document in the document storage unit 106, and the text The analysis unit 102 stores the analysis result of the added document in the text analysis result storage unit 107, and the concept dictionary creation unit 103 does not create the concept dictionary, but instead uses the existing concept used by the existing document. In association with the dictionary, the document index creation unit 104 creates document index information from the text analysis result stored in the text analysis result storage unit 107 and stores it in the document index storage unit 109, and the attribute information storage unit 110 adds Attribute information is created from the document and stored in the attribute information storage unit 110, and the relationship between the added document, the existing concept dictionary, the new document index information, and the new attribute information is managed. Stored in the information storage unit 111, these data are used for analysis processing after. For example, as shown in the row 202 of FIG. 2, when it is expected that the questionnaire 1 document and the questionnaire 2 document have relatively similar contents, the time or the storage capacity of the concept dictionary storage unit 108 is saved. Therefore, the concept dictionary is not created, and the questionnaire 1 dictionary is used for the analysis of the questionnaire 1 document and the questionnaire 2 document in the later analysis processing.

  Next, the contents of data stored in each storage unit will be described with reference to FIGS. The document storage unit 106 stores a plurality of documents, and normally stores the documents received in one registration process as a group of documents. However, depending on the designation of the registration condition 123, a new document may be added to an existing document and stored as described with reference to FIG.

  The text analysis result storage unit 107 includes words included in a registration target document (in the case of a new document, a new document may be added to the existing document as described above), and the word co-occurrence. Information (a combination of two words that appear together in a predetermined range, such as the same document, the same paragraph, and the same sentence) is stored. Note that the text analysis result storage unit 107 is a temporary storage area, and its contents are deleted after one registration process is completed.

  The concept dictionary storage unit 108 stores a plurality of concept dictionaries. FIG. 3 is an example of one concept dictionary. A column 301 is a word included in a document to be registered, and a column 302 and subsequent columns are concept vectors 306 obtained by compressing the above-described co-occurrence information using a technique called SVD (singular value decomposition).

  The document index storage unit 109 stores a plurality of document index information. One document index information is created corresponding to one registration target document. Further, one document index information is obtained by associating document index information with respect to each of documents at a finer level included in one document 124 to be registered.

  The attribute information storage unit 110 stores a plurality of attribute information. One piece of attribute information is created corresponding to one document to be registered. In addition, one attribute information is a collection of attribute information assigned to each document at a finer level included in one document to be registered.

  As described above, the management information storage unit 111 stores the license ID 122 and the related information of the registration object such as the created registration target document, concept dictionary, document index information, and attribute information. FIG. 4 is an example of the related information of the registered item, and has a license ID 401, a document name 402, a concept dictionary name 403, a document index information name 404, and an attribute information name 405.

  Furthermore, the management information storage unit 111 stores license information. FIG. 5 is an example of license information. The license information is stored in association with the license ID 501 and the user ID 502. In addition, attribute information such as the name of the license partner related to the license ID, attribute information such as a password, the name of an analyst related to the user ID, and password may be held.

  The registration restriction storage unit 116 stores a current value, which is a metered amount for metered billing, and a limit value for restricting use. FIG. 6 is an example of data stored in the registration restriction storage unit 116, and has a license ID 601, a charge type 602, a current value 603, and a restriction value 603. In the present embodiment, one accounting type 602, current value 603, and limit value 604 are defined for one license ID 601. The values of the license ID 601 and the charge type 602 are set in advance. In the current value 603, the value counted by the registration restriction confirmation unit 115 according to the charging type 602 and the usage status of the registration process is constantly updated in the flow of the registration process. The value of the limit value 604 is set in advance, and when the current value 603 exceeds the limit value, use of the registration process is limited. It should be noted that the limit value 604 can be specified without limitation. In this case, use of the registration process is not limited.

  The analysis restriction storage unit 119 stores a current value that is a pay-as-you-go charge for pay-per-use and a limit value for restricting use. FIG. 7 is a data example of the analysis restriction storage unit 119, and similarly to the registration restriction storage unit 116, it holds a license ID 601, a charge type 602, a current value 603, and a limit value 603, but for the charge type 602, A billing type suitable for analysis processing is set. In the present embodiment, one accounting type 602, current value 603, and limit value 604 are defined for one license ID 601. The values of the license ID 601 and the charge type 602 are set in advance. The current value 603 is constantly updated with a value counted according to the charge type 602 and the usage status of the registration process by the analysis restriction confirmation unit 118 in the flow of the analysis process. The value of the limit value 604 is set in advance, and when the current value 603 exceeds the limit value, use of the analysis process is limited.

  The billing information storage unit 121 stores a unit price for fee calculation. FIG. 8 shows an example of data stored in the billing information storage unit 121, which includes a billing target process 801, a billing type 602, and a bill unit price 803. For each charge type 602, a charge target process 801 and a charge unit price 803 are set in advance. The billing target process 801 sets whether the corresponding billing type 602 is a person related to the usage status of the registration process or the analysis process. The charge unit price 803 sets the charge unit price for one count of the current value 603 of the registration restriction storage unit 116 and the current value 603 of the analysis restriction storage unit 119 for the corresponding charge type 602.

  The charge type 602 defines the type of object to be used as a pay-as-you-go charge for pay-as-you-go billing, and may be a predetermined number of processing executions, the number of created products, or the amount of created products. More specific examples include document registration count, document addition count, concept dictionary creation count, document amount, word count, concept dictionary capacity, concept dictionary work range, and analysis execution count.

  The number of document registrations is the number of times a new document 124 is registered. The document addition count is the number of times a new document 124 is added to an existing document. The concept dictionary creation count is the number of times a new concept dictionary is created. The document amount is the number of words included in the new document 124. The number of words is the number of words included in the new document 124. The concept dictionary capacity is the capacity of a new concept dictionary. The unit is predetermined such as MB. The concept dictionary work range is a determination as to whether or not the new concept dictionary is a concept dictionary related to a different work range when compared with the existing concept dictionary related to the corresponding license ID 122, and will be described in detail later. The number of times of analysis execution is the number of times of analysis.

  Hereinafter, the H / W configuration of the text mining system according to the present embodiment will be described with reference to FIG. The text mining system includes an external storage device 901, a CPU (central processing unit) 902, a main memory (main storage device) 903, an input device 904, a display device 905, and a communication device 906 connected by a bus. Document storage unit 106, concept dictionary storage unit 108, document index storage unit 109, attribute information storage unit 110, management information storage unit 111, registration restriction storage unit 116, analysis restriction storage unit 119, and accounting information storage unit of the text mining system Each storage unit 121 and the like includes an external storage device 901 such as a hard disk, a floppy (registered trademark) disk, an MO, a CD, a DVD, and a magnetic tape. The text analysis result storage unit 107 and a temporary storage area necessary for executing the program are configured by a main memory (main storage device) 903 such as a RAM (Random Access Memory), or an external storage device 901, for example. The Registration accepting unit 101, text analyzing unit 102, concept dictionary creating unit 103, document index creating unit 104, attribute information creating unit 105, analysis accepting unit 112, analysis unit 113, registered billing type confirmation unit 114, registration restriction confirmation unit 115, The analysis billing type confirmation unit 117, the analysis restriction confirmation unit 118, and the charge calculation unit 120 are configured by an external storage device 901, a main memory (main storage device) 903, and a CPU (central processing unit) 902. Programs related to data processing loaded in the main memory (main storage device) 903 are sequentially read into the CPU (central processing unit) 902 for processing. Each processing unit accepts input or outputs via an input device 904 such as a mouse or a keyboard, a communication device 906 connected to a LAN or WAN, for example, a display device 805 such as a display, or the like. Can do.

  Hereinafter, the flow of registration processing in the present embodiment will be described with reference to FIGS. The registration receiving unit 101 receives the license ID 122 (step S1001). The registration receiving unit 101 refers to the management information storage unit 111 and confirms the presence of the corresponding license ID 122 (step S1002). If the corresponding license ID 122 does not exist, the registration process is interrupted and an error message is output. The registered billing type confirmation unit 114 refers to the registration restriction storage unit 116 and acquires the billing type 602 related to the registration of the corresponding license ID 122 (step S1003). The registration accepting unit 101 accepts the designation of the registration condition 123 and the document 124 from the user (step S1004).

  The registration restriction confirmation unit 115 adds 1 to the current value 603 of the corresponding license ID 122 in the registration restriction storage unit 116 when the charge type 602 is the number of document registrations and the registration condition 123 specifies new document registration. . The registration restriction confirmation unit 115 adds 1 to the current value 603 of the corresponding license ID 122 in the registration restriction storage unit 116 when the charge type 602 is the number of times of document addition and the registration condition 123 specifies additional document registration. . The registration restriction confirmation unit 115 sets the current value 603 of the corresponding license ID 122 in the registration restriction storage unit 1 to 1 when the accounting type 602 is the number of times of concept dictionary creation and the registration condition 123 designates creation of a new concept dictionary. to add. When the charge type 602 is the document amount, the registration restriction confirmation unit 115 counts the number of characters of the document 124 and adds the number of characters to the current value 603 of the corresponding license ID 122 in the registration restriction storage unit 116. (Step S1005). If the current value 603 after the addition exceeds the limit value 604, the current value 603 is restored, the registration process is interrupted, and an error message is output.

  The registration receiving unit 101 stores the document 124 in the document storage unit 106 (step S1006). When the registration condition 123 specifies additional document registration, the document 124 is added to the specified existing document.

  The text analysis unit 102 analyzes the text included in the document 124, divides it into words, and creates co-occurrence information (step S1101).

  When the charge type 602 is the number of words, the registration restriction confirmation unit 115 acquires the number of words counted by the text analysis unit 102 and adds the number of words to the current value 603 of the corresponding license ID 122 in the registration restriction storage unit 116. . (Step S1102). If the current value 603 after addition exceeds the limit value 604, the current value 603 is restored, the document registered in the document storage unit 106 is also restored, the registration process is interrupted, and an error Outputs the effect.

  When the charge type 602 is the concept dictionary capacity and the registration condition 123 specifies creation of a new concept dictionary, the registration restriction confirmation unit 115 acquires the number of words counted by the text analysis unit 102 and sets the number of words in advance. When the estimated value of the conceptual dictionary capacity is calculated by multiplying the calculated numerical value and the estimated value is added to the current value 602 of the corresponding license ID 122 in the registration restriction storage unit 116 to exceed the limit value 604, Output to that effect. (Step S1103). At this time, if the registration process is continued or interrupted, and the interrupt is accepted, the document registered in the document storage unit 106 is also restored, the registration process is interrupted, and an error message is output. To do.

  The text analysis unit 102 stores the analysis result in the text analysis result storage unit 107 (step S1104).

  The concept dictionary creation unit 103 calculates a concept vector of each word from the text analysis result and creates a concept dictionary (step S1105). When the registration condition 123 specifies that an existing concept dictionary is used, a concept dictionary is not created, and steps S1105, S1106, and S1107 are not executed.

  The registration restriction confirmation unit 115 acquires the capacity of the concept dictionary created by the concept dictionary creation unit 103 when the charge type 602 is the concept dictionary capacity and the registration condition 123 designates the creation of a new concept dictionary, and the registration restriction is obtained. The concept dictionary capacity is added to the current value 603 of the corresponding license ID 122 in the storage unit 116. The registration restriction confirmation unit 115 registers the business range of the concept dictionary created by the concept dictionary creation unit 103 when the charge type 602 is the concept dictionary business range and the registration condition 123 specifies creation of a new concept dictionary. 1 is added to the current value 603 of the corresponding license ID 122 of the registration restriction storage unit 116 if it is determined that the business range is different from any of the existing concept dictionaries. (Step S1106). If the current value 603 after addition exceeds the limit value 604, the current value 603 is restored, the document registered in the document storage unit 106 is also restored, the registration process is interrupted, and an error Outputs the effect. A method for determining the concept dictionary work range will be described later with reference to FIG.

  The concept dictionary creation unit 103 stores the created concept dictionary in the concept dictionary storage unit 108 (step S1107).

  The document index creation unit 104 creates document index information from the text analysis result and stores it in the document index storage unit 109 (step S1201). The attribute information creation unit 105 creates attribute information of the document 124 and stores it in the attribute information storage unit 110 (step S1202). The registration receiving unit 101 associates the license ID 410 with the names of the document, concept dictionary, document index, and attribute information stored in each storage unit, and stores them in the management information storage unit 111 as illustrated in FIG. Step S1203). The registration receiving unit deletes the text analysis result from the text analysis result storage unit 107 (step S1204).

  Hereinafter, the flow of analysis processing in the present embodiment will be described with reference to FIG. The analysis reception unit 112 receives the user ID 125 (step S1301). The analysis reception unit 112 refers to the management information storage unit 111 and confirms the presence of the corresponding user ID 125 and the corresponding license ID 122 (step S1302). If the user ID 125 does not exist, the registration process is interrupted. The analysis charging type confirmation unit 117 refers to the registration restriction storage unit 116 and acquires the charging type 602 related to the analysis of the corresponding license ID 122 (step S1303).

  If the charge type 602 is the number of times of analysis execution, the analysis restriction confirmation unit 118 adds 1 to the current value 603 of the corresponding license ID 122 in the analysis restriction storage unit 119 (step S1304). If the current value 603 after the addition exceeds the limit value 604, the current value 603 is restored, the analysis process is interrupted, and an error message is output.

  The analysis reception unit 112 receives the analysis target document name and the analysis condition 126 (keyword, attribute, etc.) from the user, and requests the analysis unit 113 to analyze the corresponding analysis target document 124 according to the corresponding analysis condition 126 (step S1305). .

  The analysis unit 113 acquires the concept dictionary, document index information, and attribute information corresponding to the corresponding analysis target document 124, analyzes them according to the corresponding analysis conditions 126, and outputs an analysis result 127 (step S1306).

  Hereinafter, the flow of fee calculation processing in the present embodiment will be described with reference to FIG. The process from step S1401 to step S1406 is repeated for all license IDs (step S1401). The fee calculation unit 120 acquires the current value 603 related to the corresponding license ID 122 from the registration restriction storage unit 116 and the analysis restriction storage unit 119 (step S1402). The fee calculation unit 120 refers to the billing information storage unit 121, calculates the fees related to the registration process and the analysis process, adds them all up, and sets the charge 128 of the corresponding license ID 122 (step S1403).

  The fee calculation unit 120 outputs the fee 128 (step S1404). The fee calculation unit 120 sets the current value 603 related to the corresponding license ID 122 in the registration restriction storage unit 116 and the analysis restriction storage unit 119 to 0 (step S1405). If all license IDs 122 have been calculated, the fee calculation process is completed. Otherwise, the process returns to step S1401.

  Hereinafter, the flow of the concept dictionary work range determination process in the present embodiment will be described with reference to FIG. This processing is a flow of processing in which the registration restriction confirmation unit 115 performs the conceptual dictionary business range determination when the charge type 602 is the conceptual dictionary business range, which has been described in step S1106 of FIG.

  The processing from step S1501 to step S1505 is repeated for all existing concept dictionaries (step S1501). The number of duplicate words stored redundantly in both the new concept dictionary and the existing concept dictionary is confirmed (step S1502). The duplication ratio obtained by dividing the number of duplicate words by the number of words registered in the new concept dictionary is acquired (step S1503).

If the overlap ratio is not smaller than a predetermined value, the process proceeds to step S1507. In cases other than that described here, process flow proceeds to Step S1505. (Step S1504).
If all the existing concept dictionaries have been confirmed, the process proceeds to step S1506. Otherwise, the process returns to step S1501 (step S1505).

It was confirmed that the new concept dictionary has a word overlapping ratio smaller than a predetermined value with respect to all existing concept dictionaries. That is, the new concept dictionary uses a lot of different terms compared to any of the existing concept dictionaries, and the business new concept dictionary is determined to have a different business scope from any of the existing concept dictionaries. (Step S1506).
New concept dictionary, for at least one existing concept dictionary, that overlap percentage of words is not rather smaller than a predetermined value, was confirmed. That is, the new concept dictionary uses many similar terms with respect to at least one existing concept dictionary, and the new concept dictionary is determined to be within the same business scope as any of the existing concept dictionaries. (Step S1507).

  According to the present invention, in the text mining, a pay-as-you-go system can be realized, so that the user who has not been able to judge the effect can easily make an initial introduction decision. Moreover, it is easy to be convinced to collect a lot of usage fees from users who have increased the amount of usage due to the effect.

As the accounting classification, and more that you can specify the concept dictionary business scope, if the use departments and applications of text mining system has spread in the customer, it is possible to increase the billing.

  By specifying the limit value, it is possible to prevent the user from spending more than expected.

  Since it is possible to designate unlimited as the limit value, it is possible to selectively use a case where the restriction is applied and a case where the restriction is not applied depending on the license ID.

  A plurality of user IDs can be set for each license ID. As one operation mode of the text mining system, the registration process is performed as a regular operation by the client-side administrator, and the analysis process is performed irregularly while repeating various trials and errors with various know-how. Although it is assumed to be performed, it is possible to cope with such an operation mode.

Embodiment 2. FIG.
Next, a second embodiment will be described. In this embodiment, an example in which a plurality of charging types 602 can be set for one license ID 601 will be described. The configuration of the text mining system in the present embodiment is the same as that of FIG. 1 described in the first embodiment. However, the data stored in the registration restriction storage unit 116 is different from that in the first embodiment.

  FIG. 16 is a data example of the registration restriction storage unit 116 in this embodiment. In addition to the license ID 601, the charge type 602, the current value 603, and the limit value 603, the charge combination type 1601 and the limit combination type 1602 are shown. Is held. In this embodiment, a plurality of charging types 602, a current value 603, and a limit value 604 are defined for one license ID 601, and one charge combination type 1601 and limit combination are defined for one license ID 601. A type 1602 is defined.

  The values of the license ID 601 and the charge type 602 are set in advance. In the current value 603, the value counted by the registration restriction confirmation unit 115 according to the charging type 602 and the usage status of the registration process is constantly updated in the flow of the registration process. This is executed for a plurality of billing types 602. The value of the limit value 604 is set in advance, and when the current value 603 exceeds the limit value, use of the registration process is limited. The charge combination type 1601 is set in advance, and sum, high, and low can be set as values representing how to combine charges related to a plurality of charge types 602. The limit combination type 1602 is set in advance, and all or one can be set as a value indicating how the restrictions relating to the plurality of section types 602 are matched.

  In the first embodiment, in step S1005, step S1102, and step S1106 of the registration process shown in FIGS. 10 to 12, the current value 603 is added according to the charge type 602. The same applies to the present embodiment. Then, addition is performed for a plurality of accounting types 602.

  In the first embodiment, if the current value 603 after the addition exceeds the limit value 604 in steps S1005, S1102, and S1106, the current value 603 is returned to the original value, the registration process is interrupted, and an error occurs. In the present embodiment, the process proceeds as follows according to the value of the restricted combination type 1602. When the limit combining unit 1602 is all, when the current value 603 exceeds the limit value 604 for all the charging types 602 related to the license ID 601 of the licensee who performs the registration process, the current value 603 is returned to the original value. The registration process is interrupted and an error message is output. When there is only one limit combining unit 1602, if the current value 603 exceeds the limit value 604 for one charging type 602 related to the license ID 601 of the licensee who performs the registration process, the current value 603 is returned to the original value. The registration process is interrupted and an error message is output.

  In the first embodiment, in step S1304 of the analysis process shown in FIG. 13, addition of the current value 603 corresponding to the charge type 602 and interruption of the analysis process when the current value 603 exceeds the limit value 604 are performed. In the present embodiment, only one charge type 602 is prepared for the analysis process, and therefore the operation for the analysis process is the same as that of the first embodiment. When there are a plurality of billing types 602 in the analysis process, addition is performed for all the billing types 602 as in the case of the registration processing, and the limit is set according to the limit combination type 1602.

  Further, in the first embodiment, in step S1403 of the charge calculation process shown in FIG. 14, the charge calculation unit 120 refers to the billing information storage unit 121, calculates the charges related to the registration process and the analysis process, and adds all of them. In the present embodiment, the process proceeds as follows according to the value of the charge combination type 1601.

  The fee calculating unit 120 refers to the charging information storage unit 121 and calculates a fee for each charging type 602 for all charging types 602 related to the license ID 601 to be calculated. Next, when the charge combination type 1601 is sum, the charge calculation unit 120 calculates the sum of charges for each charge type 602. Alternatively, when the charge combination type 1601 is high, the highest charge is selected from the charges for each charge type 602. Alternatively, when the charge combination type 1601 is low, the lowest price among the charges for each charge type 602 is selected. The fee calculated or selected in this way is set as the fee for the corresponding license ID.

  According to the invention of the present embodiment, in text mining, it is possible to realize flexible metered billing according to client needs using a plurality of billing types.

  Further, by using a plurality of billing types, it is possible to realize flexible usage restrictions according to client needs.

In the first embodiment or the second embodiment, the structure of the text mining system described with reference to FIG. 1 includes the document index creation unit 104, the attribute information creation unit 105, the document index storage unit 109, and the attribute information storage unit 110. May not be included.
In this case, the management information storage unit 111 described with reference to FIG. 4 does not have the document index information name 404 and the attribute information name 405, but has the license ID 401, the document name 402, and the concept dictionary name 403. . In the registration process flow described with reference to FIG. 12, steps S1201 and S1202 are not necessary. In step S1306 of the analysis processing flow described with reference to FIG. 13, the analysis unit 113 acquires a concept dictionary corresponding to the corresponding analysis target document 124, analyzes it according to the corresponding analysis condition 126, and outputs an analysis result 127. .

  Note that the text mining system described in the first embodiment or the second embodiment may allow input / output to / from a plurality of client terminals via the communication device 906 connected to a LAN, a WAN, or the like.

  According to the present invention, a text mining system can be provided to a large number of clients via a LAN or a WAN, and it is easy for the client to introduce the text mining system. Easy to manage clients.

  In addition, there are the following effects overall. Even for users who have not been able to judge the effect of the text mining system, it is easy to make an initial introduction decision. Moreover, it is easy to be convinced to collect a lot of usage fees from users who have increased the amount of usage due to the effect. When a plurality of users use a text mining device on a server in an in-house intranet, a usage fee corresponding to the scale can be collected by realizing charge-based billing. Assuming a service that operates a text mining device on a server on the Internet and generally provides analysis functions such as pre-registered technical documents and patent information, the usage fee according to the scale can be obtained by realizing metered billing. Can be collected.

It is a block diagram of the text mining system in Embodiment 1 of this invention. It is a schematic diagram explaining the registration conditions in Embodiment 1 of this invention. It is a figure of the example of data of the concept dictionary memory | storage part in Embodiment 1 of this invention. It is a figure of the example of data of the management information storage part in Embodiment 1 of this invention. It is a figure of the example of data of the management information storage part in Embodiment 1 of this invention. It is a figure of the example of data of the registration restriction memory | storage part in Embodiment 1 of this invention. It is a figure of the example of data of the analysis restriction storage part in Embodiment 1 of this invention. It is a figure of the example of data of the accounting information storage part in Embodiment 1 of this invention. It is a H / W block diagram of the text mining system in Embodiment 1 of this invention. It is a figure which shows the flow of the registration process in Embodiment 1 of this invention. It is a figure which shows the flow of the registration process in Embodiment 1 of this invention. It is a figure which shows the flow of the registration process in Embodiment 1 of this invention. It is a figure which shows the flow of the analysis process in Embodiment 1 of this invention. It is a figure which shows the flow of the charge calculation process in Embodiment 1 of this invention. It is a figure which shows the flow of the concept dictionary work range determination process in Embodiment 1 of this invention. It is a figure of the example of data of the registration restriction memory | storage part in Embodiment 2 of this invention.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 101 Registration reception part, 102 Text analysis part, 103 Concept dictionary creation part, 104 Document index creation part, 105 Attribute information creation part, 106 Document storage part, 107 Text analysis result storage part, 108 Concept dictionary storage part, 109 Document index storage 110, attribute information storage unit, 111 management information storage unit, 112 analysis reception unit, 113 analysis unit, 114 registered billing type confirmation unit, 115 registration restriction confirmation unit (restriction confirmation unit), 116 registration restriction storage unit (restriction storage unit) ) 117 analysis charge type confirmation unit, 118 analysis restriction confirmation unit (restriction confirmation unit), 119 analysis restriction storage unit (restriction storage unit), 120 fee calculation unit, 121 charge information storage unit.

Claims (7)

A registration receiving unit for receiving a license ID for confirming license holding and registration of a plurality of documents to be analyzed;
A text analysis unit that extracts words included in the document and co-occurrence information of the words;
A concept dictionary creating unit that creates a concept dictionary by associating the word with a concept vector calculated from the co-occurrence information;
A concept dictionary storage unit for storing a plurality of the concept dictionaries;
An analysis reception unit that receives the license ID, the specification of the document, and the specification of analysis conditions;
An analysis unit that outputs a result of analyzing the concept dictionary according to the document according to the analysis condition;
A limit storage unit for storing a predetermined charging type and a current value for each license ID;
In accordance with the billing type corresponding to the license ID, a limit confirmation unit that adds a predetermined number of processing executions, the number of creations, or the amount of creations to the current value;
A charge calculation unit for calculating a charge from a current value corresponding to the license ID stored in the restriction storage unit and a charge unit price determined in advance according to the charge type ;
When the charge type corresponding to the license ID is a concept dictionary service range and the concept dictionary creation unit creates a new concept dictionary, the restriction confirmation unit stores the concept dictionary in the new concept dictionary. If the ratio of overlapping words with existing concept dictionaries stored in the section is smaller than a predetermined value for any one of all the existing concept dictionaries, the current status corresponding to the concept dictionary work range A text mining system characterized by adding 1 to a value . The limit storage unit further stores a limit value for each license ID,
When the limit confirmation unit determines that the limit value is exceeded when added to the current value,
The text mining system according to claim 1, wherein the registration reception unit or the analysis reception unit interrupts processing. The text mining system according to claim 2 , wherein an unlimited value can be designated in addition to a value that can be designated as the restricted value. The limit storage unit further stores, for each license ID, a charge calculation type having a value of sum, high, or low as a charge calculation type based on a plurality of charge types,
The restriction confirmation unit adds to the current value corresponding to each of the charging types according to a plurality of charging types corresponding to the license ID,
The charge calculation unit calculates a charge corresponding to each of the charge types according to a plurality of charge types corresponding to the license ID,
If the charge calculation method is sum, the sum of the charges, if the charge calculation method is high, the highest charge, if the charge calculation method is low, the cheapest charge, The text mining system according to any one of claims 1 to 3 , wherein the fee is a fee for the license ID. The restriction storage unit further stores, for each license ID, a restriction combination type having all or one of the values as a restriction type based on a plurality of the charge types,
The restriction confirmation unit
When the restriction types are all, according to a plurality of charging types corresponding to the license ID, it is added to the current value corresponding to each of the charging types, and the value after the addition is set as a limiting value for all current values. If it exceeds, the registration reception unit will output an error,
When there is one restriction type, the registration is added to the current value corresponding to each of the charge types according to a plurality of charge types corresponding to the license ID, and when the limit value is exceeded for one current value, the registration The text mining system according to any one of claims 2 to 3 , wherein the reception unit interrupts the processing. Wherein the analysis reception unit accepts a user ID, according to any one of claims 1 to 5, wherein the user ID is characterized in that an error unless the user ID corresponding to the license ID to a predetermined Text mining system. The registration accepting unit, a plurality of clients terminal, receiving the document and the license ID,
The text analysis unit extracts the co-occurrence information of the word for each license ID,
The concept dictionary creation unit creates the concept dictionary for each license ID,
Wherein the analysis reception unit, the user ID your capital, receiving said user ID, and designation of the document, and a designation of the analysis condition,
The text mining system according to claim 6 , wherein the analysis unit transmits a result of analysis according to the analysis condition to the client terminal for each user ID.

Images