JP5217513B2 - An information analysis processing method, an information analysis processing program, an information analysis processing device, an information registration processing method, an information registration processing program, an information registration processing device, an information registration analysis processing method, and an information registration analysis processing program. - Google Patents

Description

  The present invention relates to an information analysis processing method for analyzing and classifying information such as a document, an information analysis processing program, an information analysis processing device, an information registration processing method for registering classification information for classifying information such as a document, and information Registration processing program, information registration processing device, information registration analysis processing method for registering classification information for classifying information such as documents, analyzing the information and classifying according to registered classification information, and information registration analysis processing program About.

  With the recent development of natural language processing technology and the improvement of computer processing power, it is possible to extract documents with similar contents from a large collection of accumulated documents and classify them based on the degree of similarity It has become.

  As a method for determining whether documents are similar, for example, the following methods are known. First, the target document is decomposed into elements having character strings, words, and phrases as units, and feature quantities are calculated based on combinations of the elements. Then, the similarity of the feature amount is obtained for all the combinations of documents, and if the similarity is equal to or higher than a certain level, it is regarded as similar.

  Various methods have been devised for calculating feature amounts. For example, after decomposing the target document into elements of character strings, words, and phrases, the weight of the element is obtained based on the appearance frequency of each element in the document set and the appearance frequency in the target document. A method for expressing a feature value by a vector constituted by the weight is known. The similarity is calculated by obtaining the inner product of the vectors. For classification based on similarity, the average value of feature quantities (vectors) of document groups defined as having the same classification is calculated, and the similarity between the feature quantity (vector) of the target document and the average vector is more than a certain level. If there is, there is a method for determining that the target document is the classification.

  Here, for example, as a method for classifying a query sentence, a query document classification apparatus and method and a recording medium on which a program describing the method is recorded are disclosed (see Patent Document 1). This patent document 1 describes a method for automatically classifying a document set by clustering a document set, and a method for classifying a new document using the classification result. That is, the accumulated inquiry / answer document information is classified into a plurality of categories based on the contents thereof, and the category having the closest contents is selected for a newly received inquiry.

  In addition, a security information estimation device, a security information estimation method, a security information estimation program, and a recording medium that appropriately protect information that is not associated with security information are disclosed (see Patent Document 2). In this patent document 2, in a system for estimating the security attribute of a document, a document similar to the document to be estimated is searched from existing stored documents, and the security attribute of the similar stored document found is estimated. Is presented as a system.

  Also disclosed is an image forming apparatus, an image forming system, a security management apparatus, and a security management method that maintain document management security even when a document ID is not acquired from a read original (see Patent Document 3). In this patent document 3, when copying a manuscript with a copying machine, if security information cannot be read from the manuscript, a character string or the like is extracted from the scanned image and matches a document of existing document management data. A system is described that determines whether or not to perform copying, and stops copying if they match.

  In addition, an image processing apparatus, an image processing method, an image processing program, and a recording medium that can appropriately determine text information included as an image in image data are disclosed (see Patent Document 4). In a system for estimating security attributes, when a document to be estimated is a scanned image, the scanned image is extracted by applying an OCR (Optical Character Reader), and a document similar to the text is extracted from existing stored documents. Although the search is performed, the search may not be performed well due to an OCR error. In order to solve this problem, in Patent Document 4, a print image of an existing stored document is created in advance, and a text obtained as a result of performing OCR on the print image is held. And if the text is matched, similar documents can be found more accurately.

  Further, there is disclosed a document classification support method and apparatus that, when classifying a document into categories, collects documents with similar contents based on an estimated category and sequentially presents them to a user to prompt a check (patent) Reference 5). When an electronic document is applied to a document classification system, it is necessary to check whether there is a classification error. However, in Patent Document 5, this check can be easily performed. It also describes the document classification system itself.

  In addition, products that automatically classify enormous text information are manufactured (see Non-Patent Document 1). The product of this non-patent document 1 associates the classification code (plurality) with the text so that the system learns it, and as shown in FIG. The code (s) are obtained.

  In recent years, it has been required to prevent information leakage of trade secrets and personal information handled by companies. Therefore, it is known to control access so that only authorized users can access confidential information, or encrypt confidential information so that only authorized users can view and print. On the other hand, however, there is a problem that confidential information that is accessed by an authorized user cannot be managed if it is printed.

  To deal with such problems, the contents of a document are analyzed when an e-mail is sent or when a paper document is scanned, copied, or faxed by an MFP (Multi Function Peripherals). A system has been proposed in which a confidential document similar to the document is searched from a document DB (database) to determine whether the document is similar to the confidential document, and the security attribute of the target document is estimated (for example, a patent) References 2-4).

JP 2000-148770 A JP 2006-185153 A Japanese Patent Laying-Open No. 2005-166023 JP 2006-293917 A Japanese Patent No. 3603392 http://www.justsystem.co.jp/km/product/cb102.html

  However, in reality, there is a possibility that the contents of a plurality of confidential documents are mixed in the target document to be analyzed. For example, a case where an electronic document file attached to an e-mail is created by copy and paste from a plurality of documents can be considered. In addition, there may be a case where a document to be copied by the MFP is a mixture of a confidential document and a general document.

  In Patent Documents 1 to 5 and Non-Patent Document 1, similarity determination is performed on the entire information such as a document. Therefore, if you try to determine whether an email attachment as described above or the entire original copied by MFP is similar to an existing confidential document, it may be partially similar, but as a whole In some cases, the determination result may not be similar. Rather, there is a problem that it is possible to determine that it is not similar to a confidential document by misusing it and inserting confidential information into a general document.

  In addition, for documents that are not yet registered in the confidential document DB, such as work-in-progress documents and draft documents, even if they are sent by e-mail, scanned, copied, or faxed by the MFP, the text of any confidential text In some cases, the determination result may be similar to the document. Then, there is a problem that it is determined that the document is not a confidential document because it includes the same content as the existing confidential document but the text is different.

  The present invention has been made in view of the above, and is an information analysis processing method, an information analysis processing program, an information analysis processing device, an information registration processing method, and information registration that prevents leakage of confidential information and improves convenience. It is an object of the present invention to provide a processing program, an information registration processing device, an information registration analysis processing method, and an information registration analysis processing program.

In order to solve the above-described problems and achieve the object, the invention according to claim 1 is an information analysis processing method executed by an information analysis processing device, wherein the information analysis processing device is registration information. A similar information search storage unit that stores a first feature amount that is a feature amount of partial information obtained by dividing target information, and an information classification storage unit that stores a second feature amount that is a feature amount of the registration target information for each attribute information The information analysis receiving unit receives the analysis target information that is the information to be analyzed from the external device, so that the information analysis receiving step for receiving the analysis request for the analysis target information, and the information division analysis unit, The analysis target information dividing step for dividing the analysis target information into partial analysis information that is a part of the analysis target information, and a similar information search unit based on the elements constituting the partial analysis information The partial information similar to the partial analysis information is calculated based on the calculated third feature amount and the first feature amount stored in the similar information search storage unit. And when the information classification unit determines that the partial analysis information is not similar to any of the partial information, the third feature amount of the partial analysis information and the information classification storage unit An information classification step for classifying the analysis target information into any one of the attribute information stored in the information classification storage unit based on the second feature amount stored in the information analysis processing unit, when said partial analysis information is determined to be similar to any of the partial information, and outputs the analysis result and the search result by the similar information retrieval step and said analyzed information, at least one said portion of the If the analysis information is determined not be similar to any of the partial information, to include an analysis result output step of outputting a classification result by the information classification step and the analyzed information to the search result as the analysis result It is characterized by.

  The invention according to claim 2 is the information analysis processing method according to claim 1, wherein the policy processing unit associates the analysis target information that has received the analysis result with the attribute information in association with the analysis. It further includes a policy processing step for performing processing based on a security policy that defines processing to be executed on target information.

  According to a third aspect of the present invention, in the information analysis processing method according to the first or second aspect, the analysis target information is information processed by an image forming apparatus.

According to a fourth aspect of the present invention, there is provided an information registration analysis processing method executed by an information registration analysis processing device, wherein the information registration analysis processing device is characterized by partial information obtained by dividing registration target information that is registration target information. An information registration receiving unit comprising: a similar information search storage unit that stores a first feature amount that is a quantity; and an information classification storage unit that stores a second feature amount that is a feature amount of the registration target information for each attribute information. but with the registered information and the attribute information and the including registration request accepting information registration receiving step of the registration target data, when the information division registration unit, which has received the registration request, the registration target information A registration target information dividing step for dividing the partial information, and a similar information registering unit calculates the first feature amount based on elements constituting the partial information, and the calculated first feature amount is used as the similar information. The partial information storage step for storing in the search storage unit, and the information classification registration unit calculates the second feature amount based on the elements constituting the registration target information, and the calculated second feature amount is used as the attribute information. In accordance with the classification, the classification storage step for storing in the information classification storage unit and the information analysis receiving unit receive the analysis target information that is the information to be analyzed from the external device, thereby receiving the analysis request for the analysis target information Information analysis acceptance step;
An information division analysis unit divides the analysis target information into partial analysis information that is a part of the analysis target information, and a similar information search unit is based on elements constituting the partial analysis information. A third feature amount of the partial analysis information is calculated, and based on the calculated third feature amount and the first feature amount stored in the similar information search storage unit, the similarity to the partial analysis information When the similar information search step for searching for partial information and the information classification unit determine that the partial analysis information is not similar to any of the partial information, the third feature amount of the partial analysis information and the information classification An information classification step for classifying the analysis target information into any of the attribute information stored in the information classification storage unit based on the second feature amount stored in the storage unit, and an information analysis processing unit If all the partial analysis information is determined to be similar to any of the partial information, and outputs the analysis result and the search result by the similar information retrieval step and said analyzed information, at least one of said partial analysis information An analysis result output step of outputting, as an analysis result, the analysis target information, the search result, and the classification result obtained by the information classification step when it is determined that the partial information is not similar to any of the partial information , To do.

The invention according to claim 5 is the information registration analysis processing method according to claim 4 , wherein the image data formed by the image forming apparatus is transmitted as information to be registered to the information registration receiving unit. step further comprises, the information registration receiving step, the information registration receiving unit, and the image data and the attribute information as the registered information, characterized in that accepting including pre Symbol registration request.

The invention according to claim 6 is the information registration analysis processing method according to claim 4 , wherein the image data formed by the image forming apparatus is sent as the analysis target information to the information analysis reception unit. The information analysis accepting step further includes a step of receiving the image data as the analysis target information from the external device.

The invention according to claim 7 is a program for causing a computer to execute the method according to claims 1 to 6 .

According to an eighth aspect of the present invention, there is provided a similar information search storage unit that stores a first feature quantity of partial information obtained by dividing registration target information that is registration target information, and a second feature quantity of the registration target information. An information classification storage unit for storing each information, an information analysis reception unit that receives an analysis request for the analysis target information by receiving analysis target information that is information to be analyzed from an external device, and the analysis target information. An information division analysis unit that divides into partial analysis information that is a part of the analysis target information, a third feature amount calculated by an element constituting the partial analysis information, and the first information stored in the similar information search storage unit based on the feature amount, and the partial analysis information similar information retrieval unit for retrieving the partial information that is similar to, if said partial analysis information is determined not to be similar to any of the partial information, the partial solution Based on the third feature amount of information and the second feature amount stored in the information classification storage unit, the analysis target information is classified into any of the attribute information stored in the information classification storage unit And when the partial analysis information is determined to be similar to any of the partial information, the analysis target information and the search result of the similar information search step are output as an analysis result. When it is determined that the two pieces of partial analysis information are not similar to any of the partial information , an information analysis processing unit that outputs the analysis target information, the search result, and the classification result by the information classification step as an analysis result; It is characterized by providing.

  According to the present invention, by dividing the analysis target information such as confidential information, the unit for determining the similarity of the analysis target information is made finer, and any of the partial analysis information that is a part constituting the analysis target information. It can be determined whether it is similar to the partial information. If it is determined that the partial analysis information is not similar to any registered partial information, the analysis target information can be classified into any category. Therefore, it is possible to more accurately determine whether or not the analysis target information is confidential information and to classify the information, thereby preventing leakage of confidential information and improving convenience.

  With reference to the accompanying drawings, an information analysis processing method, an information analysis processing program, an information analysis processing device, an information registration processing method, an information registration processing program, an information registration processing device, an information registration analysis processing method, and The best embodiment of the information registration analysis processing program will be described in detail.

  In the following embodiment, an example in which information is applied to document data of a confidential document will be described. Document data to be registered is referred to as registered document data, and document data to be analyzed is referred to as analysis document data. In addition, an example in which the information analysis processing device is applied to the document analysis processing unit, the information registration processing device is applied to the document registration unit, and the information registration analysis processing device is applied to the document registration analysis server is shown. It will never be done. That is, the present invention can be applied to any device or server that can register and analyze information such as documents.

  FIG. 1 is a diagram illustrating an entire configuration of a document registration analysis server and related peripheral devices according to the embodiment. As shown in FIG. 1, the document registration analysis server 1 mainly includes a similar document search DB (database) 300, a document classification DB 310, a document registration unit 100, a document analysis processing unit 200, and a document monitoring unit 400. It is configured. The document registration analysis server 1 is connected to the image log DB 2, the file server 3, and the MFP 5 mainly via a network or the like.

  First, an overview of document registration processing and document analysis processing will be described. The document monitoring unit 400 receives the setting of the folder to be monitored from the administrator D, and monitors whether or not the registered document data of the confidential document is stored in the document folder of the file server 3 according to the setting. Then, the document monitoring unit 400 acquires registered document data from the file server 3 and sends a registration request for registered document data to the document registration unit 100.

  The document registration unit 100 divides the registered document data of the confidential document received from the document monitoring unit 400 and registers it in the similar document search DB 300, or classifies the registered document data of the confidential document into one of the confidential categories. Register in DB310.

  The document analysis processing unit 200 receives from the MFP 5 document data printed by the general user E using the PC 4 or document data obtained by the general user F scanning, faxing, copying, etc. the paper document P. Then, an analysis request for analysis document data is received. The document analysis processing unit 200 selects either the received analysis document data that is similar to the registered document data of the existing confidential document or the confidential category of the existing confidential document even if it is not similar to the registered document data of the existing confidential document. Analyzes whether it can be classified into Then, based on the analysis result, the document analysis processing unit 200 processes the classified analysis document data according to the security policy set by the administrator D, that is, the notification process by e-mail or the image log DB 2. The recording process is performed.

  Next, the details of the document registration process for registered document data will be described. FIG. 2 is a configuration diagram of the document monitoring unit and the document registration unit. First, the document monitoring unit 400 monitors whether or not newly registered document data of a confidential document is stored in the document folder of the file server 3, and includes a monitoring folder setting unit 401, a document storage monitoring unit 402, a document A registration request unit 403 is mainly provided.

  The monitoring folder setting unit 401 receives a setting operation by the administrator D from the monitoring folder setting screen, and generates and outputs monitoring folder setting data according to the received setting operation. FIG. 3 is a diagram illustrating an example of a monitoring folder setting screen. As shown in FIG. 3, the monitoring folder setting screen 50 displays a confidential category name input field 51 for registered document data and a monitoring folder path input field 52. From the name input field 51 and the path input field 52, the administrator D receives an operation (setting operation) for the name of the confidential category and the path of the monitoring folder, generates related folder setting data, and stores it in a storage medium such as a memory. save. In addition, a list 53 of already set monitoring folders is displayed at the bottom of the monitoring folder setting screen 50. In FIG. 3, the registered document data stored in the monitoring folder “D: \ Documents \ Contracts \” is registered as a document of the confidential category “CONTRACT”.

  Here, the registered document data indicates, for example, document data composed of documents, but the document data to be registered is not limited to electronic document files. That is, for example, the registration process may be performed on scanned image data formed by scanning a paper document with an MFP (digital multifunction peripheral). In this case, when scanning a paper document, the configuration may be such that the confidential category of the image data is input from the operation panel of the MFP, or the “destination” of the scan delivery is mapped as the confidential category. Good.

  The document storage monitoring unit 402 reads the monitoring folder setting data generated by the monitoring folder setting unit 401, and whether registered document data of a new confidential document is stored in the document folder of the file server 3 according to the setting of the monitoring folder setting data. It is to monitor whether or not. When the registered document data of the new confidential document is stored in the file server 3, the document storage monitoring unit 402 refers to the registered document data with reference to the file path of the stored registered document data of the confidential document and the monitoring folder setting data. The confidential category corresponding to the data is sent to the document registration request unit 403.

  When the registered document data is stored in the file server 3, the document registration request unit 403 reads the registered document data from the file server 3, reads the registered document data, the file path of the registered document data, and the confidentiality of the registered document data. A registration request including a category is sent to the document registration receiving unit 101 of the document registration unit 100.

  Next, the document registration unit 100 registers the registered document data of the confidential document transmitted from the document monitoring unit 400, and includes a document registration reception unit 101, a document registration processing unit 102, and a document piece division registration unit. 103, a similar document search DB registration unit 104, and a document classification DB registration unit 105, which are connected to the similar document search DB 300 and the document classification DB 310.

  When the document registration receiving unit 101 receives a registration request including the file path of the registered document data, the confidential category of the registered document data, and the registered document data from the document registration request unit 403 of the document monitoring unit 400, the received file The path, confidential category, and registered document data are sent to the document registration processing unit 102.

  The document registration processing unit 102 receives the file path, the confidential category, and the registered document data from the document registration receiving unit 101, and sends the file path, the confidential category, and the registered document data to the document piece division registration unit 103, and the confidential category And registered document data are sent to the document classification DB registration unit 105.

  The document piece division registration unit 103 receives the file path, the confidential category, and the registered document data from the document registration processing unit 102, and converts the received registration document data into a plurality of registered document pieces based on a predetermined division rule. Dividing and assigning piece numbers. Also, the document piece division registration unit 103 sends the file path, the confidential category, the piece number, and the registered document piece to the similar document search DB registration unit 104 as a set. If the registered document data is image data, text information is obtained by performing OCR processing on the image data, and the obtained text information is divided into registered document pieces to perform registration processing.

  Here, the above-mentioned “predetermined division rule” is, for example, a rule for dividing by 500 characters, such as dividing every 500 characters, a rule for dividing a half page of registered document data as one registered document piece, There are rules for dividing by paragraph, rules for dividing by paragraphs, etc.

  The similar document search DB registration unit 104 calculates a feature amount (first feature amount) that characterizes each registered document piece by using the elements constituting each registered document piece divided by the document piece division registration unit 103, and calculates the calculated feature. The amount is stored in the similar document search DB 300 in association with the file path, the confidential category, and the piece number. As a result, a similarity search in units of registered document pieces is enabled. This first feature amount is used to search whether the registered document pieces are similar. The first feature amount is calculated by a known method based on the elements constituting each registered document piece.

  The document classification DB registration unit 105 receives the confidential category and the registered document data from the document registration processing unit 102, and calculates a classification feature amount (second feature amount) that characterizes the registered document data based on elements constituting the registered document data. The calculated classification feature amount is stored in the document classification DB 310 as a new classification feature amount for each confidential category. This enables classification of registered document data. This second feature amount is used to classify the registered document data. The second feature amount is calculated by a known method based on elements constituting each registered document data.

  The similar document search DB 300 is a recording medium such as an HDD (Hard Disk Drive) that registers the calculated feature amount of a registered document piece in association with a file path, a confidential category, and a piece number.

  The document classification DB 310 is a recording medium such as an HDD for registering the calculated classification feature amount of registered document data for each confidential category.

  Next, details of the document analysis processing of the analysis document data will be described. FIG. 4 is a configuration diagram of the document analysis processing unit. The document analysis processing unit 200 analyzes analysis document data received from an external device such as the MFP 5 and performs processing according to the security policy based on the analysis result. The document analysis reception unit 201 and the document analysis processing unit 202, a document piece division analysis unit 203, a similar document search unit 204, a document classification unit 205, a policy processing unit 206, and a policy setting unit 207.

  As described above, the document analysis reception unit 201 is analysis document data obtained by performing printing or the like according to an instruction from the PC 4 or scanning, fax transmission, copying, or the like of the paper original P, and identification information unique to the user. The analysis of the analysis document data is performed by receiving the user ID and the document processing type (for example, scan processing) indicating which processing has been performed on the analysis document data to be received from the MFP 5 via the network. It accepts requests. Then, the document analysis reception unit 201 sends the received analysis document data, user ID, and document processing type to the document analysis processing unit 202.

  Here, the analysis document data indicates, for example, document data composed of documents, but the document data to be analyzed is not limited to an electronic document file. In other words, for example, the analysis processing may be performed on the scanned image data formed by scanning a paper document with an MFP (digital multifunction peripheral).

  The document analysis processing unit 202 receives the analysis document data, the user ID, and the document processing type from the document analysis reception unit 201 and sends the analysis document data to the document piece division analysis unit 203.

  Upon receiving the analysis document data from the document analysis processing unit 202, the document piece division analysis unit 203 divides the analysis document data into a plurality of analysis document pieces based on a predetermined division rule and assigns analysis piece numbers. To do. Further, the document piece division analysis unit 203 associates the analysis piece number and the analysis document piece and sends them to the similar document search unit 204. If the analysis document data is image data, text information is acquired by performing OCR processing on the image data, and the acquired text information is divided into analysis document pieces to perform analysis processing. Here, the division rule defined in advance is the same as in the registration process.

  The similar document search unit 204 receives the analysis piece number and the analysis document piece associated with each other from the document piece division analysis unit 203, and a registered document piece similar to the received analysis document piece is registered in the similar document search DB 300. Whether or not there is a search. That is, the similar document search unit 204 calculates a feature amount (third feature amount) that characterizes the analysis document piece divided by the document piece division analysis unit 203 from the constituent elements of the analysis document piece, and the calculated feature amount and the similar document The feature amount in the registered document piece stored in the search DB 300 is compared. If any of the feature quantities of the registered document piece is similar to the feature quantity of the analysis document piece, it is determined that the analysis document piece and the registered document piece are similar, and the registered document piece If any of the feature quantities similar to the feature quantity of the analysis document piece does not exist, it is determined that the analysis document piece is not similar to the registered document piece. The similarity determination of the feature amount can be determined to be similar when the difference between the numerical values of the feature amounts to be compared is within a predetermined range, but is not limited thereto.

  When a registered document piece similar to the analysis document piece is searched as a result of the similarity search, the similar document search unit 204 includes the file path, the piece number, and the confidential category associated with the registered document piece. The search result is generated, and the generated search result is sent to the document piece division analysis unit 203.

  FIG. 5 is a diagram illustrating an example of a search result generated by the similar document search unit. As shown in FIG. 5, the search result 60 always includes an analysis document piece and an analysis piece number of the analysis document piece. Then, the file path, piece number, and confidential category are included as information on similar registered document pieces. If there are a plurality of similar registered document pieces, information on all the similar registered document pieces is included in the search result. FIG. 5 shows a case where information on two similar registered document pieces (1) and (2) is retrieved.

  On the other hand, if the registered document piece similar to the analysis document piece is not searched as a result of the similarity search, the similar document search unit 204 does not include anything in the information related to the similar registered document piece and is empty. The search result including only the analysis document piece and the piece number is generated, and the generated search result is sent to the document piece division analysis unit 203.

  Further, when the document piece division analysis unit 203 described above receives the search results for all the analysis document pieces to which the analysis piece numbers are given and sent from the similar document search unit 204, the search result is accumulated. The generated piece analysis result is generated, and the generated piece analysis result is sent to the document analysis processing unit 202. FIG. 6 is a diagram illustrating an example of a piece analysis result generated by the document piece division analysis unit. As shown in FIG. 6, the piece analysis result 61 includes all the search results (see FIG. 5) sent from the similar document search unit 204. In FIG. 6, search results (1), (2),... (N) from 1 to N are included.

  When the document analysis processing unit 202 further receives the piece analysis result sent from the document piece division analysis unit 203, the analysis piece of the search result in which a similar registered document piece could not be searched from the piece analysis result. The number and the analysis document piece are acquired, and the acquired analysis piece number and the analysis document piece are sent to the document classification unit 205.

  The document classifying unit 205 receives the analysis piece number and the analysis document piece of the search result sent out from the document analysis processing unit 202, which could not be searched for similar registered document pieces, and based on the received feature amount of the analysis document piece The analysis document data is classified into any confidential category. That is, the document classification unit 205 calculates the feature amount of the received analysis document piece, and the registered document data stored in the calculated feature amount and one of the confidential categories (confidential categories A, B, and C) of the document classification DB 310. The classified feature amount is compared to determine which classified category it is classified into, and a piece classification result obtained by performing the classification process is generated. Then, the document classification unit 205 generates a classification result obtained by accumulating the generated piece classification results, and sends the generated classification result to the document analysis processing unit 202.

  FIG. 7 is a diagram illustrating an example of the piece classification result generated by the document classification unit. As shown in FIG. 7, the piece classification result 62 always includes an analysis document piece and an analysis piece number of the analysis document piece. Then, all classified categories classified based on the feature amount of the analysis document piece calculated by the document classification unit 205 are included. FIG. 7 shows that the analysis document piece is classified into two confidential categories (1) and (2).

  FIG. 8 is a diagram illustrating an example of the classification result generated by the document classification unit. As shown in FIG. 8, the classification result 63 includes all the piece classification results (see FIG. 7) generated by the document classification unit 205. In FIG. 8, pieces classification results (1), (2),... (N) from 1 to N are included.

  Then, when the registered document piece similar to all the analyzed document pieces is searched by the similar document search unit 204, the document analysis processing unit 202 described above is similar to the user ID, the analysis document data, the document processing type, and the like. The piece analysis result searched by the document search unit 204 is sent to the policy processing unit 206 as an analysis result.

  In addition, when the similar document search unit 204 does not search for a registered document piece that is similar to at least one analysis document piece, the document analysis processing unit 202 uses the user ID, the analysis document data, the document processing type, and the similarity. The piece analysis result searched by the document search unit 204 and the classification result classified by the document classification unit 205 are sent to the policy processing unit 206 as analysis results.

  The policy processing unit 206 acquires a security policy, refers to the acquired security policy, and with respect to the user ID, analysis document data, document processing type, piece analysis result, and classification result received from the document analysis processing unit 202, Processing according to the setting contents of the security policy for each confidential category, that is, e-mail transmission processing by sending analysis document data to the mail server 7, and image log by sending analysis document data to the screen log server 6 The recording process is performed.

  The policy setting unit 207 acquires the monitoring folder setting data generated by the monitoring folder setting unit 401, displays a security policy setting screen for each confidential category set in the acquired monitoring folder setting data, and A policy setting operation by D is received, and a security policy is generated according to the received setting operation. FIG. 9 is a diagram illustrating an example of a security policy setting screen generated by the policy setting unit. FIG. 10 is a diagram showing the structure of the set security policy. As shown in FIG. 9, the security policy setting screen 70 includes a confidential category display column 71, a document processing type selection column 72 indicating which processing has been performed on the analysis document data, and a classified analysis document. An action selection field 73 indicating processing for data is displayed. In FIG. 9, a set security policy 74 is displayed at the bottom of the security policy setting screen.

  Here, in the above document analysis processing, MFP 5 (digital multifunction peripheral) is illustrated as an external device that requests document analysis. However, the analysis document data is not limited to documents processed by the digital multifunction peripheral. Absent. That is, for example, an electronic document file placed in a specific document folder on a file server is monitored and used as analysis document data, or an e-mail body passing through a mail server or an attached file to an e-mail is analyzed. It may be data.

  Next, processing by the document registration analysis server 1 according to the present embodiment configured as described above will be described. First, document registration processing by the document registration unit 100 in the document registration analysis server 1 will be described. FIG. 11 is a flowchart illustrating a procedure of document registration processing by the document registration unit according to the embodiment.

  First, the document registration receiving unit 101 receives a file path of registered document data, a confidential category of registered document data, and registered document data from the document registration request unit 403 in the document monitoring unit 400 (step S10). The path, confidential category, and registered document data are sent to the document registration processing unit 102.

  Then, the document registration processing unit 102 receives the file path, confidential category, and registered document data from the document registration receiving unit 101 (step S11). Then, the document registration processing unit 102 sends the received file path, confidential category, and registered document data to the document piece division registration unit 103. In addition, the document registration processing unit 102 sends the confidential category and registered document data to the document classification DB registration unit 104.

  Next, the document piece division registration unit 103 receives the file path, the confidential category, and the registered document data from the document registration processing unit 102, divides the received registration document data into registered document pieces, and gives a piece number (step). S12). Then, the document piece division registration unit 103 sends the file path, the confidential category, the piece number, and the registered document piece to the similar document search DB registration unit 104 as a set.

  Next, when the similar document search DB registration unit 104 receives the file path, the confidential category, the piece number, and the registered document piece from the document piece division registration unit 103, the similar document search DB registration unit 104 calculates the feature amount of each registered document piece, and calculates the calculated feature. The amount is stored in the similar document search DB 300 in association with the file path, the confidential category, and the piece number (step S13).

  On the other hand, when the document classification DB registration unit 105 receives the confidential category and the registered document data from the document registration processing unit 102, the document classification DB registration unit 105 calculates a classification feature amount of the registered document data, and calculates the calculated classification feature amount for each confidential category. Store in the classification DB 310 (step S14).

  As described above, in the document registration process by the document registration unit 100, the feature amount of the registered document piece obtained by dividing the registered document data and the classification feature amount of the registered document data are stored. As a result, it is possible to perform similarity search in units of document pieces and classification of registered document data.

  Next, document analysis processing by the document analysis processing unit 200 in the document registration analysis server 1 will be described. FIG. 12 is a flowchart illustrating a procedure of document analysis processing by the document analysis processing unit in the embodiment.

  The document analysis receiving unit 201 receives the analysis request for the analysis document data by receiving the user ID, the analysis document data, and the document processing type from the MFP 5 (step S30). Then, the document analysis reception unit 201 sends the received user ID, analysis document data, and document processing type to the document analysis processing unit 202.

  When the document analysis processing unit 202 receives the analysis document data, the user ID, and the document processing type from the document analysis reception unit 201 (step S31), the document analysis processing unit 202 sends the analysis document data to the document piece division analysis unit 203.

  When the document piece division analysis unit 203 receives the analysis document data from the document analysis processing unit 202, the document piece division analysis unit 203 divides the analysis document data, and the similar document search unit 204 performs analysis processing (step S32). That is, the document piece division analysis unit 203 divides the analysis document data received from the document analysis processing unit 202 and assigns the analysis piece number, and associates the analysis piece number with the analysis document piece to the similar document search unit 204. Send it out. Then, the similar document search unit 204 receives the associated analysis piece number and the analysis document piece from the document piece division analysis unit 203, and the registered document piece similar to the received analysis document piece is registered in the similar document search DB 300. Search whether or not. Then, the similar document search unit 204 generates a search result, and the document piece division analysis unit 203 generates a piece analysis result obtained by accumulating all the search results, and sends it to the document analysis processing unit 202.

  The document analysis processing unit 202 determines whether or not all registered document pieces similar to each search result included in the piece analysis result have been searched (step S33). When all registered document pieces similar to the piece analysis result are searched (step S33: Yes), the document analysis processing unit 202 searches the user ID, the analysis document data, the document processing type, and the similar document search unit 204. The piece analysis result thus sent is sent as an analysis result to the policy processing unit 206 (step S36).

  On the other hand, when all registered document pieces similar to the piece analysis result have not been searched, that is, when no registered document pieces similar to at least one piece analysis result have been searched (step S33: No), the document analysis processing unit 202 Acquires the analysis piece number and the analysis document piece of the search result from which the similar registered document piece was not searched from the piece analysis result (step S34). Then, the document analysis processing unit 202 sends the acquired analysis piece number and analysis document piece to the document classification unit 205.

  Upon receiving the analysis piece number and the analysis document piece from the document analysis processing unit 202, the document classification unit 205 classifies the received analysis document piece (step S35). That is, the document classification unit 205 calculates a feature amount of the received analysis document piece, determines which classified category is classified according to the calculated feature amount, and generates a piece classification result. Then, the document classification unit 205 accumulates the generated piece classification results to generate a classification result, and sends the generated classification result to the document analysis processing unit 202.

  Then, the document analysis processing unit 202 analyzes the user ID, the analysis document data, the document processing type, the piece analysis result searched by the similar document search unit 204, and the classification result classified by the document classification unit 205. As a result, it is sent to the policy processing unit 206 (step S36).

  Then, the policy processing unit 206 refers to the security policy, and with respect to the user ID, analysis document data, document processing type, piece analysis result, and classification result received from the document analysis processing unit 202, the security processing unit 206 Process according to the policy settings.

  As described above, in the document analysis process performed by the document analysis processing unit 200, a registered document piece similar to the analysis document piece is searched based on the feature amount of the analysis document piece obtained by dividing the analysis document data. If no registered document piece similar to the analyzed document piece is found, the analyzed document data can be classified into any confidential category.

  As described above, in the document registration analysis server 1 according to the present embodiment, the feature amount of the registered document piece is stored in advance in the similar document search DB for each document piece, and the registered document data of the entire document (registered document data) is stored. By storing the classification feature quantity in the document classification DB, it is possible to perform a similar search in units of analysis document pieces and further classify the confidential categories of analysis document data. That is, by dividing the analysis document data such as confidential information, the unit for determining the similarity of the analysis document data can be made finer, and it can be determined whether the analysis document piece is similar to any registered document piece. If it is determined that the analysis document piece is not similar to any registered document piece, the analysis document data can be classified into any confidential category. Therefore, it is possible to more accurately determine whether the analysis document data is confidential information. In addition, since information determined as confidential information can be processed in accordance with the security policy, leakage of confidential information can be prevented and convenience can be improved.

  In the present embodiment, an example in which the document monitoring unit and the document registration unit are configured separately is described in consideration of the case of monitoring document folders of a plurality of file servers. However, the present invention is not limited to this. That is, the document monitoring unit and the document registration unit may be integrated into one component unit.

  FIG. 13 is a diagram illustrating a hardware configuration of the document registration analysis server according to the present embodiment. The document registration analysis server 1 according to the present embodiment includes a control device such as a CPU (Central Processing Unit) 5001, a storage device such as a ROM (Read Only Memory) 5002 and a RAM (Random Access Memory) 5003, an HDD and a CD drive. An external storage device 5004 such as a device, a display device 5005 such as a display device, an input device 5006 such as a keyboard and a mouse, a communication I / F 5007, and a bus 5008 for connecting them are provided. The hardware configuration is used.

  The document registration analysis program executed by the document registration analysis server 1 of the present embodiment is a file in an installable format or an executable format, and is a CD-ROM, flexible disk (FD), CD-R, DVD (Digital Versatile). Disk) and the like are provided by being recorded on a computer-readable recording medium.

  Further, the document registration analysis program executed by the document registration analysis server 1 of the present embodiment is stored on a computer connected to a network such as the Internet and is provided by being downloaded via the network. Also good. Further, the document registration analysis program executed by the document registration analysis server 1 of the present embodiment may be provided or distributed via a network such as the Internet.

  Further, the document registration analysis program according to the present embodiment may be provided by being incorporated in advance in a ROM or the like.

  The document registration analysis program executed by the document registration analysis server 1 according to the present embodiment includes the above-described units (document registration receiving unit 101, document registration processing unit 102, document piece division registration unit 103, similar document search DB registration unit 104). , Document classification DB registration unit 105, document analysis reception unit 201, document analysis processing unit 202, document piece division analysis unit 203, similar document search unit 204, document classification unit 205, policy processing unit 206, and policy setting unit 207). As the actual hardware, the CPU (processor) reads and executes the document registration analysis program from the storage medium, and the above-described units are loaded on the main storage device. Document registration processing unit 102, document piece division registration unit 103, similar document search DB registration unit 104, document classification The B registration unit 105, the document analysis reception unit 201, the document analysis processing unit 202, the document piece division analysis unit 203, the similar document search unit 204, the document classification unit 205, the policy processing unit 206, and the policy setting unit 207 are provided on the main storage device. It is to be generated.

It is a figure which shows the whole structure of the document registration analysis server concerning an embodiment, and a related peripheral device. It is a block diagram of a document monitoring part and a document registration part. It is a figure which shows an example of the monitoring folder setting screen. It is a block diagram of a document analysis processing part. It is a figure which shows an example of the search result produced | generated by the similar document search part. It is a figure which shows an example of the piece analysis result produced | generated by the document piece division | segmentation analysis part. It is a figure which shows an example of the piece classification | category result produced | generated by the document classification | category part. It is a figure which shows an example of the classification result produced | generated by the document classification | category part. It is a figure which shows an example of the security policy setting screen produced | generated by the policy setting part. It is a figure which shows the structure of the set security policy. It is a flowchart which shows the procedure of the document registration process by the document registration part in embodiment. It is a flowchart which shows the procedure of the document analysis process by the document analysis process part in embodiment. It is a figure which shows the hardware constitutions of the document registration analysis server of this Embodiment. It is explanatory drawing of the product which classifies a vast amount of text information in a prior art automatically.

Explanation of symbols

1 Document registration analysis server 2 Image log DB
3 File server 4 PC
5 MFP
DESCRIPTION OF SYMBOLS 100 Document registration part 101 Document registration reception part 102 Document registration process part 103 Document piece division | segmentation registration part 104 Similar document search DB registration part 105 Document classification DB registration part 200 Document analysis process part 201 Document analysis reception part 202 Document analysis process part 203 Document Piece division analysis unit 204 Similar document search unit 205 Document classification unit 206 Policy processing unit 207 Policy setting unit 300 Similar document search DB
310 Document classification DB
400 Document monitoring unit 401 Monitoring folder setting unit 402 Document storage monitoring unit 403 Document registration request unit A, B, C Confidential category D Administrator E, F General user

Claims (8)

In the information analysis processing method executed by the information analysis processing device,
The information analysis processing device includes:
A similar information search storage unit that stores a first feature quantity that is a feature quantity of partial information obtained by dividing registration target information that is registration target information;
An information classification storage unit that stores, for each attribute information, a second feature amount that is a feature amount of the registration target information,
An information analysis accepting step for receiving an analysis request for the analysis target information by receiving analysis target information that is information to be analyzed from an external device;
An information division analysis unit that divides the analysis target information into partial analysis information that is a part of the analysis target information; and
The similar information search unit calculates a third feature amount of the partial analysis information based on an element constituting the partial analysis information, and the calculated third feature amount and the similar information search storage unit stored in the third information amount A similar information search step for searching for the partial information similar to the partial analysis information based on the first feature amount;
When the information classification unit determines that the partial analysis information is not similar to any of the partial information, the third feature amount of the partial analysis information and the second feature amount stored in the information classification storage unit Based on the information classification step of classifying the analysis target information into any of the attribute information stored in the information classification storage unit,
When it is determined that all the partial analysis information is similar to any of the partial information, the information analysis processing unit outputs the analysis target information and the search result by the similar information search step as an analysis result, and at least one When it is determined that the two pieces of partial analysis information are not similar to any of the partial information , an analysis result output step of outputting the analysis target information, the search result, and a classification result by the information classification step as an analysis result;
An information analysis processing method comprising:   A policy processing step in which a policy processing unit performs processing based on a security policy that defines processing to be executed on the analysis target information in association with the attribute information, on the analysis target information that has received the analysis result; The information analysis processing method according to claim 1, further comprising:   The information analysis processing method according to claim 1, wherein the analysis target information is information processed by an image forming apparatus. In the information registration analysis processing method executed by the information registration analysis processing device,
The information registration analysis processing device includes:
A similar information search storage unit that stores a first feature quantity that is a feature quantity of partial information obtained by dividing registration target information that is registration target information;
An information classification storage unit that stores, for each attribute information, a second feature amount that is a feature amount of the registration target information,
Information registration receiving unit, and the information registration receiving step of receiving a including registration request and the attribute information of the registration target data and the registration target data,
When the information division registration unit accepts the registration request, a registration target information division step for dividing the registration target information into the partial information;
A partial information storage step in which a similar information registration unit calculates the first feature amount based on an element constituting the partial information, and stores the calculated first feature amount in the similar information search storage unit;
An information classification registration unit calculates the second feature quantity based on elements constituting the registration target information, and stores the calculated second feature quantity in the information classification storage unit according to the classification of the attribute information A classification storage step;
An information analysis accepting step for receiving an analysis request for the analysis target information by receiving analysis target information that is information to be analyzed from an external device;
An information division analysis unit that divides the analysis target information into partial analysis information that is a part of the analysis target information; and
The similar information search unit calculates a third feature amount of the partial analysis information based on an element constituting the partial analysis information, and the calculated third feature amount and the similar information search storage unit stored in the third information amount A similar information search step for searching for the partial information similar to the partial analysis information based on the first feature amount;
When the information classification unit determines that the partial analysis information is not similar to any of the partial information, the third feature amount of the partial analysis information and the second feature amount stored in the information classification storage unit Based on the information classification step of classifying the analysis target information into any of the attribute information stored in the information classification storage unit,
When it is determined that all the partial analysis information is similar to any of the partial information, the information analysis processing unit outputs the analysis target information and the search result by the similar information search step as an analysis result, and at least one When it is determined that the two pieces of partial analysis information are not similar to any of the partial information , an analysis result output step of outputting the analysis target information, the search result, and a classification result by the information classification step as an analysis result;
An information registration analysis processing method characterized by comprising: An information registration sending step of sending image data formed by the image forming apparatus to the information registration receiving unit as the registration target information;
The information registration receiving step, the information registration receiving unit, information registration according to said image data and the attribute information as the registered information in the claim 4, characterized in that accepting including pre Symbol registration request Analysis processing method. An information analysis sending step of sending image data formed by the image forming apparatus to the information analysis receiving unit as the analysis target information;
5. The information registration analysis processing method according to claim 4 , wherein in the information analysis reception step, the information analysis reception unit receives the image data as the analysis target information from the external device. The program which makes a computer perform the method of Claims 1-6 . A similar information search storage unit that stores a first feature amount of partial information obtained by dividing registration target information that is registration target information;
An information classification storage unit that stores the second feature amount of the registration target information for each attribute information;
An information analysis receiving unit that receives an analysis request for the analysis target information by receiving analysis target information that is information to be analyzed from an external device;
An information division analysis unit that divides the analysis target information into partial analysis information that is a part of the analysis target information;
The partial information similar to the partial analysis information is searched based on the third feature value calculated by the elements constituting the partial analysis information and the first feature value stored in the similar information search storage unit. Similar information search unit,
When it is determined that the partial analysis information is not similar to any of the partial information, based on the third feature amount of the partial analysis information and the second feature amount stored in the information classification storage unit, An information classification unit for classifying the analysis target information into any of the attribute information stored in the information classification storage unit;
When it is determined that all the partial analysis information is similar to any of the partial information, the analysis target information and the search result by the similar information search step are output as an analysis result, and at least one partial analysis information is When it is determined that the partial information is not similar to any of the partial information , an information analysis processing unit that outputs the analysis target information, the search result, and a classification result by the information classification step as an analysis result;
An information analysis processing apparatus comprising:

Images