JP2012113606A - Protection object information masking device, protection object information masking method, and protection object information masking program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To estimate a wrong conversion place without referring to source data before masking even if wrong masking processing is performed when an electronic document (text) including protection object information is suitably masked.SOLUTION: A protection object information masking device includes a personal information dictionary DB 10 storing personal information to be protected; personal information detection means 20 for inputting a text having the personal information as a source text, and detecting a character string corresponding to the personal information stored in the personal information dictionary DB 10 as a character string to be masked from the input text; personal information mask generation means 30 for finding a character string length of the character string to be masked detected by the personal information detection means 20, and generating a character string for mask including the character string length; and personal information replacement means 40 for replacing the character string to be masked detected by the personal information detection means 20 with the character string for mask generated by the personal information mask generation means 30.

Description

  The present invention relates to a protected information masking system for detecting personal information for the purpose of protecting personal information and masking the portion of the text including personal information such as first name, last name, and address, and in particular, erroneously identifies a portion that is not personal information. The present invention relates to a personal information mask system that can estimate the state before masking even when the possibility of masking is high.

  As the importance of protecting personal information increases, there is an increasing expectation for a technique that masks the personal information part and makes it unreadable for texts containing personal information such as first and last names and addresses.

  As an existing technique for masking the personal information part, there is a technique for performing character string conversion (Patent Document 1).

JP 2007-102540 A

  However, in the existing technology, even a general name or the like in an electronic document, the wording registered in the personal information dictionary data is subject to masking and may be masked by mistake. For example, “Junko Koshino” as a brand name may be mistaken as a personal name and masked, and the character string of the corresponding brand name part may be encrypted, or “******” It was sometimes replaced with a string of characters.

  In these cases, it is difficult to distinguish between the wrongly masked part and the correctly masked part, so it is difficult to check whether the masking result data is correctly masked. It needs to be referenced.

  However, in the first place, the main purpose of applying masking to document data is to hide personal information portions in order to protect personal information. Therefore, it is desirable to avoid referring to the original data as much as possible. Therefore, there is a need for a masking technique that can estimate a character string that has been masked by mistake without referring to the original data before masking.

  An object of the present invention is to provide a protection target information masking apparatus, method, and program for solving this problem.

  In order to solve the above problems, in the present invention, a protection target information dictionary database storing protection target information and a text having protection target information are input as original text, and the protection target is input from the input text. A protection target information detecting means for detecting a character string corresponding to the protection target information stored in the information dictionary database as a mask target character string, and obtaining a character string length of the mask target character string detected by the protection target information detection means; , A protection target information mask generation unit that generates a mask character string including the character string length, and a mask generated by the protection target information mask generation unit using the mask target character string detected by the protection target information detection unit. And a protection target information replacement means for replacing with a character string.

  According to the present invention, when appropriately masking an electronic document (text) including information to be protected, even if the masking process is erroneously performed, an erroneous conversion location can be determined without referring to the original data before masking. Estimation can be performed.

  Thus, by using the data masked according to the present invention, it is possible to realize extraction of statistical information and the like while realizing protection of information to be protected, for example, personal information.

1 is an overall configuration diagram showing an embodiment of the present invention. The flowchart which shows the process of the personal information mask production | generation means in the example of 1 embodiment of this invention. The flowchart which shows the whole process in the other embodiment of this invention. Explanatory drawing which shows the example of application of the personal information mask process of this invention. Explanatory drawing of the Example which utilized this invention for marketing data extraction. Explanatory drawing of the Example using this invention for confidential document masking.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings, but the present invention is not limited to the following embodiments.

  Note that the exemplary embodiments described with reference to FIGS. 1 to 5 below show exemplary embodiments where the protection target information is personal information.

  FIG. 1 shows the overall configuration of an embodiment of the present invention. In FIG. 1, a personal information masking apparatus 100 inputs a personal information dictionary DB (database) 10 (protection target information dictionary database of the present invention) storing personal information to be protected and an original text such as an electronic document. And personal information detection means 20 (protection target information detection means of the present invention) for detecting a personal information part contained in the personal information dictionary DB 10 from the text as a character string, and for the detected personal information character string, The personal information mask generating means 30 (protection target information mask generating means of the present invention) for generating a character string for masking the personal information portion, and the personal information portion detected by the personal information detecting means 20 The personal information place which outputs the text (electronic document) after the personal information mask in place of the character string generated by the mask generating means 30 E means 40 is provided with a (protected information replacement means of the present invention) and.

  The personal information masking device 100 is configured by a computer, for example, and includes hardware resources of a normal computer such as a ROM, a RAM, a CPU, an input device, an output device, a communication interface, a hard disk, a recording medium, and a driving device thereof.

  As a result of the cooperation between the hardware resource and the software resource (OS, application, etc.), the personal information masking apparatus 100 has a personal information dictionary DB 10, personal information detection means 20, personal information mask generation means as shown in FIG. 30. The personal information replacement means 40 is implemented.

  The personal information mask generator 30 stores a character number setting information DB 31 (FIG. 2) in which the number of characters b from the beginning of the hash value, which will be described later, and a mask character for storing a mask character string and mask target character string position information. Column information DB 32 (FIG. 3), and the number-of-characters setting information DB 31, the mask character string information DB 32 and the personal information dictionary DB 10 are constructed in storage means / storage means such as a hard disk or RAM. .

  In the apparatus configured as described above, when an electronic document (original text) having personal information is input to the personal information masking apparatus 100, the personal information detecting means 20 applies the personal information dictionary to the character string in the electronic document. Based on the DB 10, a character string that is a target of personal information is detected.

  As data stored in the personal information dictionary DB 10, for example, data such as a character string (name, address, telephone number) to be protected for personal information is stored.

  The personal information detection means 20 searches the input electronic document (text) data for a character string corresponding to the personal information by using the personal information data stored in the personal information dictionary DB 10 as a key, and applies the corresponding information. A character string is detected as a mask target character string.

  At this time, for example, in the personal information dictionary DB 10, when the name data is stored as “first name” and “first name”, not only “first name” and “first name” are searched for as keys, but also “first name and last name” combined with each other. "May be used as a key.

  In the personal information mask generation means 30, a mask character consisting of a part of the character string length and hash value of the personal information target character string (mask target character string) extracted (detected) by the personal information detection means 20. Generate a column.

  Here, the process performed by the personal information mask generating means 30 will be described with reference to FIG. In FIG. 2, in step S <b> 101, the personal information data extracted by the personal information detection unit 20 is input to the personal information mask generation unit 30 as a mask target character string.

  In step S102, the number of characters in the input mask target character string is counted to obtain the character string length (a). For example, when the mask target character string is “Hanako”, the character string length is 6 and a = 6.

  In step S103, the input mask target character string is classified (character type determination). At this time, it is determined whether the mask target character string is a numerical string composed only of numerical values or a character string including characters other than numerals.

  For example, if the character string to be masked is a telephone number, it is classified as a numerical string consisting only of numbers, and if it is address information, it is classified as a character string.

  If it is determined in step S103 that the mask target character string is a character string, the process proceeds to a processing step for generating a mask character string after step S104.

  In step S104, a hash value is generated based on the mask target character string. A hash value is obtained from a character string to be masked plus a random character string.

  For example, when the mask target character string is “hanako”, when a hash value such as Sha1 (Secure hash algorithm 1) is obtained from a random character string (ex.rKRotFVB) + hanako, the following hash value is obtained. can get.

EA1A9CF54F0DDB2065D2DAA9022B7F10875AF643
In addition, when generating the hash value, any means capable of generating a hash value other than Sha1 may be used. The random character string added to the mask target character string is a value for making it difficult to estimate the original mask target character string from the output hash value.

In step S105, a mask character string is generated. Here, a character string for masking a portion of the personal information character string (mask target character string) is generated. That is, for the hash value generated in step S104, the first b characters (b is a positive arbitrary number) are combined with the character string length a obtained in step S102,
“Start symbol” + “character string length a” + “delimiter symbol” + “hash value subsequence b” + “terminal symbol”
A mask character string is generated by combining character strings in the order of.

  The start symbol and the end symbol are for making it easy to find a mask location. Symbols such as “*” are used, and symbols such as “:” are used as delimiters.

  For example, in the case of the mask target character string “hanako”, since the character string length is 6, a = 6, and when b = 4, the mask character string is “* 6: EA1A *”.

  At this time, the number of characters b from the head of the hash value used as a mask character string is a value stored in the character number setting information DB 31 (data storage area), and is a value set at the start of the personal information mask generation process. To do.

  Note that the character string of the hash value used as the mask character string may be a character string at a fixed position. For example, it may be a continuous b character 10 characters after the beginning.

  By incorporating the character string length a of the mask target character string into the mask character string, for example, in the case of a character string such as a surname or first name, there is little variation in the character string length in the same electronic document. Even if a is given, there is little effect of narrowing down the original personal information data (it is not easy to estimate the pre-mask character string), but there are various character string lengths for general names such as brand names. By specifying the character string length a, a narrowing effect is generated. Accordingly, even if the hash value character string length b is reduced, the pre-mask character string can be estimated.

  Therefore, by assigning the value of the mask target character string length a, it is possible to estimate whether it is a surname, a first name, or a general name in the same electronic document.

  The purpose of generating the mask character string using the start symbol and the end symbol “*” is to distinguish the mask conversion portion from other portions, and “*” according to the characteristics of the electronic document. May be used, other symbols (for example, “#”, etc.) may be used.

  When the mask target character string is classified (determined) as a numerical string in step S103, a numerical mask character string is generated in step S106.

Note that when the mask target character string is a numeric string, the number of hash value patterns is smaller and the possibility of estimating the original value is higher than in the case of a character string, so the hash value is not used and the fixed value “ N "shall be used. Therefore, when the mask target character string is the number of characters a, the mask character string is
“Start symbol” + “character string length a” + “separator” + “fixed value N” + “terminal symbol”
And are displayed as “* a: N *”.

  For example, when a 3-digit numerical string is a mask target character string, the mask character string is “* 3: N *”.

  Therefore, when the telephone number “012-345-6789” is a character string to be masked, the mask character string is “* 3: N * − * 3: N * − * 4: N *”.

  Through the above steps, a mask character string is generated and output from the personal information mask generating means 30 (step S107).

  Thereafter, in the personal information replacement means 40, the character string that is the target of personal information (the output of the personal information detection means 20; the mask target character string) is converted into the mask character string generated by the personal information mask generation means 30. A replacement process is performed, and an electronic document (text) after personal information masking is output.

  Next, an embodiment in which masking is performed when data registered in the personal information dictionary DB 10 has a plurality of types will be described with reference to FIG.

  Personal information location for electronic documents (original texts) that contain multiple personal information such as phone numbers, mobile phones, cards, postal codes, e-mail addresses, handle names, addresses, card names, names, names, etc. In the case of masking, it is necessary to set the priority order for each character string to be masked and extract and convert the character string to be masked.

  The personal information dictionary DB 10 in FIG. 3 includes a plurality of types of personal information character strings such as telephone numbers, mobile phone numbers, credit card numbers, postal codes, e-mail addresses, Web handle names, and addresses, and data of their character string lengths. (Note that there is also a personal information character string in which the character string length is not stored).

  First, in step S <b> 201, an electronic document (original text) in which a plurality of the aforementioned personal information exists simultaneously is input to the personal information detection means 20.

  In step S202, the personal information detection means 20 adopts the character string length when a character string length is stored for a plurality of personal information character strings in the personal information dictionary DB 10, and stores the character string length. If not, calculate the number of characters in the string. In step S203, the personal information detection means 20 sets a detection priority based on the number of characters of each personal information character string calculated or adopted.

  That is, for example, a personal information character string having a long character string length is preferentially detected. This means that if a part of the alphabet of the name is used as part of the e-mail address, the e-mail address that should be detected may be detected if it is detected from a short character string length. This is to avoid erroneous detection processing such as being impossible. When the character string lengths are the same, the numerical string is preferentially detected and the detection priority is set.

  Thereafter, in steps S204 and S205, the personal information detection means 20 performs the personal information in the personal information dictionary DB 10 from the input text for all personal information character strings in the personal information dictionary DB 10 in the order of the detection priority set. A character string corresponding to the information character string is detected as a mask target character string.

  At this time, the detection priority is high, and the previously detected character string portion is not detected again. For example, the extracted (detected) portion may be prevented from being duplicated as a detection target by replacing the corresponding portion in the input data with a blank.

  Next, in step S206, the personal information mask generation means 30 obtains the character string length (a) of the mask target character strings that are detected by the processing in step S205 and sequentially input from the personal information detection means 20.

  Here, if the character string length of the corresponding character string is stored in the personal information dictionary DB 10, the character string length is adopted, and if not stored, as in step S102 of FIG. The number of characters in the masked character string is counted to obtain the character string length.

  Next, in steps S207 to S210, the personal information mask generation means 30 uses the mask character strings for masking the mask target character strings sequentially input from the personal information detection means 20 in steps S103 to S106 of FIG. It is generated by the same process as the process.

  That is, in step S207, the corresponding character string type is determined, and in the case of only a numerical value, a numerical mask character string is generated in step S210. If it is determined in step S207 that the character string includes a value other than a numerical value, a hash value of the corresponding character string is generated in step S208, and then in step S209, the character string is masked based on the value of the target character string and the hash value. Generate a string.

  Next, in step S211, the generated mask character string and the position information of the mask target character string detected by the personal information detecting means 20 corresponding to the mask character string are displayed in the mask character string information DB 32. Save to.

  Thereafter, the processing of steps S204 to S211 is repeatedly performed in the order of detection priority order defined in step S203 for all detection target character strings.

  Next, when it is determined in step S204 that the detection has been completed for all the detection target character strings, in step S212, the personal information replacing unit 40 is based on the information stored in the mask character string information DB 32. The mask target character string is replaced with a mask character string corresponding to the mask target character string, and the text subjected to the mask processing in step S213 is output.

  In the character string detection process in step S205, when a character string formed by combining a plurality of numerical strings such as a zip code and a telephone number with a symbol such as a hyphen "-" is detected, a hyphen "-" is used. It is assumed that the detection process is performed when all of the partial numeric strings combined by symbols such as are present in the electronic document (input text) to be extracted.

  As a result, in the case of an EC site or the like, even if a part of a character string of a postal code or a telephone number matches a numerical value string of money display, it can be prevented from being erroneously masked.

  For example, when the postal code is “234-0847”, even if the amount is displayed as “234 yen”, masking is performed only when “0847” is also present in the electronic document (input text). Doing so makes it possible to avoid false masking.

  By the above masking process, the personal information of FIG. 4A to be protected is masked as shown in FIG. 4B as shown in FIG. 4 showing an example in which the present invention is applied to personal information.

  According to FIG. 4, the number of characters of the masked character can be estimated from the number next to the start symbol “*” (character string length a), and “1” before the end symbol “*”. Since “N” exists, it can be estimated that the masked character is composed only of numerical values.

  Further, by utilizing the present invention as a Web page access log analysis system, it is possible to extract marketing information. For example, it is possible to collect purchase information by collecting information on which Web site the user has purchased and what brand of product the user has purchased. An embodiment in which an access log analysis system is constructed using the present invention will be described with reference to FIG.

  In FIG. 5, reference numeral 200 denotes a user terminal (a plurality of user terminals) used by the user, each equipped with the personal information masking device 100 of the present invention.

  Reference numeral 201 denotes Web access data obtained by masking personal information included in information accessed on the Web page on the user terminal 200 by the same process as described with reference to FIGS.

  A web access data DB 202 stores the web access data 201 input via the Internet.

  203 corresponds to the brand name, the character string length of the brand name, and the masking data (b = 4) obtained as a result of performing the masking process previously described with reference to FIGS. This is the brand name list DB stored.

  Collecting access logs masking personal information by realizing a function of masking character strings related to personal information included in access information of Web pages on the user's terminal by the access log analysis system of FIG. Is possible.

  As for the brand name, there is a possibility that the personal name is used. Therefore, there is a possibility that the brand name is masked. However, masking result data on the brand name is created in advance, and the brand name list DB 203 By storing the data as dictionary data as described above, it is possible to analyze the brand name by comparing the Web access data DB 202 and the brand name list DB 203 when analyzing the Web access data.

  Further, by using the present invention as a masking system for confidential information and important information, leakage of confidential information and important information can be prevented.

  Normally, browsing authority is strictly managed for confidential documents including important information and confidential information in a company, and the users and environments that can be viewed are limited. However, for business reasons, it may be necessary to access such important electronic documents from outside the company. For example, in order to confirm the contract document urgently, there is a case where the electronic document is viewed / edited by remote access from outside the company. In such a case, important information may be leaked by being peeped by outsiders.

  Therefore, by using the masking function of the present invention, it is possible to mask a customer name or the like. An embodiment in which a confidential document masking system is constructed using the present invention will be described with reference to FIG.

  In FIG. 6, reference numeral 300 denotes a text server that includes various in-house systems, file servers, and the like, and holds confidential documents (electronic documents) including confidential information and important information.

  301 is an important information dictionary DB in which data such as character strings of confidential information and important information and their character string lengths are stored as a dictionary, similar to the personal information dictionary DB 10 of FIG.

  Reference numeral 302 denotes an important information masking server that masks confidential information and important information included in the confidential document of the text server 300 by using the important information dictionary DB 301 by the same process as described with reference to FIGS. is there.

  The text server 300, the important information dictionary DB 301, and the important information masking server 302 are connected via an in-house network.

  Reference numeral 303 denotes a plurality of remote access terminals connected via an external network.

  A firewall 304 is provided between the important information masking server 302 and the remote access terminal 303.

  In the system configured as described above, when there is an access to an internal network from an external remote access terminal 303, the important information masking server 302 applies to the confidential information and important information in the confidential document of the text server 300. By performing masking and outputting confidential information and important information after the masking process, it becomes possible to prevent leakage of confidential information and important information.

  Further, the present invention can be realized by configuring some or all of the functions of each means in the protection target information masking apparatus of the present embodiment with a computer program and executing the program using the computer. It goes without saying that the procedure in the form of the protection target information masking method can be configured by a computer program and the program can be executed by the computer, and the program for realizing the function by the computer can be read by the computer. Media such as FD (Floppy (registered trademark) Disk), MO (Magneto-Optical disk), ROM (Read Only Memory), memory card, CD (Compact Disk) -ROM, DVD (Digit) al Versatile Disk) -ROM, CD-R, CD-RW, HDD, removable disk, etc., and can be stored or distributed. It is also possible to provide the above program through a network such as the Internet or electronic mail.

10 ... Personal Information Dictionary DB
20 ... Personal information detection means 30 ... Personal information mask generation means 31 ... Character number setting information DB
32 ... Mask character string information DB
40 ... Personal information replacement means 100 ... Personal information masking device 200 ... User terminal 202 ... Web access data DB
203 ... Brand name list DB
300 ... text server 301 ... important information dictionary DB
302 ... Important information masking server 303 ... Remote access terminal 304 ... Firewall

Claims (7)

An information masking device that performs masking by detecting a protection target information portion for text having protection target information,
A protection target information dictionary database in which protection target information is stored;
Protection target information for inputting a text having protection target information as an original text and detecting a character string corresponding to the protection target information stored in the protection target information dictionary database as a mask target character string from the input text Detection means;
A protection target information mask generation unit that obtains a character string length of the mask target character string detected by the protection target information detection unit, and generates a mask character string including the character string length;
Protection target information replacement means for replacing the mask target character string detected by the protection target information detection means with the mask character string generated by the protection target information mask generation means;
A protection target information masking device characterized by comprising: The protection target information dictionary database stores a plurality of types of protection target information and character string length data of the information,
The protection target information detection means sets the detection priority of protection target information based on the number of characters of each protection target information in the protection target information dictionary database, and the input text in the set detection priority order. To detect a character string corresponding to the protection target information in the protection target information dictionary database as a mask target character string,
The protection target information mask generation unit obtains a character string length of the mask target character string sequentially detected by the protection target information detection unit, sequentially generates a mask character string including the character string length, and the generated Storing the mask character string and the position information of the mask target character string detected by the protection target information detection means corresponding to the mask character string;
The protection target information replacement unit replaces the mask target character string detected by the protection target information detection unit with a mask character string corresponding to the mask target character string based on the stored position information. The protection target information masking device according to claim 1, wherein The protection target information mask generation means obtains a hash value based on the mask target character string when the mask target character string is composed of characters other than numeric values, and a mask character string including a partial sequence of the hash value The protection object information masking device according to claim 1 or 2, wherein An information masking method for performing masking by detecting a protection target information part for text having protection target information,
Corresponding to the protection target information stored in the protection target information dictionary database storing the protection target information from the input text, the protection target information detection means inputs the text having the protection target information as the original text Protection target information detecting step for detecting a character string to be masked as a character string to be masked;
A protection target information mask generation unit that obtains a character string length of the mask target character string detected by the protection target information detection unit, and generates a mask character string including the character string length;
A protection target information replacement unit that replaces the mask target character string detected by the protection target information detection unit with the mask character string generated by the protection target information mask generation unit;
A method of masking information to be protected, comprising: The protection target information dictionary database stores a plurality of types of protection target information and character string length data of the information,
The protection target information detection step sets a detection priority of protection target information based on the number of characters of each protection target information in the protection target information dictionary database, and the input text in the set detection priority order. To detect a character string corresponding to the protection target information in the protection target information dictionary database as a mask target character string,
The protection target information mask generation step obtains a character string length of the mask target character string sequentially detected by the protection target information detection means, sequentially generates a mask character string including the character string length, and the generated Storing the mask character string and the position information of the mask target character string detected by the protection target information detection means corresponding to the mask character string;
The protection target information replacement step replaces the mask target character string detected by the protection target information detection means with a mask character string corresponding to the mask target character string based on the stored position information. The protection target information masking method according to claim 4, wherein the protection target information is masked. In the protection target information mask generation step, when the mask target character string is composed of characters other than numerical values, a hash value is obtained based on the mask target character string, and a mask character string including a substring of the hash value 6. The protection target information masking method according to claim 4 or 5, wherein: A protection target information masking program for causing a computer to function as each means according to any one of claims 1 to 3.

Images