JP2006309406A - Information concealing device, concealing system, concealing method and concealing program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an information concealing device which estimates information to be concealed by referring to information separately acquired and performs concealing processes based on the estimation. <P>SOLUTION: The information concealing device 10 for concealing text information including secret information, such as a character string has a first kind concealing required word indication part 12 which indicates first kind of information (name of a person, name of a company, etc.) as concealing required information, and a determination processing part 26 which determines efficiency of the concealment by the concealing required word indicated in the first kind concealing required indication part 12. The determination processing part 26 inspects external information by an inspection formula using inspection words extracted from the text information, and determines the efficiency of the concealment from the inspection result. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a technique for concealing text information including confidential information represented by a daily report that records business activities and the like.

  In order to share information collected in business activities and to manage and support activities in the field, it has traditionally been possible to keep activity records. For example, in sales activities, daily activities are recorded in the form of daily reports, etc., using information obtained from the sales staff in the course of business. With the advancement of information technology, activity records have begun to attract attention as a source for extracting useful information for corporate activities. For example, using information such as text mining to derive useful information from activity records It has been tried to be used for business analysis and to improve sales activities. In order to utilize these activity records, not only the organization that holds the activity records is independently carried out, but also a specialized analysis service of another person or a third party is used.

  However, there is a lot of information that should be kept confidential in the information collected for business purposes, and it is necessary to avoid the possibility of leakage or leakage of information that should be kept confidential, including accidents and virus damage. Has been. For this purpose, it is necessary to be careful in disclosing information to others, and when disclosing information to the outside, some information has been concealed in advance.

  For example, in Patent Document 1, the user's identification information is anonymized when information obtained from the service history on the Internet can be used by others for marketing research. In this type of technology, the information to be concealed before being provided to others has been concealed by a method of changing to another symbol / character string.

  Patent Document 2 discloses an authentication or access management system based on the reliability and disclosure level of personal information, which makes it possible to conceal personal information and perform access management based on reliable personal information. As a result, anonymous posting is possible, and a reliable handle name can be used. For example, access permission only for women and access permission with age restrictions are possible.

  Patent Document 3 discloses an evaluation system that can evaluate a web site while preventing leakage of user information. This can be done by converting the part that can identify the user in the access log to another character string and providing it to the analysis side, or so that the information related to the part necessary for web analysis can be maintained. It is described that a part that needs to be understood in the analysis to be accessed by the same user is converted to the same identifier.

  Patent Document 4 relates to a technique for constructing a data evaluation system used by a rating evaluation organization or the like. When rating is performed by an external organization, the contents of debts that are not disclosed to the public are often used, and the know-how of evaluation is also a trade secret. Disclosure of technology that realizes evaluation without sharing these with each other Yes.

Japanese Patent Application Laid-Open No. 2004-234381 JP 2002-132730 A Japanese Patent Laid-Open No. 2002-123452 JP 2001-052051 A

  However, the conventional information concealment process has the following problems. In business daily reports and the like in which a variety of information is recorded as compared to service histories on the Internet, it is not sufficient to conceal personal information and the like even if the above-described conventional technology is used. Since various information is described in the activity record, there is a case where information that should be concealed may be revealed by referring to other information. For example, in the fictitious business daily report shown in FIG. 8, even if the concealment process is performed on the character string representing the person “Wakayo Manager” and converted into the text information as shown in FIG. If an article as shown in FIG. 15A is on the top, it is easily determined that “AAA” in FIG. 10 is “Wakayo”. In addition, for example, individuals may be identified from the conference participant list and company name, or individual names may be identified by combining university graduation list, age, company name, etc. is there. In particular, outflow of directory information is well known, and it is necessary to sufficiently conceal individuals so that individuals are not identified even if information of that type is introduced.

  The present invention has been made in view of the above points, and by examining the possibility that information to be concealed is estimated by referring to other information that is separately available, and by concealment processing based on the inspection, An information concealment device, an information concealment system, an information concealment method, and a method for concealing business-related information including text descriptions such as business daily reports to be disclosed to others for services such as analysis and diagnosis, and The purpose is to provide an information concealment program.

  An information concealment device for concealing information including secret information such as a character string according to the present invention is based on instruction means for instructing first type information as information to be concealed, and the first type information. The first selection means for selecting the confidential word from the information to be concealed and the concealment word selected by the first selection means sufficiently conceal the information to be concealed. Determination means for determining whether or not. As a result, it is possible to prevent information that is an object of concealment from being easily specified.

  Preferably, the determination unit includes: a test word extraction unit that extracts at least one test word from the text information to be concealed; and a test formula configuration unit that configures a test formula based on the extracted test word. And an inspection execution means for inspecting external information using the inspection formula, and the determination means determines whether concealment is sufficient based on the inspection result of the inspection execution means. A more accurate determination can be made by referring to external information and examining whether there is a term that is likely to lead to deciphering of the information to be concealed.

  Preferably, the information concealment device further includes a second selection unit that selects the second type of concealment word from the text information to be concealed when the determination unit determines that the concealment is not sufficient. And the determining means determines whether or not the information to be concealed is sufficiently concealed by the anonymized word selected by the second selecting means. As a result, it is possible to prevent information that is an object of concealment from being easily specified.

  For example, the test word extraction unit extracts a feature word included in text information that is a concealment target according to the appearance frequency. The appearance frequency may include not only the number of appearances but also the bias frequency of appearance of the feature words that appear. The inspection word extraction means extracts a caution word included in the text information to be concealed according to a predetermined rule. Further, the inspection formula constructing means sets a test formula candidate from the plurality of extracted test words, causes the test execution means to perform a preliminary test on the set test formula candidate, and determines the test formula based on the preliminary test result. You may make it select. The inspection formula constructing means may set a test formula candidate based on the external inquiry result. Further, the inspection execution means executes an inspection on the information accumulated in the database for the inspection formula candidates. The database may be an external database or an internal database. The inspection execution means may execute an inspection on information included in the information disclosure service for the inspection formula candidate. The information disclosure service includes information published on the web.

  Furthermore, the information concealment device conceales the concealment word that is selected by the first selection unit and / or the second selection unit when the determination unit determines that the concealment is sufficient. Have means.

  More preferably, the information concealment apparatus includes a prohibition determination unit that determines whether or not the inspection based on the inspection formula should be prohibited. If the information concealment device determines that the inspection is prohibited, the information is collected from the outside. The inspection according to the inspection formula is executed. Thereby, it can prevent that a check type | formula leaks outside and the information which should be concealed is specified.

  The information concealment device described above may be an information concealment system including an information storage medium that stores information to be inspected by the inspection execution unit. In this case, the information storage medium can be accessed by the examination execution means via the communication means.

  An information concealment method for concealing text information including secret information such as a character string according to the present invention selects a concealment word from text information to be concealed based on the first type of information. A first selection step and a determination step of determining whether or not the text information that is the target of concealment is sufficient by the concealment word selected in the first selection step. The information concealment method further includes a concealment step of concealing a concealment-required word when it is determined that the concealment is sufficient.

  An information concealment program for concealing text information including secret information such as a character string according to the present invention selects a concealment word from text information to be concealed based on the first type of information. The first selection step, the determination step for determining whether or not the information to be concealed is sufficient by the concealment word selected in the first selection step, and concealment by the determination step A second selection step of selecting a second type of confidential word from information to be concealed, and a step of concealing the selected confidential word when it is determined that the conversion is not sufficient Have

  According to the information concealment device according to the present invention, it is determined whether the concealment is sufficient before concealing the information to be concealed, so that the concealed information can be easily decoded. Can be prevented in advance. Furthermore, when concealment is not sufficient, more strict and safe information can be concealed by referring to external information and selecting information that can lead to deciphering as a concealment word. .

  The best mode for carrying out the present invention will be described below with reference to the drawings.

  FIG. 1 is a block diagram showing the configuration of the information concealment device according to the first embodiment of the present invention. The information concealment device 10 according to the present embodiment conceals information included in a daily business report that records business activities as an example, and is configured as follows. The first type concealment required word instruction unit 12 holds designation of information to be concealed input from an input unit (not shown) and notifies the concealed word management unit 14 of this. The anonymized word management unit 14 manages a character string and a symbol to be specifically concealed, and notifies the anonymity processing unit 16 of a character string and a symbol to be concealed. The concealment processing 16 performs concealment processing according to a predetermined conversion rule for character strings, symbols, and the like that need to be concealed based on determination results of the concealment management unit 14 and a determination processing unit 26 described later.

  The business data holding unit 18 holds data related to business activities such as daily reports. The inspection word selection unit 20 refers to the information from the confidential word management unit 14 and selects a characteristic word to be inspected, that is, an inspection word, with reference to the business data in the business data holding unit 18. The test formula configuration management unit 22 receives the test word selected by the test word selection unit and communicates with the test execution unit 24 to configure the test formula. The inspection execution unit 24 executes the concealment inspection with reference to external information different from the business data holding unit 18. The determination processing unit 26 determines the degree of concealment of the inspection formula managed by the inspection formula configuration management unit 22 according to a predetermined criterion. The concealment processing unit 16 is activated in response to the result of the determination processing, performs concealment processing of the data held in the business data holding unit 18 with reference to the anonymized word management unit 14, and outputs the concealed data. 28. The second type concealment word selection processing unit 30 issues a second type concealment instruction for newly adding concealment from the inspection formula determined by the determination processing unit 26 that the concealment processing is insufficient. This is selected, and this is notified to the anonymized word management unit 14.

  Next, the flow of processing of the information concealment device according to this embodiment will be described. For the explanation, fictitious daily sales data (FIGS. 8 and 9) when a fictitious sales representative “Ue no Kefuko” visits a fictitious customer company “Cesson Co., Ltd.” is used as an example. In addition, all the proper nouns used in the examples in the examples are fictitious cut out partial character strings of the Iroha song.

  FIG. 2 shows an outline of the information concealment process. First, the first type concealed word instruction unit 12 notifies the concealed word management unit 14 of information on the character string and the symbol to be concealed of the first type (step S101). For example, the “person name” and the “customer company name” are stored as information to be concealed in the first type confidential word instruction unit 12, and the first type essential information is stored in the confidential word management unit 14. “Personal name” and “customer company name” in the daily report data are notified as instructions for the concealment word.

  When the confidential word management unit 14 receives an instruction for the first type of confidential word, the confidential word management unit 14 refers to the business data holding unit 18 to extract a specific character string or symbol to be concealed. Setting is performed (step S102), and this is stored in the internal recording means as a concealment word. This processing can be realized by using a well-known named expression extraction technique from text information, or from daily report data attribute values such as the name of a customer company held in the business data holding unit 18 and the name of a person to be visited. It can also be extracted. In this step, the personal name “Wakayo” is selected in FIG. 8 and FIG. 9, and the visited company name “Cessin” is set as a concealment word from the daily attribute data and stored.

  Next, a test word is extracted by the test word selection unit 20. For the sake of simplicity, the outline of the processing will be described using a known feature word extraction technique. Words are cut out from the text information of each daily report data (FIGS. 8 and 9), and free words such as nouns and sub-variants and unknown words are extracted. For each extracted expression, a ranking is performed by calculating a numerical value that is weighted according to the number of daily reports in which the expression appears in the entire daily report. Therefore, an expression (feature word) to be selected as a test word is extracted (step S103). The smaller the number of articles in which an expression appears (here, the number of daily reports), the more the expression can be selected.

  In this process, “Wakayo”, “Kashiwahimo-shi”, “Kendo”, “Champion”, “Chillin” in FIG. 8, and “Wakayo”, “Chillin”, “Eco-Iroha” in FIG. 9, respectively. The expression excluding “Wakayo”, which is a concealment word, is extracted as a check word by referring to the concealment word management unit 14.

  When a test word is extracted by the test word selection unit 20, the test formula configuration management unit 22 configures a test formula for each unit set by a predetermined method (step S104). FIG. 3 shows the flow of processing for constructing this check expression. First, a test formula candidate is initialized from the extracted test words (step S201). For example, four test words are selected for the data shown in FIG. 8, so select each of the four test words, and select [Kashihimoichi], [Kendo], [Winner], and [Chillin]. It is considered as an inspection formula candidate.

  Next, a preliminary inspection is performed by the inspection execution unit 24 for the initially set inspection formula candidates (step S202). The inspection control unit 24a performs preliminary inspection using inspection formula candidates on the public information, and collects this. For example, an inquiry process such as searching for external information is performed using the inspection formula candidates. Here, for the sake of simplicity of explanation, information on the web will be taken as external information, and an explanation will be given by taking as an example the realization of the information on the web by an inquiry to a search engine.

  Taking the test formula candidate “Kashihimo-shi” as an example, the test execution unit 24 makes an inquiry to an external search engine using the search formula “Kashihimo-shi”, and the number of information corresponding to the inquiry (hereinafter, hits) Number). The inspection formula configuration management unit 22 inspects the number of hits by a predetermined preliminary inspection method, and selects an inspection formula from the inspection formula candidates. (Step S203). In this preliminary inspection method, for example, the number of hits in a search is compared with a certain numerical value, and an inspection formula can be selected when the number is smaller than the numerical value. Here, when the number of hits is smaller than 1000, the inspection formula is selected. FIG. 4 shows the result of an imaginary preliminary inspection. According to the above selection criteria, no inspection formula is selected at this stage.

  Next, the test formula configuration management unit 22 determines whether a test formula candidate can be added (step S204), and prepares a new test formula candidate by combining test words using a predetermined method. For example, by adding a test word to a test formula candidate not selected as a test formula, a new test formula candidate is prepared as follows.

  Inspection formula that combines the test words of [Kashihimoichi and Kendo], [Sagahimoichi and Winner], [Kashihimoichi and Cirinu], [Kendo and Winner], [Kendo and Cirinu], and [Winner and Cirinu] Candidates are set (step S201).

  Then, the inspection formula configuration management unit 22 similarly requests a preliminary inspection from the inspection execution unit 24 (step S202), and obtains the number of hits as the inspection result. FIG. 5 shows a fictitious inspection result by this inspection formula candidate. At this stage, the following four test formula candidates having a hit count of 1,000 or less are selected and stored in the test formula configuration management unit 22 as test formulas. The inspection execution unit 24 receives notification of the inspection formula selected from the inspection formula configuration management unit 22, collects data hit by the search, and stores it in the internal storage means. Four inspection formulas are selected, [Shihimo-shi and Kendo], [Shihimo-shi and winner], [Shihimo-shi and Chirinu] and [Kendo and Chirinu] (step S203).

  Next, the inspection formula configuration management unit 22 prepares the next combination of search formula candidates. For this, a combination of test words is created based on [Kendo and Winner] and [Winner and Chillin] that were not selected as a search expression. In this case, [Kendo and Winner and Chillin] can be generated, but since “Kendo and Chillin” already selected as the search expression is included, it is not added as a search expression candidate (step S204). . At this stage, since no additional search formula candidates are generated, the check formula configuration for the daily report in FIG. 8 ends (step S205).

  The same processing is executed for the check word extracted from FIG. 9, and a check expression is selected. That is, in the data of FIG. 9, since two inspection words “Cilenu” and “Eco-Iroha” are selected, these preliminary inspections are executed. However, since the preliminary inspection has already been performed for “Chillin”, the preliminary inspection is omitted, and the preliminary inspection is performed for [Eco-Iroha]. FIG. 6 shows the result of the preliminary inspection of [Eco-Iroha].

  Since the number of search results is 1,000 or less, “Eco-Iroha” is selected as the inspection formula. For the search result of “Eco-Iroha”, the search execution unit 24 collects and stores the hit information. As described above, if [Eco-Iroha] is not selected as the inspection formula, [Eco-Iroha and Chillin] will be the next test formula candidate.

  Returning to FIG. 2 again, the inspection formula configuration management unit 22 takes out the concealed information “Wakayo” from the anonymized word management unit 14 and adds it to the inspection formula (for example, [Kashihimoichi and Kendo and Waka Yo]), and instructs the inspection execution unit 24 to execute the inspection process using the collected information (step S105). In this example, the inspection process is realized by performing a search process and obtaining the number of hits. The processing result is, for example, as shown in FIG.

  The determination processing unit 26 determines whether or not the concealment process is sufficient for the result of the inspection process (step S106). The determination processing unit 26 can determine that the concealment process is not sufficient when, for example, the inspection result to which the concealment information is required is within a predetermined numerical range. For example, the predetermined range is set to 0 or more and less than 20, and it is determined that the concealment is not sufficient when it is within this range. When the confidential information is “Wakayo”, there are three search results for [Kashihimo-shi and Kendo] and [Tatsumi-himo-shi and Winner] with the confidential information added. If not, select these two check expressions. This examination examines the risk that the evidence will be easily specified when the hypothesis is that the information that is trying to reveal the concealment might be information about "Wakayo".

  Further, the ratio of the inspection result to which the concealment word of FIG. 7 is added to the preliminary inspection result of the inspection formula candidate can be used as a criterion. For example, if the ratio exceeds 60% and it is selected that the concealment is not sufficient, the search result when the concealment information “Cesun” is added for the inspection formula [Eco-Iroha] is 80. Therefore, since the ratio is 80/96 to 83%, the inspection formula [Eco-Iroha] is selected as being insufficiently concealed. In this inspection, the risk that is estimated about what is applicable to the inspection formula candidate is examined.

  Fig. 15 (a) is an example of a fictitious article in which the name of "Wakayo Taleso" is listed as the winner of the citizen Kendo tournament on December 23rd. Similarly, Fig. 15 (b) This is an example of a fictitious article that describes "Eco-Iroha", which is set up by Cesson.

  As described above, the determination processing unit 26 determines that the inspection formulas of [Shihimoichi and Kendo], [Sasahimoichi and Winner] and [Eco-Iroha] are not sufficiently concealed, (Step S107), and then supplies these selected check expressions to the second type concealment word selection unit 30 (Step S108). The second type anonymized word selection unit 30 selects an anonymized word to be newly added from the inspection words included in the selected inspection formula. The new secret words to be selected are either “Kashihimo-shi” or “Kendo” test word, “Kashihimo-shi” or “winner” test word, and “Eco-Iroha”. Therefore, the anonymized word management unit 14 may be instructed to conceal these. At this time, a characteristic word may be selected by referring to the numerical value calculated to extract the characteristic word described above, or a word that affects a plurality of narrowing keys may be selected preferentially (example here) Let ’s talk about “Shimo-shi”). In this example, by adding the two words “Tsubakihimo-shi” and “Eco-Iroha” to the anonymized words, it repeats the process of selecting the test word, constructing the test formula, testing, and judging (step) S102 to S107), concealment can be made safe.

  In the selection of the inspection word for the second time and later, if the word that was not selected at the time of the first inspection word selection is added as the inspection word by changing the criteria for selecting the feature word, the second time It becomes easy to add a new test word even when selecting a test word thereafter. At the time of constructing the test formula, a test formula candidate is constructed by combining a newly selected test word and an already selected test word. Here, it is assumed that the test word selection criteria have not been changed, and no new test word is selected, so that a new test formula candidate or the like is not formed, and the repeated test process is completed.

  When the repeated inspection process is completed, the determination processing unit 26 instructs the concealment processing unit 16 to execute the concealment process (step S109). The anonymization processing unit 16 follows the anonymized word from the anonymized word management unit 14 in the daily report of FIG. 8 and displays “Wakayo” and “Kashihimo-shi”, and in the daily report of FIG. The character string is concealed (step S110). The output unit 28 outputs the information subjected to the concealment process. This output is concealed data as a result of verification of external public information.

  FIG. 12 and FIG. 13 show the processing results of the concealment shown in FIG. 8 and FIG. 9, respectively. FIG. 10 and FIG. 11 show the first type concealment instruction for FIG. It is the figure which showed the result at the time of concealing only the character string instruct | indicated as. FIG. 14 is a log of character string symbol conversion processing when FIGS. 12 and 13 are generated. The log is generated by the anonymization processing unit 16 and transmitted to the anonymized word management unit 14 for storage. The log of the conversion process is used when referring to the original information when receiving the analysis result of the data subjected to the concealment process. Here, for the sake of simplicity of explanation, an example in which one type of character string symbol is converted into one type of character string symbol is shown, but this does not indicate a limitation on the concealment process, and the concealment process A process in which a concealed word and a converted word correspond to each other in a combination of one-to-other, other-to-other, and other-to-one may be used.

  In the above-described processing configuration, a modifier that limits the time and place can also be added by using dependency analysis or the like at the time of the test formula configuration processing. An example of adding a modifier that limits time, place, etc. is “December” in FIG. Furthermore, since “December” is used as a modifier to limit the time, in addition to “December”, calculation for estimating the year is performed (there is a method of obtaining from information other than the text of the daily report) “2004” It is also easy to add the character string “December Year”. As a result, the preliminary inspection can be narrowed down to more easily applicable, and the information collected from the outside in the inspection execution unit can be reduced.

  In addition, a process for predicting a query result to the outside based on a past query history, a result of feature word calculation, or the like can be performed prior to an inspection formula construction process. For example, in the example of daily report data shown in FIG. 8, the expression of “kendo” or “winner” is based on past inquiry results, and even if a preliminary check formula of [kendo] or [winner] is simply configured, However, it is predicted that only a sufficiently large result will be returned, and it is possible to omit a preliminary inspection process for the outside.

  Further, in the above-described processing example, the configuration in which the confidential word is not used at the time of the preliminary inspection is shown. However, it is also possible to perform the processing in which the confidential word is added to other inspection words from the time of selecting the inspection word. . In the example, referring to the anonymized word management unit 14, “Wakayo” and “Cessin” are added to the first test word. In the configuration using the concealment word at the time of the preliminary inspection, it is possible to adopt a processing configuration in which collection of public information is not performed only for the inspection using the inquiry to the outside as well as the preliminary inspection processing.

  On the other hand, in the case of a processing configuration that uses confidential information for external inquiries, external query prohibition determination processing that determines combinations that should not be used for external inquiries should be executed before external inquiries. Is desirable. This external inquiry prohibition determination process can be easily realized by preparing a predetermined determination rule. For example, this is a process to deal with the case where Mr. Ue no Kefuko has not heard the inside information of “Cesun” as confidential information and has just registered it in the daily report. For example, “Wakayo's director seems to be the vice president of Kayota next month.” “Cesson seems to be considering the merger with Kayota.” It is desirable that inquiries such as “Wakayo AND Kayota AND Vice President” and “Cessen AND Kayota AND Merger” should not be made outside before the official announcement. For this purpose, for example, it is possible to set to prohibit the inquiry in the pattern such as {company name} AND {words indicating change of merger}, {person name} AND {specific title}, etc. . For inspection formulas that apply to patterns that prohibit these inquiries, prepare preliminary inspection formulas with relaxed conditions, collect information that was hit in the preliminary inspection from the outside, and then prohibit external inquiries about the collected information. It is also possible to inspect with a different inspection formula.

  In addition, in this embodiment, for the sake of simplicity of explanation, an example is shown in which the composition of the inspection formula is a combination of inspection words in units of daily reports, but this unit is the same for each case or the same visited place, etc. You may comprise by the unit of. By this modification, it is possible to prevent information to be concealed from being specified by combining a plurality of related daily report data. In addition, in this embodiment, for the sake of simplicity, the example in which the preliminary inspection is inquired to the outside is shown on a daily report basis. However, after collecting a large number of preliminary inspections, the order can be inquired at random. It is possible to prevent the contents of each daily report from being specified from the inquiry of the preliminary inspection.

  Next, a second embodiment of the present invention will be described. FIG. 16 is a block diagram showing the configuration of the second embodiment. In the figure, the same reference numerals are assigned to the same components as in the first embodiment. In the information concealment system 10a according to the second embodiment, the test word selection unit 40 extracts words requiring attention that are likely to specify a word to be concealed according to a rule input by an input unit (not shown). The caution required word extraction part 42 is provided. Furthermore, the information concealment system 10a includes an external examination execution management unit 100 that searches for external name list information. The external inspection execution management unit 100 causes the plurality of external name list information inspection execution units 110 to execute inspections in response to the inspection control unit 24a. A broken line P in the figure indicates the boundary between the inside and the outside of the information concealment system 10a.

  FIG. 17 is a diagram illustrating an example of an extraction rule in the word requiring attention extraction unit. FIG. 17A is an example of a caution word extraction rule corresponding to the case where the introduction company name is narrowed down from the description of product delivery / introduction in the text data. In the case of introducing product X, the extraction pattern specifies the pattern in which character strings such as “product X” (something one character), “introduction” or “delivery”, “acceptance”, “order receipt” are connected. ing. When this extraction pattern is applicable, there is a possibility that the company name may be specified. Therefore, it is specified that the product introduction company list of the product X should be inspected. The criteria for this inspection are specified in the corresponding list (in this case, there is a list of companies that have introduced the products of the product Chillin, and the number of elements (number of companies) in the list. The formula configuration management unit 22 and the determination processing unit 26 are referred to and used.

  FIG. 17B is an example of a rule for extracting a word requiring attention regarding past affiliation. The examination target is two types of name list information, and the determination criteria of the selection of the name list of Y University and the number of elements in the list in the name list and the selection of the list of the X part of Y University are described as determination items. This shows an example of changing the designation of the judgment criteria in a case where a large amount of data is included and a case where a relatively small number of elements are included like a university graduation list. And it is described that the specified item specified by the rule of FIG. 17B is a person.

  FIG. 17C is an example of an extraction rule for preventing a person from being identified from article information. Among the types of information recorded as customer special notes in a conversation with a customer, information such as awards that have been awarded in particular may be disclosed by other information sources with their names. In FIG. 17C, the article information of X of Y is set as the inspection target, and the selection of the article is designated as the determination item.

  Instructions regarding the inspection target and the inspection standard can be added to each of the attention word extraction rules. For example, the rule of FIG. 17B specifies that the name list information is to be inspected, and the inspection is performed in a plurality of stages. Whether “Y” in the character string pattern of “Y University” is specified (extracted), or if specified, the inspection control unit 24a checks the external name list information etc. via the external inspection execution management unit 100 Is executed. The external name list information inspection execution unit 110 is arranged to manage the name list of “Y” university, for example, and makes an inquiry using the inspection formula candidates. The list information inspection execution unit 110 does not return specific information in the list information in response to the inquiry, but may return a total value such as only the number of cases that match the inquiry. Furthermore, it is assumed that the answer is in the form of only determination whether the number is greater or less than the designated number, and the number designated at the time of inquiry cannot be smaller than a predetermined numerical value (for example, a value smaller than 5 cannot be designated). In this way, it is possible to carry out an inspection by inquiring while further hiding specific information relating to the list information.

  An example will be described using the fictitious daily report in FIG. Here, the rule of FIG. 17B is applied, and “Y” is “Uno” and “X” is “Kendo” for the pattern “Senior at the X Department of Y University”. The words requiring attention of “Uno” and “Kendo” are added to the test word and processed in the same manner as in FIG. 2, but the test target is specified by the rule of FIG. 17B and added. In addition, inspection of the directory information is also performed. Here, only the inspection process for the directory information will be described.

  The inspection formula configuration management unit 22 inquires of the inspection execution unit 24 whether it is possible to inquire about the name list information of “University Y”, that is, “University of Udon”, according to the instruction in the rule. The examination execution unit 24 checks whether or not there is an external name list information examination execution unit 110 grasped by the external examination execution management unit 100 that handles name information of “University of Uda”. Here, when there is no information handling the name list information of “University of Uda”, the result is notified to the inspection formula configuration management unit 22, and the inspection formula configuration management unit 22 records that fact, and the subsequent processing is determined. Leave it to the part 26. The determination processing unit 26 determines that the information without the list information is preferably insufficiently concealed, and the character string symbol “Uno” is sent to the second type concealment required word instruction unit 12. It will be added.

  If there is a list information inspection execution unit 110 grasped by the external inspection execution management unit 100 that handles the name information of “University of Uda”, the type of information that can be inquired is set as the external inspection execution management unit. 100 and is transmitted to the inspection type configuration management unit 22 through the inspection control unit 24a. Here, “Graduation Year”, “Faculty”, and “(Current) Affiliation” are returned.

  The inspection formula configuration management unit 22 selects the type of information corresponding to “graduation year”, “faculty”, and “(current) affiliation” from the inspection word and the confidential word. A correspondence relationship between the type of information returned from the external inspection execution management unit 100 and the type of information given to the test word can be checked and stored in advance. For this reason, here, “Csen” is selected as a correspondence relationship with “(current) affiliation” based on the stored correspondence relationship. The examination formula configuration management unit 22 selects [Affiliation: Csen] as the examination formula candidate, and instructs the examination execution unit 24 to make an inquiry to the external name list information examination execution unit 110 that handles name list information of “University of Uda”. In the inspection execution unit 24, numerical values obtained by referring to the determination processing unit 26 by a corresponding person whose “affiliation” is “Cessin” in the corresponding list information inspection execution unit 110 through the inspection control unit 24 a and the external inspection execution management unit 100. Check if it is smaller than. For example, if the numerical value obtained by referring to the determination processing unit 26 is 5, when a “small” answer is obtained from the list information inspection execution unit 110 that handles the list information of “Ugano University”, “U” This means that there are fewer than five graduates of “University of Samurai” who belong to “Cessun”. In this case, it means that the possibility that “Wakayo” will be narrowed by referring to the directory information of “University of Udon” if “Cesson”, which should be concealed, is identified. To do. In this example, the confidential word “Cessin” is used as an inspection formula. However, according to a policy predetermined by the determination processing unit 26, the information “Cesson” that should be concealed is displayed. The handling of the results when used in the inspection formula is predetermined.

  In order to conceal more safely, if it is desirable not to identify any other concealed information even if any of the information to be concealed is specified, “Cesson” should be Reflecting the results of use, “Unono” is classified as a second type concealment word. On the other hand, if the response “large” is obtained from the external name list information inspection execution unit 110, the determination processing unit 26 determines that the concealment is sufficient, The string is not added as a concealment candidate.

  Similarly, if there is an external list information inspection execution unit 110 that handles the list information of “Uo no University Kendo Club”, whether or not to add “Uo no” as a candidate for concealment by performing the same processing. Will be performed. There is a high possibility that “small” answers will be obtained relatively in “Uno University Kendo Club” with a small population. If the results from the “Uo University” roster information and the “Uo University Kendo Club” roster information are different, the processing may be decided in a predetermined method. By giving priority to the addition, it is possible to make it more secure and concealed. In addition, if there is no roster information inspection execution section that handles the roster information of “Uo no University Kendo Club”, for example, it is determined that the “Kendo Club” part that is the difference with “Uo University” is to be concealed. I can leave.

  In the inquiry to the list information inspection execution unit 110, selection is also made from the confidential word, but the type of confidential word can be used separately from the type used for external inquiry and the type not used. This classification rule should be saved in advance. For example, the department name, person name, telephone number, and address should not be used for external inquiries, and the company name should be used for external inquiries. Can be kept. As a result, it is possible to easily realize control so as not to make an inadvertent inquiry outside.

  Next, a third embodiment of the present invention will be described. In the information concealment system 10b according to the third embodiment, a name list information inspection execution unit 110 is arranged inside the organization as shown in FIG. As mentioned above, only the investigation of whether the information can be narrowed down without presenting the internal information of the name list information is performed, so the management of each name information holder is managed with the cooperation of the person holding the name list information. The name list information inspection execution unit 120 can also be arranged under this. Since there are people with various types of name information in an organization of a certain size, the name information inspection execution unit 120 is arranged under the right owner of name information on the premise of their cooperation. As a result, it is possible to obtain the same effect as the case of making an inquiry to the outside as shown in FIG. 16 by the inquiry to the inside, and to reduce the information presented to the outside.

  Internal collaborators can cooperate in concealing business information without disclosing individual information on legally owned directory information. Furthermore, it is possible to cooperate more safely by explicitly restricting the name list information inspection execution unit 120 in advance. For example, it does not answer inquiries to information that identifies individuals such as name and address telephone number. It is only a reply whether or not the inquiry content is smaller than the numerical value specified at the time of inquiry, and furthermore, a list that is specified by the operation etc. is arranged so that the specified numerical value can not be less than a predetermined numerical value in advance. Information leakage can be prevented.

  In the above-described embodiments, the explanation was made using pattern matching and simple feature word extraction for character string extraction, but the processing using the results of dependency analysis and semantic analysis of natural language processing is excluded. It is not a thing. In addition, although the explanation is made by taking the combination of the inspection words in one daily report as an example when constructing the inspection formula, it does not exclude that the inspection formula is configured by combining words appearing in a plurality of daily reports. For example, if the relationship between a test word and a secret word is captured using dependency analysis or semantic analysis, or if the structure of an expression is captured using a pattern as shown in FIG. It is also easy to adopt a configuration used for the inspection process.

  Taking FIG. 17C as an example, test word extraction processing is performed using the extraction pattern at the test word extraction stage, and the data of FIG. The character string “win” ”is selected in accordance with the extraction pattern. From this test word (expression), the test formula candidate is used for the question answering system, for example,“ Kendo Tournament ” Who is the “winner” or “person”? The preliminary examination using this examination formula candidate is performed using a well-known question answering system, and as a result of the preliminary examination, the ranked noun (the question answering system determines that it represents a person). A list of nouns) in the check process, does the list contain the secret word “Wakayo”? Check the information on how many are listed, and in the determination process, make a determination according to a predetermined criterion for the result of the inspection, for example, when a confidential word appears within the top 10 ranks, It can also be configured to perform processing such as determining that concealment is not sufficient.

  In the above embodiment, the daily business report has been described as an example. However, data that records other work may be targeted, and there is also little information collected at one time such as questionnaires and interview records. The embodiments can be easily applied by modification alone.

  In addition, there can be a configuration where the inspection execution unit and the inspection control unit are arranged in an external public organization that can be trusted, in that case, it can be expected to be able to inspect more directory information, etc. A safer concealment process can be realized.

  As described above, a more reliable examination can be performed by extracting an expression that is empirically known to have the possibility of specifying a concealment word. In addition, by implementing the inspection of the directory information, for example, in the above example, there are very few graduates of “Uno” university who belong to “Cessin” and there are 5 or less people. In the case where the graduate list of “Uo no University” was leaked, there was a risk that Mr. “Wakayo” would be identified, but the list which is not generally disclosed in the above embodiment A framework for inspecting information was also prepared. As a result, it is possible to verify and re-conceal the concealment in case the directory information of “University of Udon” has leaked, and to create safer concealment information Can do.

  The preferred embodiments of the present invention have been described in detail above. However, the present invention is not limited to the specific embodiments according to the present invention, and various modifications can be made within the scope of the gist of the present invention described in the claims. Deformation / change is possible.

  The information concealment apparatus according to the present invention is used in an apparatus, system, network system, program, etc. for concealing information contained in digitized information.

It is a block diagram which shows the structure of the information concealment apparatus which concerns on 1st Example of this invention. It is a figure which shows the operation | movement flow of an information concealment apparatus. It is a figure which shows the operation | movement flow of the test | inspection type | formula structure of FIG. It is a figure which shows a fictitious preliminary test result. It is a figure which shows the fictitious test result by a test formula candidate. It is the result of preliminary inspection of [Eco-Iroha]. It is a figure which shows a test process result. It is a figure which shows the example of daily report data. It is a figure which shows the example of daily report data. It is a figure which shows the example which concealed the daily report data of FIG. 8 by the 1st type concealment required word. It is a figure which shows the example which concealed the daily report data of FIG. 9 by the 1st type concealment required word. It is a figure which shows the example which concealed the daily report data of FIG. 8 by the 2nd type concealment required word. It is a figure which shows the example which concealed the daily report data of FIG. 9 by the 2nd type concealment required word. It is a figure which shows the log of the conversion process of the character string symbol in the concealment process of FIG. 12 and FIG. It is a figure which shows the example of a fictitious article. It is a block diagram which shows the structure of the information concealment apparatus which concerns on 2nd Example of this invention. It is a figure which shows an example of the rule when extracting a caution word. It is a figure which shows the example of daily report data. It is a block diagram which shows the structure of the information concealment apparatus which concerns on 3rd Example of this invention.

Explanation of symbols

10: Information concealment device 10a, 10b: Information concealment system 12: First type concealed word instruction unit 14: Concealed word management unit 16: Concealment processing unit 18: Business data holding unit 20: Check word selection Unit 22: Inspection type configuration management unit 24: Inspection execution unit 26: Judgment processing unit 28: Output unit 30: Type 2 anonymized word selection unit 100: External inspection execution management unit 110: Name list information inspection execution unit 120: Name list Information inspection execution department

Claims (35)

An information concealment device for concealing information including secret information such as character strings,
Instruction means for instructing the first type of information as information to be concealed;
First selecting means for selecting a confidential word from information to be concealed based on the first type of information;
A determination means for determining whether or not the information to be concealed is sufficiently concealed by the concealment word selected by the first selecting means;
An information concealment device. The determining means includes a test word extracting means for extracting at least one test word from the information to be concealed, a test formula forming means for configuring a test formula based on the extracted test word, and a test formula configured The information concealment device according to claim 2, further comprising: an inspection execution unit that performs an inspection of external information to be used, and the determination unit determines whether the concealment is sufficient based on the inspection result of the inspection execution unit. The information concealment device further includes a second selection unit that selects a second type of concealment word from the information to be concealed when the determination unit determines that the concealment is not sufficient.
3. The information according to claim 1, wherein the determination unit determines whether or not the information to be concealed is sufficiently concealed based on the concealment word selected by the second selection unit. Concealment device. The information concealment device according to claim 3, wherein the second selection unit selects the second type of concealment word based on the inspection result of the inspection execution unit. The information concealment device according to claim 2, wherein the test word extraction unit extracts a feature word included in the text information to be concealed according to the appearance frequency. The information concealment device according to claim 5, wherein the appearance frequency includes a bias frequency of appearance of feature words included in the text information. The information concealment device according to claim 2, wherein the test word extraction unit extracts a caution word included in the text information to be concealed according to a predetermined rule. The inspection formula construction means sets inspection formula candidates from the extracted plurality of test words, causes the inspection execution means to perform preliminary inspection on the set inspection formula candidates, and selects an inspection formula based on the preliminary inspection results The information concealment device according to claim 2. The information concealment device according to claim 8, wherein the check expression constructing means sets a check expression candidate to be a narrow search using a plurality of extracted check words. 9. The information concealment device according to claim 8, wherein the inspection formula construction means selects a test formula candidate whose hit count when the preliminary inspection is executed is included in a certain range. The information concealment device according to claim 8, wherein the inspection formula constructing unit sets a test formula candidate based on an inquiry result to the outside. The information concealment device according to any one of claims 1 to 11, wherein the inspection execution unit performs an inspection on information accumulated in a database for an inspection formula candidate. The information concealment device according to claim 12, wherein the database includes an external database or an internal database. The information concealment device according to any one of claims 1 to 11, wherein the inspection execution unit performs an inspection on information included in the information disclosure service for the inspection formula candidate. The information concealment device according to claim 14, wherein the information disclosure service includes information disclosed on the web. The information concealment device according to any one of claims 1 to 11, wherein the inspection execution unit performs inspection on the name list information disclosed externally for the inspection formula candidate. The information concealment device according to claim 1, wherein the inspection execution unit performs an inspection on the internal name list information for the inspection formula candidate. 2. A concealment unit that conceals a concealment word selected by the first selection unit and / or the second selection unit when the determination unit determines that concealment is sufficient. Or the information concealment device according to any one of 17 to 17; The information concealment apparatus according to claim 1, further comprising inquiry means for inquiring the concealment information to the outside. The information concealment device further includes a prohibition determination unit that determines whether or not the inspection based on the inspection formula should be prohibited. The information concealment device according to claim 1, wherein the information concealment device executes inspection based on a formula. The information concealment device according to any one of claims 1 to 20,
An information concealment system comprising: an information storage medium that stores information to be inspected by the inspection execution means. The information concealment system according to claim 21, wherein the information storage medium is accessible by the examination execution means via the communication means. An information concealment method for concealing text information including secret information such as character strings,
A first selection step of selecting a concealment word from information to be concealed based on the first type of information;
A determination step of determining whether or not the information to be concealed is sufficiently concealed by the anonymized word selected in the first selection step;
An information concealment method comprising: The information concealment method further includes a test word extraction step for extracting at least one test word from information to be concealed, and a test formula configuration step for configuring a test formula based on the extracted test word. The information concealment method according to claim 17, further comprising: an inspection execution step for inspecting external public information using an inspection formula, and determining whether concealment is sufficient based on an inspection result of the inspection execution step. The information concealment method further includes a second selection step of selecting a second type of concealment word from the information to be concealed when the determination step determines that the concealment is not sufficient,
A determination step of determining whether or not the information to be concealed is sufficiently concealed by the anonymized word selected in the second selection step;
An information concealment method comprising: 24. The information concealment method according to claim 23, wherein the inspection word extraction step extracts a feature word included in the text information to be concealed according to the appearance frequency. 24. The information concealment method according to claim 23, wherein the inspection word extraction step extracts words requiring attention from the text information based on a predetermined rule. In the inspection formula construction step, a test formula candidate is set from a plurality of extracted test words, a preliminary test is executed by the test execution means for the set test formula candidate, and a test formula is selected based on the preliminary test result The information concealment method according to claim 23. 24. The information concealment method according to claim 23, wherein the inspection formula construction step sets inspection formula candidates to be a narrow search using a plurality of extracted test words. 24. The information concealment method according to claim 23, wherein the test formula configuration step selects test formula candidates in which the hit count when the preliminary test is executed is included in a certain range. 24. The information concealment method according to claim 23, wherein the inspection execution step performs an inspection on information published on the web for inspection formula candidates. 24. The information concealment method according to claim 23, wherein the inspection execution step performs inspection on name list information that is publicly disclosed for the inspection formula candidate. The information concealment method according to claim 23, wherein the inspection execution step executes an inspection on the internal name list information for the inspection formula candidate. The information concealment method further comprises a concealment step of concealing the concealment-required word selected in the first and / or second selection step when it is determined that the concealment is sufficient. The information concealment method as described in any one. An information concealment program for concealing text information including secret information such as character strings,
A first selection step of selecting a concealment word from information to be concealed based on the first type of information;
A determination step for determining whether or not the information to be concealed is sufficiently concealed by the concealment word selected in the first selection step;
A second selection step of selecting a second type of concealment word from information to be concealed when it is determined that the concealment is not sufficient by the determination step;
Concealing the selected confidential word, and
An information concealment program.

Images