JP2013105207A - Information processing method and apparatus for retrieving concealed data - Google Patents

Abstract

PROBLEM TO BE SOLVED: To extract similar data with the data concealed.SOLUTION: The method includes the steps of: extracting a first numeric value and a plurality of feature words present in the periphery of the first numeric value from text data stored in a data storage unit and including the first numeric value; generating, from the extracted first numeric value, one or a plurality of second numeric value to be served as a reference when determining whether or not being approximate to the first numeric value; and performing concealing process on each of the one or the plurality of second numeric values and the plurality of feature words to generate concealed data, and storing the concealed data in the data storage unit.

Description

  The present technology relates to a concealed data search technology.

  With the spread of the cloud, information sharing and utilization utilizing the original features of the cloud is progressing by depositing information in the cloud. Among them, the utilization of confidential data in collaboration and division of labor in the cloud is expected. For example, there may be a usage method in which an individual deposits information on health in the cloud and has it analyzed and organized by a trusted public organization.

  In such situations, text data including numerical values is shared. For example, in medical relations, numerical values are included in patient examination data such as body temperature and blood pressure. Sharing such data is useful to those involved.

  On the other hand, for security and privacy protection, it is common to store such data in the cloud after concealing it. This protects security and privacy, but limits the use of data. That is, the concealment data cannot be appropriately processed by a conventional analysis application or search service. For example, even if it is desired to search for a medical treatment example similar to a patient's symptom, it is difficult to find an appropriate medical treatment example with a simple search if the medical treatment data is concealed.

  There is a technique in which a document is searched with a search keyword, and when a search keyword appears, a numerical value related to the search keyword in the document is compared with a numerical value specified together with the search keyword. However, since it is not considered to conceal data, if it is concealed, it is difficult to extract appropriate data by searching.

  There is also a technology in which data to be searched is concealed and held in a server, and the search condition is concealed in the same way during the search. However, when encryption or hash value calculation is performed, there is a problem that only data that completely matches can be extracted.

  Further, there is a technique for specifying a numerical range as search condition input data and extracting data that at least matches the numerical range. However, it is not considered to perform encryption or hash value calculation.

  There is also a technique for restoring the concealed data once in a safe place and performing a search condition and matching processing. Since the matching process is performed in a plain text state, similar data can be extracted. However, since the search condition is also input in the plain text, the search condition is known to the server that performs the matching process.

JP 2000-11001 A JP 2007-52698 A JP 2005-242740 A JP 2002-108911 A JP 2004-213649 A JP-A-1-58019

  Therefore, an object of the present technology is to provide a technology for enabling extraction of similar data while keeping it secret in one aspect.

  The information processing method according to the first embodiment of the present technology is (A) from the text data stored in the data storage unit and including the first numeric value, the first numeric value and the vicinity of the first numeric value. A step of extracting a plurality of feature words, and (B) one or a plurality of second words that serve as a reference in determining whether to approximate the first numerical value from the extracted first numerical value A generation step of generating a numerical value; and (C) a step of generating concealment data by performing concealment processing for each of one or more second numerical values and a plurality of feature words, and storing the data in a data storage unit; including.

  The information processing method according to the second technique of the present technology includes (A) a first concealment data value of a first numerical value and a second concealment data value of a plurality of first feature words 1 Or receiving a search request including a plurality of search data blocks; (B) a third concealment data value of a plurality of second numerical values and a fourth concealment data value of a plurality of second feature words The first concealment data value and the third concealment for each of the case data blocks stored in the data storage unit for storing a plurality of case data blocks including one or a plurality of data blocks including identification information and identification information The data included in the case data block to be processed from the first similarity for the numerical value calculated from the data value and the number of the fourth anonymized data value that matches the second anonymized data value Each combination of block and search data block A calculation step for calculating a third similarity that is a total value of the second similarities for the matching, and (C) identification information of the case data block whose third similarity exceeds the threshold or the third similarity Transmitting the identification information of the upper predetermined number of matter data blocks to the transmission source of the search request.

  According to one aspect of the present technology, similar data can be extracted while being kept secret.

FIG. 1 is a system configuration diagram of the embodiment. FIG. 2 is a functional block diagram of the registration device. FIG. 3 is a functional block diagram of the management apparatus. FIG. 4 is a functional block diagram of the search device. FIG. 5 is a diagram showing a processing flow at the time of registration. FIG. 6 is a diagram illustrating a processing flow of the FP generation processing. FIG. 7 is a diagram illustrating an example of confidential data. FIG. 8A is a diagram illustrating an example of a data block (first scheme). FIG. 8B is a diagram illustrating an example of a data block (second scheme). FIG. 9A is a diagram illustrating another example of the data block (first scheme). FIG. 9B is a diagram illustrating another example of the data block (second scheme). FIG. 10 is a diagram illustrating an example of data stored in the DB of the management apparatus. FIG. 11 is a diagram showing a processing flow at the time of search. FIG. 12 is a diagram illustrating a processing flow of the second FP generation processing. FIG. 13 is a diagram illustrating an example of confidential data serving as a search condition. FIG. 14A is a diagram illustrating an example of search FP data (first example of the first scheme). FIG. 14B is a diagram illustrating an example of search FP data (second example of the first scheme). FIG. 15A is a diagram illustrating an example of search FP data (first example of the second method). FIG. 15B is a diagram illustrating an example of search FP data (second example of the second scheme). FIG. 16 is a diagram illustrating a processing flow of search processing. FIG. 17 is a diagram illustrating a processing flow of similarity calculation processing. FIG. 18 is a diagram for explaining comparison of numerical hash values when the first method is employed. FIG. 19 is a diagram for explaining comparison of numerical hash values when the second method is employed. FIG. 20 is a diagram illustrating a processing flow of similarity calculation processing. FIG. 21 is a diagram illustrating a processing flow of similarity calculation processing. FIG. 22 is a diagram illustrating an output example. FIG. 23 is a diagram illustrating another example of the FP generation process. FIG. 24 is a functional block diagram of a computer.

  A configuration example of a system according to an embodiment of the present technology is illustrated in FIG. As shown in FIG. 1, a registration device 3, a management device 5, and a search device 7 are connected to a network 1 that is, for example, the Internet. The registration device 3 is a device that performs processing described below to conceal confidential data and registers it in the management device 5, and the number of registration devices 3 is not limited. Further, the search device 7 performs the process described below to conceal confidential data related to the search condition, and transmits a search request including the concealed data and other search conditions to the management device 5. This is a device that receives search results, and there is no limit to the number of search devices 7. The registration device 3 and the search device 7 may be dedicated devices, function as the registration device 3 when registering confidential data, and function as the search device 7 when performing search. May be.

  FIG. 2 shows a functional block diagram of the registration device 3. The registration device 3 includes an input unit 31, a confidential data storage unit 32, an FP (Finger Print) generation unit 33, an FP rule data acquisition unit 34, an FP rule data storage unit 35, an FP data storage unit 36, And a transmission unit 37. In response to an instruction from the user, the input unit 31 stores data to be stored in the management apparatus 5 in the confidential data storage unit 32 or accepts an instruction to select confidential data from the user, and outputs the selection instruction to the FP generation unit. To 33. The FP generation unit 33 generates FP data according to the FP rule data stored in the FP rule data storage unit 35 and stores it in the FP data storage unit 36. When FP rule data is not stored in the FP rule data storage unit 35, the FP generation unit 33 instructs the FP rule data acquisition unit 34 to acquire FP rule data from the management device 5. . The FP rule data acquisition unit 34 acquires FP rule data from the management device 5 in accordance with an instruction from the FP generation unit 33 and stores it in the FP rule data storage unit 35. The transmission unit 37 transmits the FP data stored in the FP data storage unit 36 to the management device 5.

  FIG. 3 shows a functional block diagram of the management apparatus 5. The management device 5 includes an FP rule data storage unit 51, an FP rule data distribution unit 52, an FP registration unit 53, a database (DB) 54, a search processing unit 55, a search request reception unit 56, and a search result transmission. Part 57. The FP rule data distribution unit 52 distributes the FP rule data stored in the FP rule data storage unit 51 in response to a request. The FP registration unit 53 receives FP data from the registration device 3 and stores it in the DB 54. The search request receiving unit 56 receives the search request from the search device 7 and outputs the received search request data to the search processing unit 55. When the search result transmission unit 57 receives the search result from the search processing unit 55, the search result transmission unit 57 transmits the search result to the search device 7 that is the transmission source of the search request. The search processing unit 55 performs search processing using the concealment data and search conditions included in the search request received from the search request receiving unit 56 according to the FP rule data, and sends the search result to the search result transmitting unit 57. Output.

  FIG. 4 shows a functional block diagram of the search device 7. The search device 7 includes an input unit 71, a confidential data storage unit 72, an FP generation unit 73, an FP rule data acquisition unit 74, an FP rule data storage unit 75, a search condition data storage unit 76, and an FP data storage. A section 77, a search request section 78, and an output section 79. In response to an instruction from the user, the input unit 71 stores confidential data for search in the confidential data storage unit 72 or accepts an instruction to select confidential data from the user, and sends the selection instruction to the FP generation unit 73. Output. The input unit 71 also receives search condition data from the user and stores it in the search condition data storage unit 76.

  The FP generation unit 73 generates FP data and the like according to the FP rule data stored in the FP rule data storage unit 75 and stores the FP data in the FP data storage unit 77. When FP rule data is not stored in the FP rule data storage unit 75, the FP generation unit 73 instructs the FP rule data acquisition unit 74 to acquire FP rule data from the management device 5. . The FP rule data acquisition unit 74 acquires FP rule data from the management device 5 in accordance with an instruction from the FP generation unit 73 and stores it in the FP rule data storage unit 75. The search request unit 78 reads out the FP data and the like stored in the FP data storage unit 77 and the search condition data stored in the search condition data storage unit 76 to generate a search request, and sends it to the management device 5. Send. In addition, when the search request unit 78 receives the search result from the management device 5, the search request unit 78 outputs the search result to the output unit 79 and displays the search result on a display device, for example.

  Next, processing contents of the apparatus shown in FIGS. 1 to 4 will be described. First, the FP data registration process will be described with reference to FIGS. First, the input unit 31 receives designation of confidential data to be generated by FP (FIG. 5: step S1). For example, the confidential data stored in the confidential data storage unit 32 may be listed and selected, or the designated confidential data is acquired from another computer or the like and stored in the confidential data storage unit 32. Anyway. Then, the input unit 31 notifies the FP generation unit 33 of the designated confidential data.

  The FP generation unit 33 checks whether FP rule data is stored in the FP rule data storage unit 35 (step S3). When the FP rule data is not stored (step S5: No route), the FP generation unit 33 causes the FP data acquisition unit 34 to acquire the FP rule data and store it in the FP rule data storage unit 35 (step S5). S7).

  On the other hand, when the FP rule data is stored in the FP rule data storage unit 35 (step S5: Yes route), or after step S7, the FP generation unit 33 performs the confidential data designated by the user according to the FP rule data. The FP generation process is performed (step S9). The FP generation process will be described in detail later. As a result, the generated FP data is stored in the FP data storage unit 36.

  Then, the transmission unit 37 transmits the FP data stored in the FP data storage unit 36 to the management device 5 (step S11). On the other hand, when receiving the FP data from the registration device 3, the FP registration unit 53 of the management device 5 stores the received FP data, identification information, and the like in the DB 54 (step S13). The identification information includes, for example, the registrant ID of the registration device 3 and the registration date, and also includes the FPID issued by the FP registration unit 53.

  By repeating such processing, FP data is accumulated in the DB 54.

  Next, the FP generation process will be described with reference to FIGS. The FP generation unit 33 performs normalization processing on the designated confidential data (FIG. 6: Step S21). The confidential data in the present embodiment is text data including numerical values. However, the numerical value may be represented by a half-width number, a full-width number, a Chinese number, an Arabic number, and the like, and may include a difference in units. The normalization process in the present embodiment is a process for unifying such different expressions. For example, “7000” in full-width is converted into “7000” in half-width, and “10,000 yen” is converted into “10000” in half-width numbers. This normalization process is well known and will not be described further.

  Thereafter, the FP generation unit 33 extracts numerical values and feature words in the designated confidential data and stores them in a storage device such as a main memory (step S23). For example, the text of confidential data is decomposed into morphemes by morphological analysis, and further numerical values and feature words (for example, general nouns, proper nouns, etc.) are extracted therefrom.

  For example, consider the case of processing text as shown in FIG. In this example, “patient” “basic” “information” “main complaint” “morning” “body temperature” “38.5” “fever” “symptom” “examination” “heart rate” “measurement” “result” “85” “ Above, "value", "blood test". . . “Treatment” and “policy” are extracted.

  Next, the FP generation unit 33 identifies one unprocessed numerical value among the extracted numerical values (step S25). Then, the FP generation unit 33 generates a numerical value for the FP from the specified numerical value according to the FP rule data stored in the FP rule data storage unit 35 and stores it in a storage device such as a main memory (step) S27). In the present embodiment, in order to be able to determine the approximation of the numerical value, the numerical value is not simply concealed, but the specified numerical value is developed by one of two methods, for example.

In the first method, the specified numerical value is represented by a plurality of significant digits. For example, if “38.5”, the number of significant digits is “3 × 10 ¹ ”, if the number of significant digits is 2, “3.8 × 10 ¹ ”, and the number of significant digits is 3. For example, the expression is changed to “3.82 × 10 ¹ ”. Data on the number of significant digits to be used is included in the FP rule data. As a result, a numerical value indicating the width of the approximation determination is generated.

  In the second method, it is determined to which of the predetermined numerical ranges the specified numerical value belongs, and the upper limit value and the lower limit value of the range to which the specified numerical value belongs are specified. As auxiliary data, a difference from the lower limit value and a difference from the upper limit value are further calculated. For example, when the range is defined in increments of 10, the upper limit value “40” and the lower limit value “30” are specified because “38.2” belongs to the range of 30 to 40. For the auxiliary data, a difference “8.2” from the lower limit value and a difference “−1.2” from the upper limit value are calculated. The FP rule data includes a definition for a range of numerical values. In this way, a numerical value indicating the range of approximation determination and its auxiliary numerical value are generated.

Then, the FP generation unit 33 generates a hash value for the secret part in the numerical value for the generated FP, and stores it in a storage device such as a main memory (step S29). Encryption may be used instead of the hash value. When a key is used, a common key is used for the registration device 3 and the search device 7. In the case of the first method, if the number of significant digits is 1 to 3, hash values are respectively obtained for “3 × 10 ¹ ”, “3.8 × 10 ¹ ”, and “3.82 × 10 ¹ ”. calculate. In the case of the second method, a hash value is calculated for each of the upper limit value and the lower limit value of the range to which the specified numerical value belongs. No hash value is calculated for auxiliary data.

  Thus, in the case of FP data registration processing, a plurality of hash values are calculated for a plurality of numerical values. By simply calculating one numerical hash value, it can only be determined whether or not the numerical values completely match. However, according to the first method, it is possible to determine the coincidence / mismatch within the range of the effective digits, such as one-digit match, two-digit match, and three-digit match. According to the second method, it is possible to specify the case where the upper limit value or the lower limit value of the range to which the specified numerical value belongs, and to determine the difference between the numerical values as described below. Can also be judged.

  The FP generation unit 33 specifies a predetermined number of feature words around the specified numerical value (step S31). The predetermined number is defined in FP rule data, for example. Then, the FP generation unit 33 calculates a hash value for each identified feature word and stores it in a storage device such as a main memory (step S33).

  Then, the FP generation unit 33 stores a data block including a hash value of a numerical value or the like (auxiliary data when there is auxiliary data) and a hash value of a feature word in the FP data storage unit 36 (step S35). .

  When processing is performed so far, the data block as shown in FIG. 8A is stored in the FP data storage unit 36. In the example of FIG. 8A, hash values are calculated for four feature words around the numerical value. FIG. 8A shows an example in which the first method is adopted. Hash (X) represents the hash value of X. On the other hand, when the second method is adopted, a data block as shown in FIG. 8B is generated.

  Thereafter, the FP generation unit 33 determines whether there is an unprocessed numerical value among the numerical values extracted from the confidential data (step S37). If there is an unprocessed numerical value, the process returns to step S25. On the other hand, if there is no unprocessed numerical value, the process returns to the calling process. When the confidential data in FIG. 7 is processed by the first method, another data block as shown in FIG. 9A is generated. On the other hand, when the second method is used, another data block as shown in FIG. 9B is generated. Thus, the FP data includes one or a plurality of data blocks.

  In general, the DB 54 of the management apparatus 5 stores data as shown in FIG. 10, for example. In the example of FIG. 10, FPID, registrant ID, registration date, and FP data are registered. The FP data includes a block number that is an ID of a data block, a numerical value portion, and a feature word portion. The numerical portion of each data block includes a plurality of numerical hash values (NUM (1,1), NUM (1,2), etc.) and auxiliary data (AUX1, etc.) if there is auxiliary data. Further, the feature word portion includes hash values of a plurality of feature words (hash values of M feature words such as KW (1, 1) and KW (1, 2)). In the example of FIG. 10, N data blocks are included, and the number of feature words of each data block is M.

  Next, processing performed at the time of search will be described with reference to FIGS. First, the input unit 71 of the search device 7 receives the designation of confidential data related to the search from the user, and outputs the designation of confidential data to the FP generation unit 73 (FIG. 11: Step S41). When the confidential data is not stored in the confidential data storage unit 72, designated confidential data may be acquired from another computer, for example, and output to the FP generation unit 73.

  Further, the input unit 71 accepts input of search conditions from the user and stores them in the search condition data storage unit 76 (step S43). The user specifies parameters used in the specific search process below. For example, a threshold value of similarity and the number of output results are specified. Which parameter should be specified may be included in the FP rule data, for example. In addition, parameters used for generating FP data are output to the FP generation unit 73.

  Then, the FP generation unit 73 determines whether FP rule data is stored in the FP rule data storage unit 75 (step S45). When the FP rule data is not stored in the FP rule data storage unit 75 (step S47: No route), the FP generation unit 73 causes the FP data acquisition unit 74 to acquire the FP rule data and store the FP rule data. The data is stored in the unit 75 (step S49).

  On the other hand, when the FP rule data is stored in the FP rule data storage unit 75 (step S47: Yes route), or after step S49, the FP generation unit 73 performs the confidential data designated by the user according to the FP rule data. The second FP generation process is performed (step S51). The second FP generation process will be described with reference to FIGS. 12 to 15B.

  The FP generation unit 73 performs normalization processing on the designated confidential data (FIG. 12: step S71). This is the same as step S21.

  Thereafter, the FP generation unit 73 extracts numerical values and feature words in the designated confidential data and stores them in a storage device such as a main memory (step S73). This is the same as step S23.

  For example, consider the case of processing text as shown in FIG. In this example, “patient” “basic” “information” “main complaint” “body temperature” “38” “measurement” “fever” “determination” “examination” “heart rate” “measurement” “value” “80” “normal value” ". . . “Treatment” and “policy” are extracted.

  Next, the FP generation unit 73 identifies one unprocessed numerical value among the extracted numerical values (step S75). Then, the FP generation unit 73 generates one or a plurality of numerical values for the FP from the specified numerical values according to the FP rule data stored in the FP rule data storage unit 75, and stores them in a storage device such as a main memory. Store (step S77). This step is basically the same as step S27.

  However, in this embodiment, when the first method is adopted, designation of the number of significant digits may be included in the search condition. In that case, instead of generating a plurality of numerical values for the numerical value for the FP, a numerical value having a designated effective number of digits is generated.

In the case of the example in FIG. 13, for “38”, if the designated number of significant digits is “2”, a numerical value with an expression such as “3.8 × 10 ¹ ” is generated.

  When the second method is adopted, a method similar to the method described above may be adopted. For example, when the range of numerical values is defined in increments of 10, the value “38” belongs to the range of 30 to 40, so the upper limit value “40” and the lower limit value “30” are specified. For the auxiliary data, a difference “8” from the lower limit value and a difference “−2” from the upper limit value are calculated. In this case, the lower limit value and the upper limit value are treated as the calculation target numerical values of the hash value as representative values.

  However, in the case of the second method, since a range that is determined to be approximate is designated as a search condition, a specified numerical value may be developed based on the range that is determined to be approximate. For example, when a range of plus or minus 1 is designated as a range to be determined to be approximate, if the specified numerical value is “38”, it is determined that the approximation is made if “37” to “39”. Therefore, since the specified numerical value does not exceed the range of the numerical value to which it belongs, it is not determined to be approximate, so the lower limit value “30” of the belonging range is specified as the representative value, and the difference “8” from the representative value is supplemented. Specified as data. On the other hand, if the specified numerical value is “41” and the range of plus or minus 3 is designated as the range to be determined to be approximate, if it is from “38” to “44”, it is determined to be approximate. The Therefore, as the representative value, the lower limit value “30” of the range immediately below the range to which it belongs and the lower limit value “40” of the range to which it belongs are specified as the representative value, and as the auxiliary data, from the first lower limit value And a difference “1” from the second lower limit value is calculated.

  In the example described above, there is a premise that the range that is determined to be approximate beyond the range of the numerical value range is not defined. If such a premise is not satisfied, the following is performed. For example, if the range is defined in increments of 10, but the specified numerical value is “123” and plus or minus 15 is designated as the range to be approximated, “108” to “138” are determined to be approximate. The Therefore, “100”, “110”, “120”, and “130” are specified as representative values, and the difference between them is generated as auxiliary data.

  Then, the FP generation unit 73 generates a hash value for the secret part in the generated numerical value or values and stores it in a storage device such as a main memory (step S79). This is the same as step S29. In the case of the second method, the auxiliary data does not calculate a hash value.

  As described above, when a search is performed, one or a plurality of numerical values are generated as numerical values for the FP with respect to the specified numerical values. However, FP data may be generated as in FP registration. The options as described above are defined in the FP rule data.

  The FP generation unit 73 specifies a predetermined number of feature words around the specified numerical value (step S81). The predetermined number is defined in FP rule data, for example. Then, the FP generation unit 73 calculates a hash value for each identified feature word, and stores it in a storage device such as a main memory (step S83).

  Then, the FP generation unit 73 stores a data block including a numerical hash value or the like (auxiliary data when there is auxiliary data) and a hash value of the feature word in the FP data storage unit 77 (step S85). . This is the same as step S35.

  Thereafter, the FP generation unit 73 determines whether there is an unprocessed numerical value among the numerical values extracted from the confidential data (step S87). If there is an unprocessed numerical value, the process returns to step S75. On the other hand, if there is no unprocessed numerical value, the process returns to the calling process.

  For example, when the simple first method is adopted for the confidential data shown in FIG. 13, FP data as shown in FIG. 14A is generated. When the number of significant digits is “2” in the first method, for example, FP data as shown in FIG. 14B is generated. On the other hand, according to the simple second method, FP data as shown in FIG. 15A is generated. Furthermore, if the range determined to be approximate by the search condition is plus or minus 3, FP data as shown in FIG. 15B is generated.

  In this way, it is possible to determine whether the numerical values are approximated as well as exact matches.

  Returning to the description of the processing in FIG. 11, the search request unit 78 of the search device 7 includes the FP data stored in the FP data storage unit 77 (hereinafter referred to as search FP data for distinction) and the search condition data storage unit. A search request including the data stored in 76 is transmitted to the management apparatus 5 (step S53).

  When receiving the search request including the search FP data and the search condition from the search device 7 (step S55), the search request receiving unit 56 of the management device 5 outputs the search request data to the search processing unit 55. When receiving the search request data, the search processing unit 55 performs a search process (step S57). This search process will be described with reference to FIGS.

  The search processing unit 55 reads FP rule data from the FP rule data storage unit 51 (FIG. 16: step S91). Then, the search processing unit 55 initializes a similar FP array that stores identification information about the FP determined to be similar (step S93). Further, the search processing unit 55 sets a threshold T for similarity determination from FP rule data or search conditions (step S95). The threshold value may be fixed and is included in the FP rule data in that case.

  Thereafter, the search processing unit 55 identifies unprocessed FP data in the DB 54 (step S97). Then, the search processing unit 55 performs a similarity calculation process for the specified FP data and the searched FP data (step S99). The similarity calculation process will be described with reference to FIGS.

  First, the search processing unit 55 calculates a variable c1 for calculating the cumulative value of the common degree of feature words according to the numerical similarity and a variable c2 for counting the number of data blocks included in the specified FP data. It is initialized to 0 (FIG. 17: step S111). In addition, the search processing unit 55 specifies the numerical data N1 of the unprocessed data block among the data blocks included in the search FP data (step S113). When there are a plurality of hash values, and there is auxiliary data, it is specified as N1 including those.

  Further, the search processing unit 55 specifies the numerical data N2 of the unprocessed data block among the data blocks included in the specified FP data (step S115). Also here, if there are a plurality of hash values, and there is auxiliary data, it is specified as N2 including these.

  Then, the search processing unit 55 compares the numerical data N1 and the numerical data N2, and sets the numerical similarity Sim (step S119). In the present embodiment, there are the two methods described above. First, a simple comparison method will be described.

  In the case of the first method, the numerical data N1 includes one or a plurality of hash values, and the numerical data N2 includes a plurality of hash values. For example, FIG. 18 shows an example in which the numerical data N1 includes a plurality of hash values. In the example of FIG. 18, the original numerical value of the numerical data N1 is 38.2, and the hash value when the number of significant digits is 1 to 3 is included in the numerical data N1. On the other hand, FIG. 18 shows the numerical data N2 of the data block (A) whose original numerical value is 38.2, the numerical data N2 of the data block (B) whose original numerical value is 38, and the original numerical value of 39. Numerical data N2 of the data block (C) is shown as a comparison target. In this way, when a plurality of hash values are included in the numerical data N1, if any hash value matches the hash value included in the numerical data N2 to be compared, 1 is set in Sim, If the hash value does not match the hash value included in the numerical data N2 to be compared, 0 is set to Sim. In the example of FIG. 18, since all of the data blocks (A) to (C) match for the number of significant digits “1”, Sim = 1 is set.

On the other hand, when the numerical data N1 includes only one hash value having the designated number of significant digits, it is determined whether or not it matches the one hash value. For example, when only the number of significant digits “3.8 × 10 ¹ ” is included for the numerical data N1, the hash values for the number of significant digits 2 match for the data blocks (A) and (B). However, it is determined that there is no matching hash value for the data block (C).

  In the case of the second method, the numerical data N1 includes one or more hash values and corresponding auxiliary data, and the numerical data N2 includes a plurality of hash values and corresponding auxiliary data. FIG. 19 schematically shows a comparative example between the numerical data N1 and the numerical data N2. For example, the numerical data N2 of the original numerical value “38.2” includes Hash (30), auxiliary data “8.2”, Hash (40), and auxiliary data “−1.8”. On the other hand, the numerical data N1 of the original numerical value “39.1” includes Hash (30), auxiliary data “9.1”, Hash (40), and auxiliary data “−0.9”. . It should be noted that data in a range determined to be approximate is designated as a search condition, and here, plus or minus 1 is designated.

  In this case, the hash value included in the numerical data N1 and the hash value included in the numerical data N2 are compared to determine whether there is a match. In the example of FIG. 19, it is determined that both hash (30) and hash (40) match. In the case of hash (30), it is determined whether or not the difference between the auxiliary data “8.2” of the numerical data N2 and the auxiliary data “9.1” of the numerical data N1 is within a specified range. to decide. In this case, since | 9.1-8.2 | = 0.9, it is within the specified range, and therefore in this embodiment, the numerical similarity Sim = 1 is set. If the difference between the auxiliary data exceeds the designated range, the numerical similarity Sim = 0 is set. Since the same value is obtained for hash (40), it is not necessary to process it.

  Next, a method for setting a real number from 0 to 1 as the numerical similarity Sim according to the numerical similarity will be described. In the case of the first method, as shown in FIG. 18, the numerical data N1 includes a plurality of hash values, and the numerical data N2 also includes a plurality of hash values. Therefore, hash values having the same number of significant digits are compared, and the number of times of matching is counted. For example, in the case of the data block (A), since the original numerical values match, they match three times. In the case of the data block (B), it matches up to 2 significant digits, so it matches twice. Since the data block (C) matches up to 1 significant digit, it matches once. Therefore, for data block (A), Sim = 3 times / 3 (= number of types of significant digits) = 1 is set, and for data block (B), Sim = 2 times / 3 = 0.67. For data block (C), Sim = 1 times / 3 = 0.33 is set.

  On the other hand, in the case of the second method, since the difference between the auxiliary data is calculated as described above, it is calculated by (specified range-absolute value of auxiliary data difference) / (specified range). In the example described above, Sim = | 1-0.9 | /1=0.1 is calculated.

  After that, the search processing unit 55 determines whether the numerical similarity Sim exceeds 0 (step S121). When the numerical similarity Sim is 0, the process proceeds to step S137 in FIG. This is because the similarity for the data block is determined by multiplication with the numerical similarity Sim, so that if the numerical similarity Sim = 0, the data block will be 0 even if the comparison is made. Because. On the other hand, if the numerical similarity Sim> 0, the process proceeds to step S123 in FIG.

  Shifting to the description of the processing in FIG. 20, the search processing unit 55 specifies the hash value KW1 of the unprocessed feature word among the feature words included in the data block corresponding to the numerical data N1 (step S123). Further, the search processing unit 55 specifies the hash value KW2 of the unprocessed feature word among the feature words included in the data block corresponding to the numerical data N2 (step S125). Then, the search processing unit 55 compares the hash value KW1 and the hash value KW2 (step S127).

  In this embodiment, it is assumed that the similarity is 1 when the similarity is calculated for the same FP data. However, in general, the same hash value may appear in different data blocks for numerical values. In this case, there is no particular problem when hash values of different feature words are associated with each other, but when hash values of the same feature words are associated with each other, the similarity is calculated for the same FP data as a whole. As a result, the degree of similarity exceeds 1. Therefore, when it is found that a combination of a hash value for a numerical value and a hash value for a feature word has already appeared, the comparison result is not reflected in the similarity.

  Therefore, the search processing unit 55 determines whether KW1 = KW2 and the combination of the numerical data N1 and KW1 is the first appearance (step S129). If KW1 and KW2 do not match, or if the combination of numerical data N1 and KW1 has already been made, the process proceeds to step S133.

  On the other hand, when KW1 = KW1 and the combination of the numerical data N1 and KW1 is the first appearance, the search processing unit 55 adds the numerical similarity Sim to the variable c1 and sets it as the value of the new variable c1 (step S1). S131). When the numerical similarity Sim is 0 or 1, the number of common feature words is set in the variable c1. On the other hand, when the numerical similarity Sim changes from 0 to 1, the number of common feature words weighted by the numerical similarity Sim for each data block is accumulated in the variable c1.

  Then, the search processing unit 55 determines whether or not the hash value KW2 of the unprocessed feature word exists in the data block corresponding to the numerical data N2 (step S133). If there is a hash value of an unprocessed feature word, the process returns to step S125. On the other hand, if there is no hash value of an unprocessed feature word, the search processing unit 55 determines whether an unprocessed feature word exists in the data block corresponding to the numerical data N1 (step S135). If there is a hash value of an unprocessed feature word, the process returns to step S123. On the other hand, if there is no hash value of an unprocessed feature word, the process proceeds to step S137 in FIG.

  Shifting to the description of the processing in FIG. 21, the search processing unit 55 increments the variable c2 by 1 (step S137). The search processing unit 55 determines whether there is an unprocessed data block in the data block included in the specified FP data (step S139). If an unprocessed data block exists in the identified FP data, the process returns to step S115 in FIG. On the other hand, if there is no unprocessed data block in the specified FP data, the search processing unit 55 determines whether there is an unprocessed data block in the data block included in the search FP data (step S141). . If there is an unprocessed data block in the search FP data, the processing returns to step S113 in FIG. On the other hand, when there is no unprocessed data block in the search FP data, the search processing unit 55 calculates the similarity by c1 / (c2 × block size), and associates it with the identification information of the FP data, for example, the main data block. The data is stored in a storage device such as a memory (step S143). The block size is the number of feature words included in one data block. Then, the process returns to the calling process.

  By performing such processing, it is possible to search confidential data having similar feature words based on numerical values while maintaining confidentiality. Whether the numerical value is approximated or not can be determined while keeping it secret. Further, the search FP data is also concealed, and the management apparatus 5 is kept secret about what kind of search is being performed.

Note that the similarity calculated in step S143 is focused on the specified FP data, and the number of blocks c2 included in the data is included in the similarity calculation formula. In addition, the number of blocks N _Q included in the search FP data is used instead of c2, and the degree of commonness c1 of the feature word occupies the size of the search FP data size (number of blocks N _Q × block size). The degree of similarity can also be considered. The calculation formula is represented by the following formula. Similarly, by the use scene it is also contemplated to use larger max of c2 and _{N Q (c2, N Q)} , or the smaller min to (c2, N _Q) instead of c2.

Here, Q represents search FP data, and D represents FP data to be compared. Block_size is the block size described above, and N _Q represents the number of data blocks of the search FP data. Num _Qi represents the numerical data of the i-th data block in the search FP data, and Num _Dj represents the numerical data of the j-th data block in the FP data to be compared. Sim (Num _Qi , Num _Dj ) represents the similarity Sim between the numerical data of the i-th data block in the search FP data and the numerical data of the j-th data block in the FP data to be compared. B _Qi ∩B _Dj is a hash common to the hash value of the feature word included in the i-th data block in the search FP data and the hash value of the feature word included in the j-th data block in the FP data to be compared. Represents the number of values.

  Returning to the description of the processing in FIG. 16, the search processing unit 55 determines whether or not the calculated similarity exceeds the threshold T specified by the search condition (step S <b> 101). When the similarity exceeds the threshold T, the search processing unit 55 converts the bibliographic data including identification information (such as FPID, registrant ID, and registration date in FIG. 10) of the identified FP data into a similar FP array. (Step S103). The bibliographic data may include the similarity value itself for reference by the searcher. On the other hand, when the similarity is equal to or less than the threshold value T, the process proceeds to step S105.

  When it is determined in step S101 that the similarity is equal to or less than the threshold value T or after step S103, the search processing unit 55 determines whether unprocessed FP data exists in the DB 54 (step S105). If unprocessed FP data exists, the process returns to step S97. On the other hand, when there is no unprocessed FP data, the search processing unit 55 outputs the data of the similar FP array to the search result transmission unit 57 (step S107). Then, the process returns to the calling process. Note that the detailed data of the registrant may be added to the search result transmission unit 57 in an added form.

  In this way, not only perfect matching but also similar FP data is specified, and data related to the FP data is extracted.

  Returning to the description of the processing in FIG. 11, the search result transmission unit 57 transmits the data of the search result received from the search processing unit 55 to the search device 7 that is the transmission source of the search request (step S59). The search request unit 78 of the search device 7 receives the search result from the management device 5 and outputs it to the output unit 79 (step S61). Then, the output unit 79 outputs the search result to a display device or the like. For example, data as shown in FIG. 22 is displayed on the display device. In the example of FIG. 22, FPID, registrant, registration date, and similarity are displayed. As described above, the sorted results may be presented in descending order of similarity.

  As a result, the searcher can specify a registrant of similar FP data, and can request specific information from the registrant.

  For example, when registering FP data for medical data, since the medical data itself is not disclosed, there is no problem in terms of privacy protection and security protection, and registration of FP data in the management apparatus 5 is promoted. . On the other hand, the patient's data is kept secret on the search side, and there is no problem in terms of privacy protection and security protection, so the use can be promoted. If the presence of a similar case can be confirmed, information such as a treatment method can be acquired at an early stage by making a separate inquiry, which is also effective for patients.

  Although the embodiment of the present technology has been described above, the present technology is not limited to this. For example, processing for specifying a feature word from confidential data has been performed, but in addition to this, a processing flow as shown in FIG. 23 may be performed. Steps S201 to S205 are the same as in FIG. The FP generation unit 33 extracts synonyms of the identified feature words from the dictionary (step S201), and calculates a hash value for each feature word and each synonym (step S203). Further, the FP generation unit 33 stores a data block including a synonym hash value in addition to the feature word in the FP data storage unit 36 (step S205). In this way, a hash value for a synonym may also be included in the FP data.

  Furthermore, although the example in which the threshold value T is included in the search condition has been described above, for example, the upper designated number of FP data may be extracted in descending order of similarity.

  Furthermore, the functional block diagram shown above is an example, and may not necessarily match the actual program module configuration. Further, regarding the processing flow, as long as the processing result does not change, the order of the processing steps may be changed or may be executed in parallel.

  The FP rule data may be managed by a device other than the management device 5.

  The registration device 3, the management device 5, and the search device 7 described above are computer devices, and as shown in FIG. 24, a memory 2501, a CPU (Central Processing Unit) 2503, a hard disk drive (HDD: Hard). Disk Drive) 2505, a display control unit 2507 connected to the display device 2509, a drive device 2513 for the removable disk 2511, an input device 2515, and a communication control unit 2517 for connecting to a network are connected by a bus 2519. . An operating system (OS: Operating System) and an application program for executing the processing in this embodiment are stored in the HDD 2505, and are read from the HDD 2505 to the memory 2501 when executed by the CPU 2503. The CPU 2503 controls the display control unit 2507, the communication control unit 2517, and the drive device 2513 according to the processing content of the application program, and performs a predetermined operation. Further, data in the middle of processing is mainly stored in the memory 2501, but may be stored in the HDD 2505. In an embodiment of the present technology, an application program for performing the above-described processing is stored in a computer-readable removable disk 2511 and distributed, and installed from the drive device 2513 to the HDD 2505. In some cases, the HDD 2505 may be installed via a network such as the Internet and the communication control unit 2517. Such a computer apparatus realizes various functions as described above by organically cooperating hardware such as the CPU 2503 and the memory 2501 described above and programs such as the OS and application programs. .

  The above-described embodiment can be summarized as follows.

  The information processing method according to the first embodiment of the present embodiment is based on (A) text data stored in the data storage unit and including the first numerical value, and the first numerical value and the surroundings of the first numerical value. A process of extracting a plurality of feature words existing in (B), and (B) one or a plurality of first reference values used as a reference in determining whether to approximate the first numerical value from the extracted first numerical value Generation processing for generating a numerical value of 2, and (C) concealment processing is performed on each of one or more second numerical values and a plurality of feature words to generate concealment data, and store the data in the data storage unit Processing.

  If the second numerical value is generated and the concealment data of the second numerical value is generated in this way, it is possible to detect not only a perfect match but also an approximate numerical value. If a plurality of second numerical values are generated instead of only one, it becomes easier to detect a more approximate numerical value. This is the same during data registration and data retrieval. In addition, it can be said that the numerical value used as a reference in determining whether or not the first numerical value is similar is a numerical value indicating the range of approximation determination.

  In addition, the generation process described above may be a process of generating a plurality of second numerical values in which the extracted first numerical values are represented by different effective digits. In this way, the accuracy of approximation can be adjusted by the number of significant digits.

  Further, the generation processing described above specifies a plurality of second numerical values that are an upper limit value and a lower limit value of a predetermined numerical range including the extracted first numerical value, and the first numerical value and the lower limit value. And a process of calculating a difference between the first numerical value and the upper limit value and storing the difference in the data storage unit. In this way, it becomes easy to calculate the difference from the original numerical value during the search.

  Further, the generation process described above specifies one or more second numerical values representing the first numerical value from the extracted first numerical value and the setting of the numerical value range for classifying the numerical value. The process may include a process of calculating a difference between the first numerical value and the first numerical value and the first numerical value and storing the difference in the data storage unit. For example, when generating the data for search, if the second numerical value is generated in consideration of the range determined to be approximate, the concealment data for the numerical value approximated at the time of the search can be accurately specified. Become.

  In addition, the generation process described above may be a process of generating one second numerical value in which the extracted first numerical value is represented by the designated effective number of digits. By specifying the number of significant digits in this way at the time of search, approximation can be determined with a desired accuracy.

  Furthermore, the information processing method according to the first embodiment of the present embodiment generates processing of extracting synonyms of a plurality of feature words and concealing synonyms to generate concealed data, and a data storage unit May further include a process of storing in. According to this, it becomes easy to extract similar concealment data.

  The information processing method according to the second aspect of the present embodiment includes (A) a first concealment data value of a first numerical value and a second concealment data value of a plurality of first feature words. Processing for receiving a search request including one or a plurality of search data blocks, and (B) a third concealment data value of a plurality of second numerical values and a fourth concealment of a plurality of second feature words For each of the case data blocks stored in the data storage unit for storing a plurality of case data blocks including one or more data blocks including the data value and identification information, the first concealed data value and the third Included in the case data block to be processed from the first similarity degree for the numerical value calculated from the concealment data value and the number of the fourth concealment data value that matches the second concealment data value Each set of data block and search data block A calculation process for calculating a third similarity that is a total value of the second similarities for the combination, and (C) identification information of the case data block whose third similarity exceeds the threshold or the third similarity Includes a process of transmitting the identification information of the upper predetermined number of item data blocks to the transmission source of the search request.

  In this way, both the case data block stored in the data storage unit and the data block included in the search request are concealed, and the numerical similarity can be specified, and the overall similarity is calculated. it can. Therefore, it becomes possible to specify a case data block having a higher similarity.

  In addition, when the calculation process described above has the 3rd encryption data value which corresponds to a 1st encryption data value, a 1st similarity is set to 1 and the 1st encryption data If there is no third anonymized data value that matches the value, a process of setting the first similarity to 0 may be included. For example, if a third concealment data value is prepared for a plurality of second numerical values representing approximation of the original value, it is easy to detect the presence of the numerical value to be approximated.

  Further, the first concealment data value described above may be a concealment data value of a representative value for the original numerical value of the first numerical value. The search request described above may further include a first auxiliary numerical value that is a difference between the first numerical value and the representative value of the original numerical value, and range data for approximation determination. . The plurality of second numerical values are the lower limit value and upper limit value of the range to which the original numerical value belongs, and the data block described above includes the lower limit value of the range to which the original numerical value of the second numerical value belongs, A second auxiliary numerical value that is a difference from the original numerical value and a third auxiliary numerical value that is a difference between the original numerical value and the upper limit value may be further included. In such a case, if there is a third concealment data value that matches the first concealment data value, the calculation process described above is the first auxiliary for the first concealment data value. A process of calculating a difference between the numerical value and the second auxiliary value or the third auxiliary value for the third anonymized data value that matches the first anonymized data value; the first auxiliary value and the second If the difference between the auxiliary numerical value or the third auxiliary numerical value is within the range for approximation determination, the first similarity is set to 1, and the first auxiliary numerical value and the second auxiliary numerical value or third If the difference from the auxiliary numerical value is not within the range for approximation determination, the first similarity may be set to 0.

  Furthermore, when the first concealment data value of the first numerical value is included in the plurality of data blocks, the calculation processing described above is performed by the third concealment data value that matches the first concealment data value. A process of setting the similarity according to the number of the first similarity as the first similarity may be included. In this way, the first similarity that is not only 0 or 1 can be set.

  Further, the first concealment data value is the concealment data value of the representative value for the original numerical value of the first numerical value, and the search request includes a representative value for the original numerical value of the first numerical value. You may make it further contain the 1st auxiliary | assistant numerical value which is a difference, and the data of the range for an approximation determination. Further, the plurality of second numerical values described above are the lower limit value and the upper limit value of the range to which the original numerical value belongs, and the data block described above includes the range of the range to which the original numerical value of the second numerical value belongs. A second auxiliary numerical value that is a difference between the lower limit value and the original numerical value, and a third auxiliary numerical value that is a difference between the original numerical value and the upper limit value may be further included. When the calculation process described above has a third concealed data value that matches the first concealed data value, the first auxiliary numerical value for the first concealed data value, A process of calculating a difference between the second auxiliary value or the third auxiliary value for the third anonymized data value that matches the first anonymized data value, and the first auxiliary value and the second auxiliary value Alternatively, a process of setting the similarity according to the difference between the third auxiliary numerical value and the numerical value representing the range for the approximate determination as the first similarity may be included. In this way, the first similarity that is not only 0 or 1 can be set.

  It is possible to create a program for causing a computer to carry out the processing described above, such as a flexible disk, an optical disk such as a CD-ROM, a magneto-optical disk, and a semiconductor memory (for example, ROM). Or a computer-readable storage medium such as a hard disk or a storage device.

  The following supplementary notes are further disclosed with respect to the embodiments including the above examples.

(Appendix 1)
A process of extracting a first numerical value and a plurality of feature words existing around the first numerical value from text data stored in the data storage unit and including the first numerical value;
A generation process for generating one or more second numerical values serving as a reference in determining whether to approximate the first numerical value from the extracted first numerical value;
A process of generating concealment data by performing concealment processing for each of the one or more second numerical values and the plurality of feature words, and storing the data in the data storage unit;
Processing method in which a computer executes a process including:

(Appendix 2)
The generation process is
The information processing method according to supplementary note 1, which is a process of generating a plurality of second numerical values in which the extracted first numerical values are expressed by different effective digits.

(Appendix 3)
The generation process is
A process of specifying a plurality of second numerical values that are an upper limit value and a lower limit value of a predetermined numerical value range including the extracted first numerical value;
A process of calculating a difference between the first numerical value and the lower limit value and a difference between the first numerical value and the upper limit value and storing the difference in the data storage unit;
The information processing method according to appendix 1, including:

(Appendix 4)
The generation process is
A process of specifying one or a plurality of second numerical values representing the first numerical value from the extracted first numerical value and setting of a numerical value range for classifying the numerical value;
A process of calculating a difference between the one or more second numerical values and the first numerical value and storing the difference in the data storage unit;
The information processing method according to appendix 1, including:

(Appendix 5)
The generation process is
The information processing method according to claim 1, which is a process of generating one second numerical value in which the extracted first numerical value is represented by a designated number of significant digits.

(Appendix 6)
Processing to extract synonyms of the plurality of feature words;
Processing to generate concealment data by concealing the synonym and store in the data storage unit;
The information processing method according to any one of appendices 1 to 5, wherein the processing further includes:

(Appendix 7)
A process of receiving a search request including one or more search data blocks including a first concealment data value of a first numerical value and a second concealment data value of a plurality of first feature words;
A case data block including one or a plurality of data blocks including a plurality of second concealed data values of a second numerical value and a fourth concealed data value of a plurality of second feature words and identification information; 1st similarity about the numerical value calculated from said 1st concealment data value and said 3rd concealment data value about each of said case data block stored in the data storage part to store two or more And the number of the fourth anonymized data values that coincide with the second anonymized data value, the number of each combination of the data block and the search data block included in the case data block to be processed A calculation process for calculating a third similarity that is a total value of two similarities;
A process of transmitting identification information of a case data block in which the third similarity exceeds a threshold or identification information of a case data block having a third highest degree of similarity to the transmission source of the search request;
Processing method in which a computer executes a process including:

(Appendix 8)
The calculation process
If the third concealed data value that matches the first concealed data value exists, the first similarity is set to 1, and the first concealed data value matches the first concealed data value. The information processing method according to appendix 7, including a process of setting the first similarity to 0 when there is no 3 concealed data value.

(Appendix 9)
The first concealment data value is a concealment concealment data value of a representative value for the original numerical value of the first numerical value;
The search request further includes a first auxiliary numerical value that is a difference from the representative value of the original numerical value of the first numerical value, and range data for approximation determination,
The plurality of second numerical values are a lower limit value and an upper limit value of a range to which the original numerical value belongs,
The data block is a difference between a second auxiliary value that is a difference between the lower limit value of the range to which the original value of the second value belongs and the original value, and the original value and the upper limit value. A third auxiliary numerical value,
The calculation process
If there is a third concealed data value that matches the first concealed data value, the first auxiliary value for the first concealed data value and the first concealment A process of calculating a difference between the second auxiliary value or the third auxiliary value for the third concealed data value that matches a data value;
If the difference between the first auxiliary value and the second auxiliary value or the third auxiliary value is within the range for the approximation determination, the first similarity is set to 1, A process of setting the first similarity to 0 when the difference between the first auxiliary numerical value and the second auxiliary numerical value or the third auxiliary numerical value is not within the range for the approximation determination;
The information processing method according to appendix 7, including:

(Appendix 10)
A plurality of first concealment data values of the first numerical value are included in the data block;
The calculation process
The information processing method according to appendix 7, including a process of setting a similarity according to the number of the third anonymized data values matching the first anonymized data value as the first similarity.

(Appendix 11)
The first concealment data value is a concealment concealment data value of a representative value for the original numerical value of the first numerical value;
The search request further includes a first auxiliary numerical value that is a difference from the representative value of the original numerical value of the first numerical value, and range data for approximation determination,
The plurality of second numerical values are a lower limit value and an upper limit value of a range to which the original numerical value belongs,
The data block is a difference between a second auxiliary value that is a difference between the lower limit value of the range to which the original value of the second value belongs and the original value, and the original value and the upper limit value. A third auxiliary numerical value,
The calculation process
If there is a third concealed data value that matches the first concealed data value, the first auxiliary value for the first concealed data value and the first concealment A process of calculating a difference between the second auxiliary value or the third auxiliary value for the third concealed data value that matches a data value;
Similarity according to the difference between the first auxiliary numerical value, the second auxiliary numerical value or the third auxiliary numerical value, and the numerical value representing the range for the approximation determination is set as the first similarity. Processing to
The information processing method according to appendix 7, including:

(Appendix 12)
A data storage unit;
A first numerical value and a plurality of feature words existing around the first numerical value are extracted from text data stored in the data storage unit and including the first numerical value, and the extracted first 1 or a plurality of second numerical values serving as a reference in determining whether to approximate the first numerical value or not from the numerical values of the first numerical value, the one or more second numerical values and the plurality of characteristics Generating a concealment data by performing concealment processing for each word, and storing in a second data storage unit;
An information processing apparatus.

(Appendix 13)
A receiving unit for receiving a search request including one or a plurality of search data blocks including a first concealment data value of a first numerical value and a second concealment data value of a plurality of first feature words;
A case data block including one or a plurality of data blocks including a plurality of second concealed data values of a second numerical value and a fourth concealed data value of a plurality of second feature words and identification information; 1st similarity about the numerical value calculated from said 1st concealment data value and said 3rd concealment data value about each of said case data block stored in the data storage part to store two or more And the number of the fourth anonymized data values that coincide with the second anonymized data value, the number of each combination of the data block and the search data block included in the case data block to be processed A search processing unit that calculates a third similarity that is a total value of two similarities;
A transmission unit that transmits identification information of a case data block in which the third similarity exceeds a threshold or identification information of a case data block having a third predetermined higher degree of similarity to the transmission source of the search request;
An information processing apparatus.

3 Registration Device 31 Input Unit 32 Confidential Data Storage Unit 33 FP Generation Unit 34 FP Rule Data Acquisition Unit 35 FP Rule Data Storage Unit 36 PF Data Storage Unit 37 Transmission Unit 5 Management Device 51 FP Rule Data Storage Unit 52 FP Rule Data Distribution Unit 53 PF Registration Unit 54 DB
55 Search processing unit 56 Search request reception unit 57 Search result transmission unit 7 Search device 71 Input unit 72 Confidential data storage unit 73 FP generation unit 74 FP rule data acquisition unit 75 FP rule data storage unit 76 Search condition data storage unit 77 FP data Storage unit 78 Search request unit 79 Output unit

Claims (13)

A process of extracting a first numerical value and a plurality of feature words existing around the first numerical value from text data stored in the data storage unit and including the first numerical value;
A generation process for generating one or more second numerical values serving as a reference in determining whether to approximate the first numerical value from the extracted first numerical value;
A process of generating concealment data by performing concealment processing for each of the one or more second numerical values and the plurality of feature words, and storing the data in the data storage unit;
Processing method in which a computer executes a process including: The generation process is
The information processing method according to claim 1, further comprising: generating a plurality of second numerical values representing the extracted first numerical values with different effective digits. The generation process is
A process of specifying a plurality of second numerical values that are an upper limit value and a lower limit value of a predetermined numerical range including the extracted first numerical value;
A process of calculating a difference between the first numerical value and the lower limit value and a difference between the first numerical value and the upper limit value and storing the difference in the data storage unit;
The information processing method according to claim 1 including: The generation process is
A process of specifying one or a plurality of second numerical values representing the first numerical value from the extracted first numerical value and setting of a numerical value range for classifying the numerical value;
A process of calculating a difference between the one or more second numerical values and the first numerical value and storing the difference in the data storage unit;
The information processing method according to claim 1 including: The generation process is
The information processing method according to claim 1, wherein the first numerical value is a process of generating one second numerical value representing the indicated effective number of digits. Processing to extract synonyms of the plurality of feature words;
Processing to generate concealment data by concealing the synonym and store in the data storage unit;
The information processing method according to any one of claims 1 to 5, wherein the processing further includes: A process of receiving a search request including one or more search data blocks including a first concealment data value of a first numerical value and a second concealment data value of a plurality of first feature words;
A case data block including one or a plurality of data blocks including a plurality of second concealed data values of a second numerical value and a fourth concealed data value of a plurality of second feature words and identification information; 1st similarity about the numerical value calculated from said 1st concealment data value and said 3rd concealment data value about each of said case data block stored in the data storage part to store two or more And the number of the fourth anonymized data values that coincide with the second anonymized data value, the number of each combination of the data block and the search data block included in the case data block to be processed A calculation process for calculating a third similarity that is a total value of two similarities;
A process of transmitting identification information of a case data block in which the third similarity exceeds a threshold or identification information of a case data block having a third highest degree of similarity to the transmission source of the search request;
Processing method in which a computer executes a process including: The calculation process
If the third concealed data value that matches the first concealed data value exists, the first similarity is set to 1, and the first concealed data value matches the first concealed data value. The information processing method according to claim 7, further comprising: a process of setting the first similarity to 0 when there is no 3 concealed data value. The first concealment data value is a concealment concealment data value of a representative value for the original numerical value of the first numerical value;
The search request further includes a first auxiliary numerical value that is a difference from the representative value of the original numerical value of the first numerical value, and range data for approximation determination,
The plurality of second numerical values are a lower limit value and an upper limit value of a range to which the original numerical value belongs,
The data block is a difference between a second auxiliary value that is a difference between the lower limit value of the range to which the original value of the second value belongs and the original value, and the original value and the upper limit value. A third auxiliary numerical value,
The calculation process
If there is a third concealed data value that matches the first concealed data value, the first auxiliary value for the first concealed data value and the first concealment A process of calculating a difference between the second auxiliary value or the third auxiliary value for the third concealed data value that matches a data value;
If the difference between the first auxiliary value and the second auxiliary value or the third auxiliary value is within the range for the approximation determination, the first similarity is set to 1, A process of setting the first similarity to 0 when the difference between the first auxiliary numerical value and the second auxiliary numerical value or the third auxiliary numerical value is not within the range for the approximation determination;
The information processing method according to claim 7. A plurality of first concealment data values of the first numerical value are included in the data block;
The calculation process
The information processing method according to claim 7, further comprising: setting a similarity according to the number of the third concealed data values that matches the first concealed data value as the first similarity. The first concealment data value is a concealment concealment data value of a representative value for the original numerical value of the first numerical value;
The search request further includes a first auxiliary numerical value that is a difference from the representative value of the original numerical value of the first numerical value, and range data for approximation determination,
The plurality of second numerical values are a lower limit value and an upper limit value of a range to which the original numerical value belongs,
The data block is a difference between a second auxiliary value that is a difference between the lower limit value of the range to which the original value of the second value belongs and the original value, and the original value and the upper limit value. A third auxiliary numerical value,
The calculation process
If there is a third concealed data value that matches the first concealed data value, the first auxiliary value for the first concealed data value and the first concealment A process of calculating a difference between the second auxiliary value or the third auxiliary value for the third concealed data value that matches a data value;
Similarity according to the difference between the first auxiliary numerical value, the second auxiliary numerical value or the third auxiliary numerical value, and the numerical value representing the range for the approximation determination is set as the first similarity. Processing to
The information processing method according to claim 7. A data storage unit;
A first numerical value and a plurality of feature words existing around the first numerical value are extracted from text data stored in the data storage unit and including the first numerical value, and the extracted first 1 or a plurality of second numerical values serving as a reference in determining whether to approximate the first numerical value or not from the numerical values of the first numerical value, the one or more second numerical values and the plurality of characteristics Generating a concealment data by performing concealment processing for each word, and storing in a second data storage unit;
An information processing apparatus. A receiving unit for receiving a search request including one or a plurality of search data blocks including a first concealment data value of a first numerical value and a second concealment data value of a plurality of first feature words;
A case data block including one or a plurality of data blocks including a plurality of second concealed data values of a second numerical value and a fourth concealed data value of a plurality of second feature words and identification information; 1st similarity about the numerical value calculated from said 1st concealment data value and said 3rd concealment data value about each of said case data block stored in the data storage part to store two or more And the number of the fourth anonymized data values that coincide with the second anonymized data value, the number of each combination of the data block and the search data block included in the case data block to be processed A search processing unit that calculates a third similarity that is a total value of two similarities;
A transmission unit that transmits identification information of a case data block in which the third similarity exceeds a threshold or identification information of a case data block having a third predetermined higher degree of similarity to the transmission source of the search request;
An information processing apparatus.

Images