JPWO2008126149A1 - Document anonymization device - Google Patents

Abstract

A proper noun in the document is extracted, and for each extracted proper noun, at least one attribute information related to the proper noun is obtained, and a character string to be replaced with the proper noun from which the attribute information is obtained is obtained as the proper noun. Is generated by adding the attribute information to the anonymized expression, and the generated character string is replaced with a proper noun. As a result, the anonymization of the document in which the decrease in the amount of information is suppressed is performed.

Description

  The present invention relates to a technique for anonymizing proper nouns present in a document.

In recent years, digitization of documents has progressed, and an enormous amount of documents has been accumulated in the form of electronic data. The number of documents that can be browsed via a communication network tends to increase rapidly.
A document may have proper nouns, which are personal information that is not desirable to publish. For example, in an organization such as a company, in order for each employee to take a more appropriate response, the episodes on which information should be shared are disclosed in documents. Such documents often have proper nouns that should not be disclosed in real names. For this reason, a document anonymization device that anonymizes such proper nouns has been devised.

  As a conventional document anonymization device, for example, there is one described in Patent Document 1. In the conventional document anonymization apparatus described in Patent Document 1, a proper noun is extracted from a document to be anonymized, it is determined whether the extracted proper noun should be anonymized, and the proper noun determined to be anonymized Has become anonymized. Accordingly, for example, when a document as shown in FIG. 1A is anonymized, the document as shown in FIG. 1B is generated by the anonymization. In the document shown in FIG. 1B, “Mr. Sato” is replaced with “Mr. A”, “Maruhishi Electric” is replaced with “Company J”, “Mr. Omura” is replaced with “T”, and “Mr. Sakamoto” is replaced with “Mr. B”. Has been. For example, “Mr. Sato” may be replaced with “Mr. XXX”.

  Such substitutions make the proper nouns anonymized unspecified. However, if the proper noun is simply anonymized, the amount of information decreases due to the anonymization, and the decrease may make it very difficult to grasp the meaning of the original document. Therefore, the important meaning of the original document was likely to be lost.

  As a conventional document anonymization device capable of suppressing a decrease in the amount of information due to anonymization, for example, there is one described in Patent Document 2. In the conventional document anonymization apparatus described in Patent Document 2, a proper noun can be replaced with an arbitrary character string by setting a character string to be replaced with the proper noun for each proper noun. .

However, in order to realize such substitution, a character string must be set for each proper noun. Naturally, the character string must be set in consideration of other character strings to be set. This is because it is necessary to avoid replacing a plurality of personal names existing in a document with the same character string. Therefore, it is not possible to simply set a character string, and if the number of proper nouns to be replaced increases, a very complicated operation must be performed for setting the character string. For this reason, it is considered important to be able to reduce the amount of information due to anonymization without individually setting character strings to be replaced with proper nouns.
JP 2002-269081 A JP 2002-259368 A

  An object of the present invention is to provide a document anonymization device capable of suppressing a decrease in the amount of information due to anonymization of proper nouns without individually setting character strings to be replaced with proper nouns.

  The program according to the first aspect of the present invention includes a proper noun extraction function for extracting proper nouns in a document to a computer that can be used as a document anonymization device for anonymizing proper nouns in a document, and proper noun extraction. An information acquisition function for acquiring at least one attribute information related to the proper noun extracted by the function, and when the attribute information is acquired by the information acquisition function, the proper noun extracted by the proper noun extraction function using the attribute information; A character string generation function for generating a character string to be replaced and a character string replacement function for replacing the proper noun extracted from the material by the proper noun extraction function with the character string generated by the character string generation function are realized.

  The program according to the second aspect further includes a selection function for selecting a character string to be generated by the character string generation function in addition to the function provided in the first aspect. A character string is generated using the attribute information acquired by the information acquisition function in response to the user's selection instruction via.

  The document anonymization apparatus of the present invention is based on the premise that the proper nouns in the document are anonymized, proper noun extraction means for extracting proper nouns in the document, and attribute information related to proper nouns extracted by the proper noun extraction means A character string for generating a character string to be replaced with the proper noun extracted by the proper noun extraction means using the attribute information when the information acquisition means acquires the attribute information. And a character string replacement means for replacing the proper noun extracted from the material by the proper noun extraction means with the character string generated by the character string generation means.

  The document anonymization method of the present invention is for anonymizing proper nouns in a document by a computer, extracts proper nouns in a document, and for each extracted proper noun, attribute information related to the proper noun The character string to be replaced with the proper noun that acquired the attribute information is generated by adding the attribute information to the anonymized expression of the proper noun, and the generated character string is generated as the proper noun. Replace with

  In the present invention, a proper noun in a document is extracted, and at least one attribute information related to the proper noun is obtained for each of the extracted proper nouns, and a character string to be replaced with the proper noun obtained from the attribute information is obtained. Then, the proper noun is generated by adding the attribute information to the anonymized expression, and the generated character string is replaced with the proper noun. Since the attribute information is added to the character string replaced with the proper noun, a decrease in the amount of information can be suppressed as compared with the case where the proper noun is simply turned over. The need to perform complicated operations such as individually setting character strings to be replaced with proper nouns is avoided. Since the number, type, or combination of the attribute information used for generating the character string can be arbitrarily changed, a desired character string can be generated.

It is a figure which shows the example of a document. It is a figure which shows the document anonymized by the conventional document anonymization apparatus described in patent document 1. FIG. It is a figure which shows the structure of the document anonymization apparatus by embodiment of this invention. It is a figure which shows the structure of an applicable organization system information table. It is a figure which shows the structure of a corresponding customer organization system information table. It is a figure which shows the structure of a company information table. It is a figure which shows the structure of an in-house employee information table. It is a figure which shows the document after anonymizing the document of FIG. 1A. It is a flowchart of a document anonymization process. It is a flowchart of an additional expression production | generation process. It is a flowchart of an organization name additional expression generation process. It is a flowchart of a personal name additional expression production | generation process. It is a figure which shows the production | generation method of an organization name additional expression. It is a figure which shows the production | generation method of personal name additional expression. It is a figure which shows the example of anonymization of a document. It is a figure which shows the anonymization example which excludes the redundancy of data. It is a figure explaining the anonymization method of the proper noun which makes it possible to exclude redundancy. It is a flowchart of a proper noun replacement process. It is a figure which shows the selection method of the attribute information addition expression to produce | generate (organization). It is a figure which shows the selection method of the attribute information additional expression to produce | generate (individual). It is a flowchart of an additional expression selection process. It is a figure which shows an example of the hardware constitutions of the computer which can implement | achieve the document anonymization apparatus by this Embodiment.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
FIG. 1 is a diagram showing a configuration of a document anonymization apparatus according to an embodiment of the present invention. The document anonymization device 2 is constructed to anonymize a document that an organization (company) makes available to employees or external persons, for example. As shown in FIG. 1, the information acquisition unit 21, proper noun extraction unit 22, proper noun conversion unit 33, information storage unit 24, character string replacement unit 25, input unit 26, and display control unit 27 are provided. ing. The display control unit 27 realizes an anonymized document and various types of information for anonymization on the display device DP.

The operation in the above configuration will be described.
The material acquisition unit 21 acquires material as a document to be anonymized. The acquired materials are output to the proper noun extraction unit 22 and the character string replacement unit 25, respectively. The proper noun extraction unit 22 performs morphological analysis of the material input from the material acquisition unit 21 and extracts a proper noun. The extraction result is output to the proper noun conversion unit 23. Since the material acquisition unit 21 acquires a material as a document to be anonymized, “material” is hereinafter used in the meaning of the document to be anonymized unless otherwise specified. The document itself to be anonymized is not limited.

  The proper noun conversion unit 23 generates a character string to be replaced with the proper noun for each proper noun extracted by the proper noun extraction unit 22. For this purpose, an information acquisition unit 23a and an additional expression generation unit 23b are provided. The information acquisition unit 23a acquires information used for generating a character string from the information storage unit 24, and the additional expression generation unit 23b uses the information acquired by the information acquisition unit 23a to replace characters with proper nouns. A sequence (additional expression) is generated.

  The information storage unit 24 stores a corresponding organization system information table, a corresponding customer organization system information table, a company information table, and an in-house employee information table for character string generation. Here, these tables will be specifically described with reference to FIGS. 3A to 3D.

  FIG. 3A is a diagram showing the configuration of the corresponding organizational structure information table. The table summarizes information about a system (business division, department, group, etc.) that anonymizes materials in a company organization. As shown in FIG. 3A, items of a role name, a person in charge name, a belonging organization, and an attribute are provided. Thus, for each employee who is given a role, the role name, the name, the organization to which the user belongs, and the personal information of the attribute are stored in one record.

  FIG. 3B is a diagram showing a configuration of a corresponding customer organization system information table. The table is a summary of information on the system that actually requests work in an organization (company) corresponding to a customer. As shown in FIG. 3B, the configuration is basically the same as the corresponding organizational structure information table.

  FIG. 3C is a diagram illustrating a configuration of a company information table. The table summarizes information about each company that may appear in the material. As shown in FIG. 3C, information on each item of company name, annual sales, location, relationship, and type of business is stored for each company. The relationship is based on a company (in-house) with a system for anonymizing materials.

  FIG. 3D is a diagram illustrating a configuration of an in-house employee information table. The table summarizes the personal information for each employee of the company. As items for storing the personal information, as shown in FIG. 3D, a name, employee number, affiliation, title, telephone number, and the like are provided.

  The relevant organizational structure information table and the in-house employee information table are actually created from a database (DB) prepared for managing employees working in the company or from the DB. The other two tables are actually prepared from the DB prepared for managing information necessary for the company to develop the business or from the DB. In the present embodiment, such a table is used, and more specifically, a character string to be replaced with a proper noun is generated using information stored in the table. Since the generation of the character string using the information compensates for the decrease in the amount of information associated with the anonymization of the proper noun, it is possible to reliably avoid losing important meaning.

Returning to the description of FIG.
The information acquisition unit 23a performs a search using each proper noun as a key for each table, extracts a record having the key, and makes the additional expression generation unit 23b. The additional expression generation unit 23b generates a character string to be replaced with a proper noun, adds record information to the character string, and generates a character string to be actually replaced. The generation method will be described in detail with reference to FIG. 9 and FIG. In order to avoid confusion, hereinafter, a character string to which no information is added is referred to as “name after conversion”, and a character string after information is added to “attribute information addition expression”.

  Proper nouns are broadly divided into organizations and individuals. The table for acquiring information is different among the tables shown in FIGS. 3A to 3D between the organization and the individual. Therefore, in this embodiment, it is determined whether the proper noun is an organization or an individual, and the attribute information additional expression is generated. Accordingly, the attribute information additional expression generated for the proper noun representing the individual is referred to as “individual name additional expression”, and the attribute information additional expression generated for the proper noun representing the organization is expressed as “organization name additional expression”.

  FIG. 9 is a diagram illustrating a method for generating an organization name additional expression. As shown in FIG. 9, the organization name additional expression is generated by generating a post-conversion name table, an anonymization information table, and an anonymization conversion table. This is the same when the personal name additional expression is generated.

  The post-conversion name table is generated for managing the post-conversion name for each proper noun. The converted name itself is generated according to a predetermined rule. As a result, the post-conversion name is set such that “Company J” is the first organization name that appears in the document, and “K Company” is the next organization name.

  An anonymization table is produced | generated in order to manage the information for organization name additional expression production | generation for every proper noun showing an organization name. In the table, items of proper noun, converted name, relationship, attribute 1 and attribute 2 are provided. Information stored in the record extracted from the company information table is stored in each item of relationship, attribute 1, and attribute 2. Each item of attribute 1 and attribute 2 stores information stored in the business type item in the company information table. This is because the business category item is an item in which attribute information is stored for the company. The item of industry can store a plurality of attribute information.

  The anonymization conversion table is generated for managing the organization name additional expression (attribute information additional expression) for each proper noun. The organization name additional expression is simply generated according to a fixed conversion rule. According to the conversion rule, as shown in FIG. 9, as shown in FIG. 9, as shown in FIG. 9, related item information (here, “customer company” is equivalent) is added before the converted name, and attribute 1 item is added after the converted name. And / or it is generated by adding a character string in which information of attribute 2 items is described in parentheses.

  The organization name to be replaced with the attribute information additional expression may not be the company name. It may be a part of a company (organization). Therefore, in FIG. 9, the record which described "information system part" in the proper noun item is shown in the anonymization information table. Such a record can be generated, for example, by preparing a table in which information on each department constituting the organization is collected for each organization. To avoid confusion, we will not mention it here any further.

FIG. 10 is a diagram illustrating a method for generating a personal name additional expression. Here, description will be given focusing only on the difference from the case of generating the organization name additional expression.
In the search using the personal name as a key, a record storing the personal name is extracted from either the corresponding organizational structure information table or the corresponding customer organizational structure information table. A person with that personal name belongs to any organization. Therefore, as shown in FIG. 10, the anonymization table uses a record extracted from any one of the corresponding organizational structure information table and the corresponding customer organizational structure information table, and a record extracted from the company information table. Generated. In the items provided in the anonymization information table, the record of the record extracted from the company information table is stored in the related item.

  The personal name additional expression is also performed according to a fixed conversion rule, like the organization name additional expression. According to the conversion rule, the personal name addition expression adds the role item information (here, “promanet” and “application reader” are equivalent) before the converted name, as shown in FIG. Is added by adding a character string in which information on related items and / or belonging items is written in parentheses. In order to avoid becoming too redundant, information on related items is not added to individuals belonging to the company. As a result, the personal name additional expression (attribute information additional expression) whose proper noun (original expression) is “Sato” is “Promanet A”.

  Although not shown in particular, the converted name of the personal name that does not exist in any of the corresponding organizational structure information table, the corresponding customer organizational structure information table, and the in-house employee information table is an attribute information additional expression.

  As described above, the additional expression generation unit 23b determines whether the proper noun is an organization or an individual, and generates an attribute information additional expression for each proper noun according to the determination result. The generation result is output to the character string replacement unit 25. Thereby, the character string replacement unit 25 anonymizes the proper noun existing in the material input from the material acquisition unit 21 by replacing it with the specified attribute information additional expression from the proper noun conversion unit 23.

  The anonymization converts the document shown in FIG. 1A into the one shown in FIG. The document shown in FIG. 4 expresses the meaning of the original document more clearly than the document shown in FIG. 1B. For this reason, even if the document (material) is anonymized, the meaning of the original document can be accurately conveyed to the reader.

  Acquisition of the material and anonymization of the material are performed by a user operation on the input unit 26. The anonymized material can be displayed on the display device DP by the display control unit 27 in accordance with a user instruction via the input unit 26.

  FIG. 16 is a diagram illustrating an example of a hardware configuration of a computer capable of realizing the document anonymization apparatus 2. Here, with reference to FIG. 16, a configuration of a computer capable of realizing the document anonymization apparatus 2 will be specifically described. In order to avoid confusion, the document anonymization apparatus 2 will be described on the assumption that it is realized by a single computer having the configuration shown in FIG.

  The computer shown in FIG. 16 includes a CPU 61, a memory 62, an input device 63, an output device 64, an external storage device 65, a medium drive device 66, and a network connection device 67, which are connected to each other via a bus 68. It has become. The configuration shown in the figure is an example, and the present invention is not limited to this.

The CPU 61 controls the entire computer.
The memory 62 is a memory such as a RAM that temporarily stores a program or data stored in the external storage device 65 (or the portable recording medium 69) during program execution, data update, or the like. The CPU 61 performs overall control by reading the program into the memory 62 and executing it.

  The input device 63 includes, for example, an interface connected to an input device such as a keyboard and a mouse, or all of them. A user operation on the input device is detected, and the detection result is notified to the CPU 61.

  The output device 64 is, for example, a display control device connected to the display device DP of FIG. The data sent under the control of the CPU 61 is output to the display device DP of FIG.

  The network connection device 67 is for communicating with an external device via a network such as an intranet or the Internet. The external storage device 65 is, for example, a hard disk device. Mainly used for storing various data and programs.

The storage medium driving device 66 accesses a portable recording medium MD such as an optical disk or a magneto-optical disk.
The document anonymization apparatus 2 according to the present embodiment is realized by the CPU 61 executing a program (hereinafter referred to as “document anonymization software”) equipped with functions necessary for the document anonymization apparatus 2. The anonymization software may be recorded and distributed on the recording medium MD, or may be acquired by the network connection device 67. Here, it is assumed that the data is stored on the external storage device 65.

  The acquisition of the material is performed via the medium driving device 66 or the network control device 67 under the control of the CPU 61 and stored in the memory 62 or the external storage device 65. 2 is realized by the CPU 61, the memory 62, the external storage device 65, the medium driving device 66, the network control device 67, and the bus 68. If the anonymized material can be stored in the recording medium MD or output via the network control device 67, the proper noun extraction unit 22, the proper noun conversion unit 23, and the character string replacement unit 25 are also combined by the same constituent elements. Realized. The information storage unit 24 corresponds to, for example, the external storage device 65 or the medium driving device 66 on which the recording medium MD is mounted. The input unit 26 is realized by, for example, the CPU 61, the memory 62, the input device 63, the external storage device 65, and the bus 68. The display control unit 27 is realized by, for example, the CPU 61, the memory 62, the output device 64, the external storage device 65, and the bus 68.

  Generation of attribute addition expressions and replacement of proper nouns as described above are realized by the CPU 61 executing the following processing. Next, processing executed by the CPU 61 will be described in detail with reference to FIGS. Each process is realized by the CPU 61 executing the document anonymization software.

  FIG. 5 is a flowchart of the document anonymization process. First, the document anonymization process will be described in detail with reference to FIG. The document anonymization process is started when the user designates a document via the input unit 26. Although a plurality of materials can be specified, here, in order to avoid confusion, the description will be made on the assumption that only one material is specified.

  First, in step S1, proper nouns to be anonymized are extracted from the material (document). The proper noun is extracted by performing morphological analysis and referring to the tables shown in FIGS. 3A to 3D. In subsequent step S2, for each extracted proper noun, an additional expression generation process for generating an attribute information additional expression to be replaced is executed. After the execution, the process proceeds to step S3, where the anonymized material is generated by replacing the proper noun in the material with the generated attribute information additional expression, and stored in the external storage device 65, for example. After the storage, the document anonymization process is terminated.

  FIG. 6 is a flowchart of the additional expression generation process executed as step S2. Next, the generation process will be described in detail with reference to FIG. The result of the proper noun extraction is passed from the document anonymization process to the generation process.

  First, in step S11, one of the extracted proper nouns is selected, and it is determined whether the proper noun represents an individual or an organization. The determination is performed by specifying, for example, each table in FIGS. 3A to 3D having a record storing the proper noun. Thereby, when the record having the proper noun exists in any of the corresponding organizational structure information table, the corresponding customer organizational structure information table, and the in-house employee information table, it is determined that the proper noun represents an individual, The process proceeds to step S13. If not, that is, if a record having a proper noun exists in the company information table, it is determined that the proper noun represents an organization, and the process proceeds to step S12.

  In step S12, an organization name additional expression generation process for generating an organization name additional expression which is an attribute information additional expression to be replaced with a proper noun is executed. In step S13, personal name additional expression generation processing is executed to generate a personal name additional expression that is an attribute information additional expression to be replaced with a proper noun. After executing one of the generation processes, the process proceeds to step S14 to determine whether all proper nouns have been processed, that is, whether all attribute information additional expressions to be replaced with proper nouns have been generated. If there remains a proper noun for which an attribute information additional expression is to be generated, the determination is no, the process returns to step S11, and a determination process is performed by selecting one of the remaining proper nouns. Otherwise, the determination is yes, and the additional expression generation process ends here.

  As described in detail with reference to FIG. 9 and FIG. 10, in this embodiment, anonymization information tables having different configurations are generated according to whether the proper noun represents an organization or an individual. It has become. For this purpose, determination processing is performed in step S11, and different processing is executed according to the determination result. Hereinafter, each additional expression generation process executed according to the determination result will be described in detail.

  FIG. 7 is a flowchart of the organization name additional expression generation process executed as step S12. First, the additional expression generation processing will be described in detail with reference to FIG. Each time this generation process is executed, one record is added to the post-conversion name table, the anonymization information table, and the anonymization conversion table shown in FIG. The generation of these tables is not particularly illustrated, but is performed, for example, when the generation process is first executed.

  First, in step S21, a record storing a target proper noun (organization name) is extracted from the company information table (FIG. 3C). In the subsequent step S22, the post-conversion name to be replaced with the proper noun is set, one record is added to the post-conversion name table, and one record storing the information for generating the attribute information additional expression of the proper noun is anonymous. Add to the conversion information table. In step S23, the attribute information additional expression is generated using the information stored in the one record added to the anonymization information table, and the additional record and the proper noun (original expression) are stored. Is added to the anonymization conversion table. After the addition, the organization name additional expression generation process is terminated.

  FIG. 8 is a flowchart of the personal name additional expression generation process executed as step S13. Next, the additional expression generation processing will be described in detail with reference to FIG. Each time this generation process is executed, one record is added to the post-conversion name table, the anonymization information table, and the anonymization conversion table shown in FIG. The generation of these tables is not particularly illustrated, but is performed, for example, when the generation process is first executed.

  First, in step S31, a record storing a target proper noun (personal name) is selected from the corresponding organizational structure information table (FIG. 3A), the corresponding customer organizational structure information table, and the in-house employee information table (FIG. 3D). Extract from either. In the subsequent step S32, a record storing company information specified from the table from which the record is extracted is extracted from the company information table (FIG. 3C). After the extraction, the process proceeds to step S33.

  In step S33, a post-conversion name to be replaced with the proper noun (individual) is set, one record is added to the post-conversion name table, and one record storing information for generating attribute information additional expressions of the proper noun Is added to the anonymization information table. The record added to the anonymization information table is performed using information stored in the record extracted from the company information table. In step S34, the attribute information additional expression is expressed using information stored in one record added to the anonymized information table depending on whether the individual represented by the proper noun is a member of the own organization. Generate and add one record storing the additional expression and proper noun (original expression) to the anonymization conversion table. After the addition, the personal name additional expression generation process is terminated.

  As shown in FIG. 10, in the attribute information additional expression of members belonging to the own organization, the attribute information is not added in parentheses, unlike the attribute information additional expression of other members. Thereby, it is possible to specify whether or not an individual is a member belonging to his / her own organization depending on whether or not attribute information is added in parentheses.

  In the present embodiment, the generation of attribute addition expressions and the substitution of proper nouns are basically performed as described above. The following two functions are installed so that anonymization of materials can be performed more appropriately for the user. Whether or not both of these functions are used (valid) is selected by operating the input unit 26.

  There are many materials (documents) in which multiple identical proper nouns appear. If the same proper noun is replaced with the same attribute information additional expression in such materials, the anonymized document may feel redundant as shown in FIG. 11A. One of the two functions is to avoid redundant documents due to anonymization.

  By the function, the material to be anonymized as shown in FIG. 11A is anonymized as shown in FIG. 11B. In order to perform anonymization as shown in FIG. 11B, two attribute information additional expressions are generated. One is as described above, and the other is a simple one (simply expressed) without adding attribute information. Accordingly, a plurality of the same proper nouns appearing first are replaced with a normal attribute information additional expression, and those appearing after the second time are replaced with a concise expression (name after conversion). In this way, the redundancy is avoided and anonymization as shown in FIG. 11B is realized. Hereinafter, this function is referred to as “redundancy elimination function”.

FIG. 12 is a diagram for explaining a method for anonymizing proper nouns that can eliminate redundancy. The anonymization method shown in FIG. 12 is for a case where an organization name is anonymized as a proper noun.
As described above, the attribute information additional expression is generated using information stored in the anonymization information table and stored in the anonymization conversion table (FIGS. 9 and 10). When the redundancy avoidance function is used, as shown in FIG. 12, a “concise expression” item and a “counter” item are added to the anonymization conversion table. The concise expression item is an item that stores a concise expression, and the counter item is an item that stores a count result of the number of appearances of proper nouns. “0” shown in FIG. 12 is an initial value. For this reason, if the value (count value) is not 0, it means that a plurality of proper nouns appear. Thus, either the attribute information additional expression or the concise expression is selected according to whether the count value is 0, and the proper noun is replaced with the selected expression. This is the same when the proper noun is an individual name.

  The addition of the record to the anonymization conversion table as shown in FIG. 12 is performed in step S23 of FIG. 7 if the proper noun is an organization name, and in step S34 of FIG. 8 if the proper noun is a personal name. . The replacement with the proper noun is realized by executing the proper noun replacement process shown in FIG. 13 in step S3 of FIG. Here, the replacement process will be described in detail with reference to the flowchart shown in FIG. The replacement of proper nouns in this replacement processing is performed by sequentially processing words determined as proper nouns by morphological analysis in step S1 of FIG. 5 from the top of the material.

  First, in step S41, it is determined whether or not the proper noun to be processed exists as an original expression in the anonymization conversion table. If the record storing the proper noun as the original expression can be extracted from the anonymization conversion table, the determination is yes and the process proceeds to step S43. Otherwise, the determination is no, the process target is changed to the next proper noun (word) in step S42, and then the determination in step S41 is performed again.

  In step S43, it is determined whether or not the value of the counter item in the extracted record is zero. If the value is 0, the determination is YES, the proper noun (original character string) is replaced with the attribute information additional expression of the record in step S44, and the counter item value is further incremented in step S45. Control goes to step S47. Otherwise, the determination is no, and after replacing the proper noun (original character string) with the concise expression of the record in step S46, the process proceeds to step S47.

  In step S47, it is determined whether or not the replaced proper noun is located at the end of the material (sentence). If there are no proper nouns to be processed, the determination is yes, and the proper noun replacement process ends here. Otherwise, the determination is no and the process moves to step S42.

Next, the remaining functions will be described in detail.
As described above, the attribute information additional expression is generated by adding one or more attribute information to the post-conversion name. Because of such generation, it is possible to select the number and type of attribute information to be added. Another function focuses on this and allows the user to select the attribute information additional expression to be generated. Henceforth, the function will be referred to as “expression selection function”.

  14A and 14B are diagrams illustrating a method of selecting an attribute information additional expression to be generated. FIG. 14A shows the case where the proper noun is an organization name, and FIG. 14B shows the case where the proper noun is an individual name.

  As shown in FIG. 14A and FIG. 14B, in this embodiment, a display device is provided with a slider for allowing the user to select the amount of information (detail level) possessed by the attribute information additional expression, that is, the number of attribute information used for the generation. It is displayed on the DP, and the user can operate the input unit 26 to move the position of the slider knob to select the information amount of the attribute information additional expression to be generated. The anonymization conversion table generated at the maximum information amount is shown on the left side of the slider, and the anonymization conversion table generated at the minimum information amount is shown on the right side. As is clear from these anonymization conversion tables, the amount of information is changed depending on the position of the knob by increasing or decreasing the number of attribute information to be written in parentheses.

  The target to be selectable is not limited to the amount of information, and the attribute information itself used for generating the attribute information additional expression may be selectable. As a result, if it is an organization, it may be possible to select a location, a business type, or a scale, and if it is an individual, it may be possible to select an affiliation, gender, job title, or the like.

  In the present embodiment, the generation content of the attribute information additional expression can be selected for each type of organization name and personal name. This is because detailed information about the organization is often not very important for people within the organization, and detailed information about the individual is often not important for people outside the organization. That is, information required for each attribute information additional expression of the personal name and the organization name is often different depending on the person browsing the material.

  As described above, the attribute information additional expression is generated using information stored in the anonymization information tables shown in FIGS. Since the generation content of the additional expression can be changed, the actually generated additional expression is displayed to allow the user to confirm. The confirmation is realized by executing an additional expression selection process shown in the flowchart of FIG. Next, the selection process will be described in detail with reference to FIG.

  The selection process is executed before the additional expression generation process is ended, for example, when the determination in step S14 of FIG. 6 is YES. Thereby, the attribute information additional expression (anonymization conversion table) is generated according to a predetermined amount of information (detail level), and is generated again by changing the detail level by the user. There are two types of anonymization conversion tables, one for organization and one for personal use, and the degree of detail can be changed. However, for the sake of convenience, description will be made on the assumption that there is one anonymization conversion table.

  First, in step S51, the generated anonymization conversion table is displayed, and the sliders shown in FIGS. 14A and 14B are displayed, and the degree of detail is changed according to the user's operation on the input unit 26. The process proceeds to step S52 when the user changes the degree of detail. When the user does not change the level of detail, that is, when the user instructs anonymization in the last generated anonymization conversion table, the additional expression selection process is not shown here. Exit.

  In step S52, a rule for generating an attribute information additional expression (indicated as “attribute information additional expression conversion rule” in the figure) is selected according to the degree of detail after the change. Here, the rule is a rule for selecting what is actually used from among the attribute information that can be used to generate the attribute information additional expression. In the subsequent step S53, an anonymization conversion table is generated (created) from the anonymization information table using the selected rule, and the conversion table is displayed in the next step S54.

  In step S55 to which the process proceeds after displaying the conversion table, it is determined whether or not the input has been completed, that is, whether or not the user has performed an operation for instructing anonymization in the conversion table. If the user performs the operation, the determination is yes, and the additional expression selection process ends here. Otherwise, the determination is no and the process returns to step S51. Thereby, it corresponds to the change of the degree of detail when the user operates the slider.

  When the additional expression selection function is enabled, the additional expression selection process as described above is executed. The additional expression selection function can be made effective together with the redundancy elimination function. When both the redundancy elimination functions are valid, the anonymization conversion table as shown in FIG. 12 is generated.

  In this embodiment, in the attribute information addition expression, one attribute information is directly added to the name after conversion, and the other attribute information is added in parentheses. It is not limited to such a thing. The addition method may be arbitrarily changed. Proper nouns are broadly divided into two types: organizations and individuals. For example, organizations are divided into more types, focusing on the company, the departments that make up the company, the individual's presence or absence, and the status of each individual. The method of addition may be determined.

Claims (8)

In a computer that can be used as a document anonymization device that anonymizes proper nouns in a document,
A proper noun extraction function for extracting proper nouns in the document;
An information acquisition function for acquiring at least one attribute information related to the proper noun extracted by the proper noun extraction function;
When the attribute information is acquired by the information acquisition function, using the attribute information, a character string generation function for generating a character string to be replaced with the proper noun extracted by the proper noun extraction function;
A character string replacement function for replacing the proper noun extracted from the material by the proper noun extraction function with the character string generated by the character string generation function,
A program to realize The character string generation function generates an expression in which the proper noun is anonymized for each proper noun that has acquired the attribute information by the information acquisition function, and adds the attribute information to the generated expression to add the character information. Generate columns,
The program according to claim 1. The character string generation function generates a simple expression that is a character string with a small amount of information compared to the character string for each proper noun from which the information acquisition function has acquired the attribute information,
The character string replacement function replaces the first proper noun that appears in the document with the character string when the same proper noun exists in the document, and the second proper proper noun is Replace with the concise expression,
The program according to claim 1. A selection function for selecting the character string to be generated by the character string generation function;
The character string generation function generates the character string using the attribute information acquired by the information acquisition function in response to a user's selection instruction via the selection function.
The program according to claim 1. In a document anonymization device that anonymizes proper nouns in a document,
Proper noun extraction means for extracting proper nouns in the document;
Information acquisition means for acquiring at least one attribute information related to the proper noun extracted by the proper noun extraction means;
When the information acquisition unit acquires the attribute information, using the attribute information, a character string generation unit that generates a character string to be replaced with the proper noun extracted by the proper noun extraction unit;
Character string replacement means for replacing the proper noun extracted from the material by the proper noun extraction means with the character string generated by the character string generation means,
A document anonymization device comprising: The character string generation unit generates, for each proper noun for which the information acquisition unit has acquired the attribute information, a concise expression that is a character string with a small amount of information compared to the character string,
The character string replacement means replaces the proper noun that appears first in the document with the character string when there are a plurality of the same proper nouns in the document, and the proper noun that appears second or later is Replace with the concise expression,
The document anonymization device according to claim 5, wherein: A selecting means for selecting the character string to be generated by the character string generating means;
The character string generation means generates the character string using the attribute information acquired by the information acquisition means in response to a user's selection instruction via the selection means.
The document anonymization device according to claim 5, wherein: In a document anonymization method for anonymizing a proper noun in a document by a computer,
Extracting proper nouns in the document;
For each extracted proper noun, obtain at least one attribute information related to the proper noun,
A character string to be replaced with the proper noun obtained from the attribute information is generated by adding the attribute information to an anonymized expression of the proper noun,
Replacing the generated string with the proper noun,
A document anonymization method characterized by the above.

Images