KR20230105569A - A method and a device for determining predicted values for anonymous values - Google Patents

Abstract

According to an embodiment, in a method for determining a predicted value for an anonymous value by a device, a first record including an anonymized value and an identifiable first identification value according to a preset de-identification level from original data and a first record obtaining a plurality of second records composed of 2 identification values; determining one or more reference records corresponding to the first record among the plurality of second records, based on a comparison result between the first identification value and the second identification value; and determining a predicted value for the anonymous value within the range of the de-identification level using the one or more reference records.

Description

Method and device for determining predicted values for anonymity {A METHOD AND A DEVICE FOR DETERMINING PREDICTED VALUES FOR ANONYMOUS VALUES}

The present disclosure relates to a method and device for determining a predictive value for an anonymous value, and more particularly, to provide predictive information so that anonymous information processed to a level in which personal information cannot be identified can be used as meaningful data It's about technology.

As the type and amount of data processed increases with the development of technologies such as big data and cloud computing, data including personal information is actively collected and shared in various industries. However, since such personal information has the possibility of identifying a specific individual, there is a high risk of accidents due to information leakage. Therefore, data containing personal information must be anonymized in a form in which a specific individual cannot be identified based on the Personal Information Protection Act, etc. do. Examples of methods of anonymizing data include K-anonymity technology, which reduces database relevance so that individuals are not identified in a data set.

As described above, the conventional technology for utilizing data containing personal information is used for purposes such as statistical writing and academic research by anonymizing personal information in a form in which a specific individual cannot be identified.

However, this prior art can guarantee the anonymity of data by utilizing anonymized information, but has a disadvantage in that the value as a statistical characteristic of the data is reduced. For example, as anonymous information is subdivided, statistical values are also subdivided, reducing the amount of data that can actually be utilized, making it difficult to utilize as data.

Accordingly, a demand for a technology for overcoming the above-mentioned disadvantages and limitations and increasing the utilization value of anonymized information as data is gradually increasing.

Korean Patent Publication No. 10-2017-0126804 (2017.11.20.), Anonymized Statistical Database Question System and Method

An embodiment of the present disclosure is intended to solve the above-mentioned problems of the prior art, and to provide prediction information for anonymized information so that the anonymized information can be used as more meaningful data.

The purpose of the present disclosure is not limited to the purpose mentioned above, and other objects not mentioned will be clearly understood from the description below.

A method for determining a predicted value for an anonymous value by a device according to a first aspect of the present disclosure includes a first record including an anonymized value and an identifiable first identification value according to a predetermined de-identification level from original data, and a first record including an identifiable first identification value. obtaining a plurality of second records composed of 2 identification values; determining one or more reference records corresponding to the first record among the plurality of second records, based on a comparison result between the first identification value and the second identification value; and determining a predicted value for the anonymous value within the range of the non-identification level by using the one or more reference records.

The determining of the prediction value may include determining the prediction value based on at least one of the presence or absence of higher data for the first record, a statistical analysis result of the one or more reference records, and the number of the first records. can

In the determining of the prediction value, when there is no higher data for the first record, the prediction value may be determined using a unit price per case determined based on the one or more reference records.

Also, in each of the one or more reference records, the second identification value may be different from the first identification value only for one of a plurality of attributes defining each record.

In addition, the determining of the predicted value may include determining a plurality of unit prices per case based on each of the plurality of reference records when the number of the one or more reference records is plural, and an average or weighted average of the plurality of unit prices per case The prediction value may be determined using one value.

In addition, the determining of the prediction value may include determining a total anonymous value indicating a sum of the plurality of anonymous values based on the reference record when there is upper data for the first record and there are a plurality of anonymous values. step; and determining each of the plurality of anonymous values to be 1 when the total anonymous value is equal to the number of the plurality of anonymous values.

In addition, the determining of the prediction value may include determining a total anonymous value indicating a sum of the plurality of anonymous values based on the reference record when there is upper data for the first record and the anonymous value is one. step; and determining the anonymous value as the total anonymous value.

In addition, the determining of the prediction value may include determining a total anonymous value indicating a sum of the plurality of anonymous values based on the reference record when there is upper data for the first record and there are a plurality of anonymous values. step; and if the sum anonymous value is greater than the number of the plurality of anonymous values, determining a combination of the plurality of anonymous values within a range of the sum anonymous value; and determining the plurality of anonymous values according to one of combinations of the plurality of anonymous values using the unit price per case determined based on the one or more reference records.

The determining of the plurality of anonymous values may include determining a first prediction value for each of the plurality of anonymous values using a unit price per case determined based on the one or more reference records; determining a second prediction value for each of the plurality of anonymous values using a difference between the total anonymous value and the number of the plurality of anonymous values; and determining a prediction value for each of the plurality of anonymous values by using the first prediction value and the second prediction value.

A device for determining a predicted value for an anonymous value according to a second aspect of the present disclosure includes a first record including an anonymized anonymized value and an identifiable first identification value according to a preset de-identification level from original data and a second identification value. a receiving unit acquiring a plurality of second records composed of values; and determining one or more reference records corresponding to the first record among the plurality of second records based on a comparison result between the first identification value and the second identification value, and using the one or more reference records to determine the reference records. A processor for determining a prediction value for an anonymous value within the range of the non-identification level; may include.

In addition, the processor may determine the prediction value based on at least one of the presence or absence of higher data for the first record, a statistical analysis result for the one or more reference records, and the number of the first records.

In addition, when there is no upper data for the first record, the processor may determine the predicted value using a unit price per case determined based on the one or more reference records.

In addition, when there is upper data for the first record and there are a plurality of anonymous values, the processor determines a total anonymous value indicating a sum of the plurality of anonymous values based on the reference record, and determines the sum anonymous value and the number of the plurality of anonymous values are the same, each of the plurality of anonymous values may be determined to be 1.

In addition, when there is upper data for the first record and there are a plurality of anonymous values, the processor determines a total anonymous value indicating a sum of the plurality of anonymous values based on the reference record, and determines the sum anonymous value When is greater than the number of the plurality of anonymous values, a combination of the plurality of anonymous values is determined within the range of the total anonymous value, and a unit price per case determined based on the one or more reference records is used to determine the plurality of anonymous values. The plurality of anonymous values may be determined according to one of combinations of anonymous values.

A third aspect of the present disclosure may provide a computer readable recording medium recording a program for executing the method according to the first aspect in a computer. Alternatively, the fourth aspect of the present disclosure may provide a computer program stored in a recording medium to implement the method according to the first aspect.

According to an embodiment of the present disclosure, it is possible to provide a prediction value that is estimated to be close to the actual value for an anonymous value whose value cannot be identified due to anonymization, and thus provide reliable data for anonymity guaranteed. By providing predicted values, the value of use as data of anonymous data can be greatly improved.

In addition, reliability is improved by estimating the predicted value to be closer to the actual value in a more appropriate way depending on the situation, considering various factors such as the presence or absence of upper data, the comparison result between the reference record and the target record, the number of anonymous values, and the total anonymous value. can enhance

The effects of the present disclosure are not limited to the above effects, and should be understood to include all effects that can be inferred from the configuration of the invention described in the detailed description or claims of the present disclosure.

1 is a block diagram illustrating a configuration of a device according to an exemplary embodiment.
2 is a flowchart illustrating a method for a device to determine a prediction value for an anonymous value according to an embodiment.
3 is a diagram illustrating a first record, a second record, and a reference record according to an exemplary embodiment.
4 is a diagram for explaining an operation in which a device according to an embodiment determines a prediction value for an anonymous value according to the first embodiment.
5 is a diagram for explaining an operation in which a device according to an embodiment determines a prediction value for an anonymous value according to a second embodiment.
6 is a diagram for explaining an operation in which a device according to an embodiment determines a prediction value for an anonymous value according to a third embodiment.

The terms used in the embodiments have been selected as general terms that are currently widely used as much as possible while considering the functions in the present disclosure, but they may vary depending on the intention or precedent of a person skilled in the art, the emergence of new technologies, and the like. In addition, in a specific case, there is also a term arbitrarily selected by the applicant, and in this case, the meaning will be described in detail in the description of the invention. Therefore, terms used in the present disclosure should be defined based on the meaning of the term and the general content of the present disclosure, not simply the name of the term.

In the entire specification, when a part is said to "include" a certain component, it means that it may further include other components, not excluding other components unless otherwise stated. In addition, as described in the specification, "... wealth", "… A term such as “module” refers to a unit that processes at least one function or operation, and may be implemented as hardware or software or a combination of hardware and software.

Throughout the specification, 'providing' includes a process in which an object acquires specific information or directly or indirectly transmits/receives it to a specific object, and can be interpreted as comprehensively including the performance of related operations required in this process.

Hereinafter, with reference to the accompanying drawings, embodiments of the present disclosure will be described in detail so that those skilled in the art can easily carry out the present disclosure. However, the present disclosure may be implemented in many different forms and is not limited to the embodiments described herein.

Hereinafter, embodiments of the present disclosure will be described in detail with reference to the drawings.

1 is a block diagram showing the configuration of a device 100 according to an embodiment.

Referring to FIG. 1 , a device 100 may include a receiving unit 110, a processor 120, and a storage unit 130.

The receiving unit 110 may obtain anonymized anonymized data from original data. Here, anonymization can be understood as a concept including deletion. For example, 'anonymization' includes permanently deleting personal identification information or replacing all or part of the personal identification information with a unique identifier of the institution concerned. can do. Here, personal identification information may be understood as a concept encompassing an identifier capable of identifying an individual with only a single attribute, a quasi-identifier capable of identifying an individual through a combination of multiple attributes, and the like, and in one embodiment, identification information (e.g., resident number, mobile phone number), payment information (eg, purchase history), credit information (eg, credit rating), and the like.

In one embodiment, the original data may include personal identification information and may include a plurality of records defined through a plurality of attributes and composed of identifiable identification values. In one embodiment, the anonymous data includes an anonymized value obtained by anonymizing personal identification information of the original data, and a first record in which identification values of some attributes of at least one of a plurality of records constituting the original data are anonymized; may include a second record that is the remaining record from which is omitted. For example, in anonymous data, an individual can be identified with only a single attribute in the original data, and the values of the attribute corresponding to the identifier are removed. The values of some properties corresponding to may be in an anonymized state.

In one embodiment, the receiving unit 110 may receive pre-stored anonymous data from the storage unit 130 . In another embodiment, the receiving unit 110 may receive pre-stored original data from the storage unit 130, and the processor 120 may acquire anonymized data by performing a preset anonymization process on the original data. In one embodiment, the anonymization process may be performed based on a pre-stored K-anonymity algorithm. An anonymized first record may be obtained from the original data according to.

The processor 120 may determine a predicted value of an anonymous value included in the first record based on a comparison result between the first record and the second record included in the anonymous data. In one embodiment, the processor 120 compares and analyzes the identification value for each attribute between the first record and the second record according to a predetermined criterion, and the anonymous value whose value cannot be identified according to the K-anonymity algorithm (eg : The number of uses and the number of customers below the K value) can determine the predicted value that is estimated to be close to the actual value. Also, the processor 120 may update the first record so that the anonymous value is replaced with the predicted value. In this way, as a reliable predictive value is provided for anonymity-guaranteed data, the use value of anonymized data can be greatly improved.

Hereinafter, with further reference to FIGS. 2 to 6 , an embodiment in which the device 100 determines a prediction value for an anonymous value will be described in more detail.

2 is a flowchart illustrating a method of determining a predicted value for an anonymous value 12 by a device 100 according to an embodiment, and FIG. 3 is a flowchart illustrating a first record 10 and a second record ( 20) and a reference record 30.

Referring to FIGS. 2 to 3(a) , in step S210, the receiving unit 110 may obtain one or more first records 10 and a plurality of second records 20. For example, the receiving unit 110 may load one or more first records 10 and a plurality of second records 20 from the storage unit 130 based on a user input, and for another example, a processor ( One or more first records 10 and a plurality of second records 20 obtained from original data through the anonymization process by 120 may be received.

Here, the first record 10 includes a first identification value 11 and an anonymous value 12, and, for example, as described above, an identification value of some attribute for at least one of a plurality of records constituting the original data may correspond to anonymized records. The first identification value 11 means a value that can be identified in the first record 10, and is equal to, for example, an identification value for a corresponding attribute of an original record corresponding to the first record 10 in original data. In addition, the anonymized value 12 means a value anonymized from the original data according to a predetermined non-identification level (eg, K value), and, for example, the corresponding attribute of the original record corresponding to the first record 10 in the original data It may be a null value deleted from the identification value for , or a unique identifier replaced.

In addition, the second record 20 is composed of the second identification value 21, and, for example, as the anonymization condition is satisfied by the first record 10, the remainder of the plurality of records constituting the original data where anonymization is omitted. may correspond to the record. The second identification value 21 means a value that can be identified in the second record 20, and is equal to, for example, an identification value for a corresponding attribute of an original record corresponding to the second record 20 in original data. Also, throughout the specification, 'value' may be understood as a concept encompassing letters, symbols, combinations thereof, and the like, in addition to numbers.

Meanwhile, although the number of first records 10 is shown as plural in FIG. 3, it is not limited thereto and may be singular according to embodiments. In addition, when the number of first records 10 is plural, one of the plurality of first records 10 for which a prediction value is to be determined in the current step is determined using a preset method (eg, random, sequential, user input, etc.) After selecting according to, subsequent steps may proceed. Hereinafter, for convenience of description, it is assumed that an anonymous value 12 to be determined as a prediction value and a first record 10 including the anonymous value 12 are specified.

In one embodiment, the plurality of attributes defining each record may include at least one of the base year and month, metropolitan city and province, city, county, township, dong, type of business, gender, amount of use, and number of uses, but are not limited thereto. The various items belonging thereto may be used in various combinations. Also, in one embodiment, each of a plurality of attributes may correspond to a quasi-identifier capable of identifying an individual through a combination of some attributes, although a single attribute alone cannot identify an individual.

In one embodiment, the plurality of attributes may be divided into a first attribute used to determine one or more reference records 30 and a second attribute used to determine a predicted value for the anonymous value 12 . In one embodiment, the first attribute may correspond to a categorical variable and may be expressed as discrete data that may have only one of a plurality of preset values, for example, base year and month, metropolitan city, city, county, township, and township. , may include at least one of industry and gender. In addition, in one embodiment, the second attribute may correspond to a numerical variable and may be expressed as continuous data having a value within a preset numerical range, for example, at least one of the amount of use and the number of uses. can include

Referring to FIG. 3(b) , in step S220, the processor 120 selects a first of a plurality of second records 20 based on a comparison result between the first identification value 11 and the second identification value 22. One or more reference records 30 corresponding to the record 10 may be determined. Here, one or more reference records 30 represent records used to determine the prediction value of the anonymous value 12 of the first record 10 among a plurality of second records 20 .

In one embodiment, the processor 120 determines whether the first identification value 11 and the second identification value 22 are the same or similar to each other at a predetermined level or more for attributes equal to or greater than a preset number among a plurality of attributes. , one or more reference records 30 corresponding to the first record 10 among the plurality of second records 20 may be determined. For example, the anonymous value 12 (hereinafter, the first anonymous value '), as shown in FIG. 3 (b), a plurality of records having the same identification values for the base year and month, metropolitan city, city, county, and industry among a plurality of attributes are selected as a plurality of second records (20) can be filtered to determine one or more reference records (30).

In one embodiment, the second identification value 21 of each of the one or more reference records 30 may be different from the first identification value 11 only for one of a plurality of attributes. That is, records that differ in only one attribute and have identical identification values for the remaining attributes may be filtered and determined as one or more reference records 30 .

In one embodiment, each of the one or more reference records 30 may have a second identification value 21 different from the first identification value 11 only for one of a plurality of first attributes corresponding to a categorical variable. . For example, compared to the 1-1 record, if the values of the base year and month, metropolitan city, city, county, district, type of business, and gender are all the same, or if the values are different for at least one of them, it can be determined as the reference record 30 there is.

In one embodiment, each of the one or more reference records 30 consists of a third identification value 31 . Here, the third identification value 31 means a value that can be identified in the reference record 30, and is equal to, for example, an identification value for each attribute of a record corresponding to the first record 10 among the second records 20. .

In step S230, the processor 120 may determine a predicted value for the anonymous value 12 within the range of the non-identification level using one or more reference records 30. In one embodiment, the processor 120 may determine a predicted value for the anonymous value 12 based on a statistical analysis result for one or more reference records 30 . For example, when the K value applied to anonymization is 5, the processor 120 calculates the sum (eg, 12 = 7+5) of the third identification values 31 for the item (eg, the number of uses) to which the anonymization value 12 belongs. ), it is possible to determine a prediction value for the anonymous value 12 within the range of 0 or more and 4 binomial natural numbers that do not exceed the K value.

In one embodiment, the statistical analysis result for the reference record 30 is the number of reference records 30, the sum of the number of uses derived from the plurality of reference records 30 when the reference records 30 are plural, the reference record It may include the unit price per case derived from (30). For example, when the item to which the anonymous value 12 belongs is the number of uses, the processor 120 calculates the sum of the number of uses through an operation of summing the third identification value 31 for the number of uses of each reference record 30. The unit price per case may be calculated through an operation of dividing the third identification value 31 for the number of uses from the third identification value 31 for the amount of use of each reference record 30 .

In one embodiment, the statistical analysis result for the reference record 30 is obtained through various operations such as summation, average, weighted average, and median of third identification values 31 for a plurality of second attributes corresponding to numerical variables. can be obtained

In one embodiment, the processor 120 determines whether or not there is higher data for the first record 10, a statistical analysis result for one or more reference records 30, and at least one of the number of first records 10. , it is possible to determine the predicted value for the anonymous value (12).

In one embodiment, the processor 120 determines whether or not there is higher data for the first record 10, a statistical analysis result for one or more reference records 30, and at least one of the number of first records 10. , it is possible to determine the predicted value for the anonymous value (12). For example, when there is no top data for the first record 10, a predicted value for the anonymous value 12 is determined based on a statistical analysis result for one or more reference records 30, and the first record 10 If there is top data for , the top data for the first record 10, the statistical analysis result for one or more reference records 30, and the number of first records 10 are all used to determine the anonymous value 12. predictive values can be determined.

Hereinafter, with further reference to FIGS. 4 to 6 , first to third embodiments in which the device 100 determines a prediction value for an anonymous value will be described in detail.

FIG. 4 is a diagram for explaining an operation in which the device 100 according to an embodiment determines a prediction value for an anonymous value 12 according to the first embodiment. The first embodiment shows a method for the device 100 to determine a predicted value for an anonymous value 12 when there is no upper data for the first record 10 .

Referring to FIG. 4 , the processor 120 predicts an anonymous value 12 using a unit price per case determined based on one or more reference records 30 when there is no upper data for the first record 10. value can be determined. For example, if the upper data for the first record 10 is not stored in the storage 130, the processor 120 calculates the sum of the amount of use and the sum of the number of uses for the reference record 30. Then, the unit price per case in the reference record 30 is calculated by dividing the sum of the number of uses from the sum of the amount of use, and the anonymous value (12 ) can determine the predicted value for.

In one embodiment, the upper data for the first record 10 is based on original data having the same value as the first identification value 11 of the first record 10 for a predetermined number or more of a plurality of attributes. represent statistics. For example, the top data for the first record 10 is the number of uses of a record having the same values for the base year and month, metropolitan city, city, county, district, industry, and gender as those of the first record 10 among a plurality of records included in the original data. It may include the sum of, the sum of the amount of use, etc. (see FIG. 5 (a)).

For example, in the 1-1 record, the processor 120 determines the unit price per case of the group (2021.09, Seoul, Jung-gu, clothing, male) among the reference records 30 having the same attribute values such as the base year and month, region, and industry. It is calculated as 33,428 won (= 234,000 won / 7 cases) (refer to the 1st record in FIG. 3(b)), and 1.19 is calculated by dividing 33,428 won from 40,000 won in the 1-1 record. The predicted value of the first anonymous value may be estimated as one or two cases by estimating 1.19 within the natural number range, and may be determined as one case obtained by rounding off 1.19.

For example, for the anonymous value 12 (hereinafter referred to as 'second anonymous value') of the 1-2 records (the second record among the plurality of first records 10 shown in FIG. 3(a)), The processor 120 calculates 2.09 by dividing 33,428 won from 70,000 won of the usage amount of records 1-2 using 33,428 won as the unit price per case of the group (2021.09, Seoul, Jung-gu, clothing, male) calculated above. can do. The predicted value of the second anonymous value may be estimated as 2 or 3 cases by estimating 2.09 within the natural number range, and may be determined as 2 cases obtained by rounding 2.09.

For example, for the anonymous value 12 (hereinafter referred to as 'third anonymous value') of records 1-3 (the third record among the plurality of first records 10 shown in FIG. 3(a)), Among the base records (30) with the same attribute values such as base year and month, region, and industry, the unit price per case of the group (2021.09, Seoul, Jung-gu, clothing, female) is calculated as KRW 39,091 (= KRW 430,000 / 11 cases) (also 3(b), 2nd and 3rd records), 2.04 can be calculated by dividing 39,091 won from 80,000 won in records 1-3. The predicted value of the third anonymous value may be estimated as 2 or 3 cases by estimating 2.04 in the range of natural numbers, and may be determined as 2 cases obtained by rounding 2.04.

In the first embodiment, the processor 120 may determine a prediction value for the anonymous value 12 based on Equation 1 below based on the above-described example.

[Equation 1]

y = f ₁ ( x ₁ / x ₂ )

(Here, x ₁ and x ₁ are the sum of the usage amount of one or more records having the same identification value for the base year and month, metropolitan city, city, county, district, industry, and gender among the reference records 30 as the first record 10 and usage Each represents the sum of the cases, f ₁ ( ) represents a function that rounds to an internal value, and y represents a predicted value for an anonymous value (12)

In the first embodiment, when the number of one or more reference records 30 is plural, the processor 120 determines a plurality of unit prices per case based on each of the plurality of reference records 30, and for the plurality of unit prices per case An average or weighted average can be used to determine the predicted value for the anonymized value (12). For example, the processor 120 divides the number of uses from the amount of use for each of the reference records 30, and the unit price per case is 33,428 won (= 234,000 won / 7 cases), 46,667 won (= 280,000 won / 6 cases), 30,000 won (= KRW 150,000 / 5 cases), calculate the unit price per case as the average of these, KRW 36,698, and compare with the usage amount of KRW 40,000 in the case of the 1-1 record, and the usage amount in the case of the 1-2 record Compared with 70,000 won and compared with the usage amount of 80,000 won in the case of records 1-3, prediction values of the first to third anonymity values may be determined as 1 case, 2 cases, and 2 cases, respectively.

In the first embodiment, the processor 120 may determine a predicted value for the anonymity value 12 using a weighted average of a plurality of unit prices per case according to a weight given to a high number of uses. For example, as described above, the processor 120 calculates unit prices of 33,428 won, 46,667 won, and 30,000 won per reference record 30, respectively, and then assigns the largest first weight to 33,428 won, which has the largest number of uses (7 cases). After assigning a second weight smaller than the first weight to KRW 46,667 with 6 cases of use and assigning a third weight smaller than the second weight to KRW 30,000 with five cases of use, weighted average per case Based on the unit price, a predicted value for the anonymous value (12) can be determined.

In an embodiment, a weight given to each of the plurality of unit prices per case may be determined to be proportional to the size of the number of uses or the difference between the number of uses. In another embodiment, a weight assigned to each of a plurality of unit prices may be determined based on whether the value of a specific attribute is the same, and, for example, a higher weight may be assigned when the gender is the same. In another embodiment, weights assigned to each of a plurality of unit prices per case may be determined in proportion to the number of attributes having the same value.

5 is a diagram for explaining an operation in which the device 100 according to an embodiment determines a prediction value for an anonymous value 12 according to the second embodiment. The second embodiment shows one of the methods for the device 100 to determine the predicted value for the anonymous value 12 when there is higher data for the first record 10 .

Referring to FIG. 5 , when there is upper data for the first record 10 and there are a plurality of anonymous values 12, the processor 120 uses the total anonymous value representing the sum of the plurality of anonymous values 12 as a standard. A decision can be made based on the record (30). In one embodiment, the upper data for the first record 10 may be classified according to gender, as shown in FIG. 5(a). For example, the processor 120, when upper data for the first record 10 is stored in the storage unit 130 and the number of anonymous values 12 for the male attribute is two (e.g. : 1-1 to 1-2 records), the reference record (30 A total anonymity value of 2 can be calculated by subtracting 7 cases, which is the sum of the number of uses of the (2021.09, Seoul, Jung-gu, clothing, male) group in (2021.09, Seoul, Jung-gu, clothing, male). Accordingly, each of the plurality of anonymous values 12 may be determined within the range of the corresponding total anonymous value.

In the second embodiment, the processor 120 may determine each of the plurality of anonymous values 12 as 1 when the total anonymous value and the number of the plurality of anonymous values 12 are the same. According to the above example, since the total anonymity value 2 and the number of anonymity values 12 for the male attribute are equal to 2, the processor 120 calculates the first anonymous value a of record 1-1 and the number of anonymous values 12 of record 1-2. The second anonymous value b of each record may be determined to be 1.

In the second embodiment, if there is upper data for the first record 10 and the anonymous value 12 is one, the processor 120 determines the total anonymous value based on the reference record 30, and determines the anonymous value 12. The value (12) can be determined as a sum anonymous value. For example, the processor 120 may store upper data for the first record 10 in the storage 130 and the number of anonymous values 12 for the female attribute is 1, which is singular (e.g. : 1-3 records), in the 13 cases, which is the sum of the number of uses of the (2021.09, Seoul, Jung-gu, clothing, female) group in the upper data for the first record (10), in the reference record (30) (2021.09 , Seoul, Jung-gu, clothing, female) group, the total anonymity value of 2 can be calculated by subtracting 11 cases, which is the sum of the number of uses. Also, since the number of anonymity values 12 for the female attribute is 1, the processor 120 may determine the third anonymity value c of records 1-3 to be 2, which is equal to the total anonymity value.

FIG. 6 is a diagram for explaining an operation in which the device 100 according to an embodiment determines a prediction value for an anonymous value 12 according to a third embodiment. The third embodiment shows another one of methods for the device 100 to determine a predicted value for the anonymous value 12 when there is higher data for the first record 10 .

Referring to FIG. 6 , the processor 120, when there is upper data for the first record 10, there are a plurality of anonymous values 12, and the total anonymous value is greater than the number of the plurality of anonymous values 12, A combination of a plurality of anonymous values within the range of the total anonymous value may be determined. For example, if the number of anonymous values 12 for the male attribute is 2, which is plural (eg, records 1-1 and 1-2), in the upper data for the first record 10 ( 2021.09, Seoul, Jung-gu, Clothing, Male) group's total number of uses of 11 cases, in the reference record (30) (2021.09, Seoul, Jung-gu, Clothing, Male) group's total number of uses of 7 cases, A total anonymity value of 4 can be calculated. In this way, if the total anonymous value 4 is greater than the number 2 of the anonymous value 12, the combination of (first anonymous value, second anonymous value) within the range of the total anonymous value 4 is {(1, 3), (2 , 2), (3, 1)}.

In the third embodiment, the processor 120 may determine the plurality of anonymous values 12 according to any one of the combinations of the plurality of anonymous values using the unit price per case determined based on one or more reference records 30. there is. For example, the processor 120 calculates the unit price per case of the 2021.09, Seoul, Jung-gu, clothing, male) group as 33,428 won based on the reference record 30 in a manner similar to the first embodiment, and Among the combinations of a plurality of anonymous values for records 1-2 {(1, 3), (2, 2), (3, 1)}, the combination with the smallest difference from the unit price per case when each combination is applied (1, 3) may be detected, and prediction values for the first and second anonymous values (a, b) may be determined as 1 and 3, respectively, according to the detected combination of (1, 3).

In the third embodiment, the processor 120 may determine a prediction value for each of the plurality of anonymous values 12 using the first prediction value according to the first embodiment and the second prediction value according to the second embodiment. there is.

More specifically, the processor 120 may determine a first predicted value for each of the plurality of anonymous values 12 using the unit price per case determined based on one or more reference records 30 . For example, in a manner similar to the first embodiment, the unit price per case of the group (2021.09, Seoul, Jung-gu, clothing, male) among the reference records 30 having the same attribute values such as base year, month, region, and industry is 31,272 won (= 344,000 won / 11 cases), divide the usage amount of 40,000 won of the 1-1 record to 31,272 won, and estimate the first predicted value for the 1st anonymous value as 1 or 2 cases, The first prediction value for the second anonymity value may be determined as two or three cases by dividing the amount of use of the record from 70,000 won to 31,272 won.

Also, the processor 120 may determine a second prediction value for each of the plurality of anonymous values 12 by using a difference between the total anonymous value and the number of the plurality of anonymous values 12 . For example, in a manner similar to the second embodiment, in 11 cases, which is the sum of the number of uses of the (2021.09, Seoul, Jung-gu, clothing, male) group in the upper data for the first record 10, the reference record (30 ) in (2021.09, Seoul, Jung-gu, clothing, male) by subtracting 7 cases, which is the sum of the number of uses of the group, to calculate the total anonymity value 4, and sum the second predicted value for the first and second anonymity values. Within the range of value 4, it can be determined as {(1, 3), (2, 2), (3, 1)}, which is a combination of the first to second anonymous values.

Also, the processor 120 may determine prediction values for the plurality of anonymous values 12 using the first prediction value and the second prediction value. In an embodiment, the processor 120 may determine a prediction value for the anonymous value 12 according to a numerical value common to the first prediction value and the second prediction value. For example, for a first anonymity value a, the first prediction value is 1 or 2, and the second prediction value is (a, b) = {(1, 3), (2, 2), (3, 1 )}, it can be assumed that it is a (1, 3) or (2, 2) combination among {(1, 3), (2, 2), (3, 1)} combinations. Also, for the second anonymity value b, the first prediction value is 2 or 3, and the second prediction value is (a, b) = {(1, 3), (2, 2), (3, 1)} Since it is 3 to 1 depending on , it can be estimated that it is a (1, 3) or (2, 2) combination among {(1, 3), (2, 2), (3, 1)} combinations.

In the third embodiment, the processor 120 determines, when there are a plurality of combinations of common prediction values between the first prediction value and the second prediction value, the composition ratio of the first record 10 to the amount of use and the combination of the number of uses. A plurality of anonymous values 12 may be determined according to any one of a plurality of common predictive value combinations based on the star composition ratio. Here, the composition ratio to the amount of use represents the ratio of the amount of use of each first record 10 to the sum of the amount of use of the plurality of first records 10, and the composition ratio for each combination of the number of uses is the sum of the anonymous represents the ratio of each anonymous value (12) to the value. For example, when the combination of the first and second anonymous values (a, b) is estimated as a combination of (1, 3) or (2, 2), the composition ratio to the amount of use of the 1-1 record is 0.36 (= 40,000 / (40,000 + 70,000)), the composition ratio to the amount of use of the 1-2 records can be calculated as 0.64 (= 70,000 / (40,000 + 70,000)). In addition, the composition ratio for each combination of the number of uses of the 1-1 record is 0.25 (= 1 for each of a and b for the combination of (1, 3) among the common combinations between the first prediction value and the second prediction value) / (1 + 3)), 0.75 (=3 / (1 + 3)), and for the (2, 2) combination, 0.5 for each of a and b (=2 / (2 + 2)), 0.5 It can be calculated as (=2 / (2 + 2)).

In the third embodiment, the processor 120 may determine predicted values for the plurality of anonymous values 12 such that a difference between the composition ratio for the amount of use and the composition ratio for each combination of the number of uses is minimized. For example, in the case of the (1, 3) combination, the difference between the composition ratio for the amount of use and the composition ratio for each combination of the number of uses is 0.11 (= |0.36 - 0.25|), 0.11 (= | 0.64 - 0.75|), and for the (2, 2) combination, 0.14 (= |0.36 - 0.5|) and 0.14 (= |0.64 - 0.5|), so the difference is smaller depending on the (1, 3) combination Prediction values for the first and second anonymity values a and b may be determined to be one and three, respectively.

In the third embodiment, the processor 120 may determine a prediction value for the anonymous value 12 based on Equation 2 below based on the above-described example. For example, the processor 120 calculates z for each combination of a plurality of prediction values common between the first prediction value and the second prediction value, and the sum of z values calculated for each of a and b among the plurality of combinations is A predicted value for an anonymous value (12) can be determined according to one of the smallest combinations.

[Equation 2]

z ₁ = x ₄ / x ₃

z ₂ = x ₆ / x ₅

z = f ₂ ( z ₁ - z ₂ )

(Here, z ₁ represents the composition ratio of the first record 10 to the amount of money used, x ₃ represents the sum of the amount of money used by a plurality of first records 10, and x ₄ represents each first record ( 10) In addition, z ₂ represents the composition ratio of each combination of the number of uses, x ₅ represents the total anonymity value, and x ₆ represents each anonymity value 12. In addition, z represents the use Represents the difference between the composition ratio for the amount and the composition ratio for each combination of the number of uses, and f ₂ ( ) indicates a function that converts the internal value into an absolute value)

According to an embodiment of the present invention, it is possible to provide a prediction value estimated to be close to the actual value for the anonymous value 12 whose value cannot be identified in order to satisfy the anonymity condition. By providing reliable prediction values for guaranteed data, the value of use as data of anonymous data can be greatly improved.

The receiving unit 110 according to an embodiment is connected to components of the device 100 or other devices (eg, a terminal, a server) through a network or electrically to transmit and receive various information described throughout the specification through a wired/wireless connection. A communication device may be included. Here, the network may be configured through various communication networks such as wired and wireless, for example, a local area network (LAN), a metropolitan area network (MAN), and a wide area network (WAN). Network) can be composed of various communication networks.

The processor 120 according to an embodiment may perform a series of operations for determining a prediction value for an anonymous value, and is electrically connected to the receiver 110, the storage unit 130, and other components to obtain these It is possible to control the flow of data between the devices, and for this purpose, it may be implemented as a central processor unit (CPU) that controls the overall operation of the device 100 .

The storage unit 130 according to an embodiment may store data used throughout the operation of the device 100, and may store, for example, the first record 10 and the second record 20. In one embodiment, the storage unit 130 may store original data about the customer's personal information or anonymized data from the original data, and may store a prediction value determined through the above-described operation. In one embodiment, the storage unit 130 may be implemented in various forms such as a memory or a database, and may be implemented as a cloud or a separate storage server to provide data and storage space necessary for the device 100 through a wired or wireless communication network. may be

In addition, those skilled in the art can understand that other general-purpose components other than the components shown in FIG. 2 may be further included in the device 100 . For example, the device 100 may further include an input/output interface for receiving user input or outputting information. In addition, the device 100 corresponds to a computing device capable of determining a predicted value for an anonymous value, and in an embodiment, is implemented as a server such as a computer that operates through a computer program for realizing the functions described in this specification. It can be.

The order and combination of the steps shown above is an embodiment, and the order, combination, branch, function, and subject of execution thereof may be added, omitted, or modified within the scope that does not deviate from the essential characteristics of each component described in the specification. It can be seen that this can be done.

Various embodiments of the present disclosure are implemented as software including one or more instructions stored in a storage medium (eg, memory) readable by a machine (eg, a display device or a computer). It can be. For example, a processor (eg, processor 120 ) of the device may call at least one of one or more instructions stored from a storage medium and execute it. This enables the device to be operated to perform at least one function according to the invoked at least one instruction. The one or more instructions may include code generated by a compiler or code executable by an interpreter. The device-readable storage medium may be provided in the form of a non-transitory storage medium. Here, 'non-temporary' only means that the storage medium is a tangible device and does not contain signals (e.g., electromagnetic waves), and this term refers to the case where data is stored semi-permanently in the storage medium. It does not discriminate when it is temporarily stored.

According to one embodiment, the method according to various embodiments disclosed in the present disclosure may be included and provided in a computer program product. Computer program products may be traded between sellers and buyers as commodities. A computer program product is distributed in the form of a device-readable storage medium (e.g. compact disc read only memory (CD-ROM)), or through an application store (e.g. Play Store™) or on two user devices (e.g. It can be distributed (eg downloaded or uploaded) online, directly between smartphones. In the case of online distribution, at least part of the computer program product may be temporarily stored or temporarily created in a device-readable storage medium such as a manufacturer's server, an application store server, or a relay server's memory.

The description of the present disclosure described above is for illustrative purposes, and those skilled in the art can understand that it can be easily modified into other specific forms without changing the technical spirit or essential features of the present disclosure. will be. Therefore, the embodiments described above should be understood as illustrative in all respects and not limiting. For example, each component described as a single type may be implemented in a distributed manner, and similarly, components described as distributed may be implemented in a combined form.

The scope of the present disclosure is indicated by the following claims, and all changes or modifications derived from the meaning and scope of the claims and equivalent concepts thereof should be construed as being included in the scope of the present disclosure.

100: device
110: receiver 120: processor
10: first record
11: first identification value 12: anonymous value
20: second record 21: second identification value
30: reference record 31: third identification value

Claims (15)

A method for a device to determine a predicted value for an anonymous value,
obtaining a first record including an anonymized anonymized value and an identifiable first identification value and a plurality of second records composed of a second identification value from original data according to a preset de-identification level;
determining one or more reference records corresponding to the first record among the plurality of second records, based on a comparison result between the first identification value and the second identification value; and
Determining a predicted value for the anonymous value within the range of the non-identification level by using the one or more reference records.
According to claim 1,
The step of determining the predicted value is
The method of determining the prediction value based on at least one of the presence or absence of higher data for the first record, a statistical analysis result for the one or more reference records, and the number of the first records.
According to claim 1,
The step of determining the predicted value is
If there is no upper data for the first record, determining the prediction value using a unit price per case determined based on the one or more reference records.
According to claim 3,
Each of the one or more reference records is
wherein the second identification value differs from the first identification value only for one of a plurality of attributes defining each record.
According to claim 3,
The step of determining the predicted value is
When the number of the one or more reference records is plural, determining a unit price per case based on each of the plurality of reference records;
The method of determining the predicted value using an average or a weighted average value for the plurality of unit costs per case.
According to claim 1,
The step of determining the predicted value is
When there is upper data for the first record and there are a plurality of anonymous values,
determining a total anonymous value representing a sum of the plurality of anonymous values based on the reference record; and
and determining each of the plurality of anonymous values as 1 when the total anonymous value and the number of the plurality of anonymous values are the same.
According to claim 1,
The step of determining the predicted value is
When there is upper data for the first record and the anonymous value is 1,
determining a total anonymous value representing a sum of the plurality of anonymous values based on the reference record; and
determining the anonymous value as the total anonymous value.
According to claim 1,
The step of determining the predicted value is
When there is upper data for the first record and there are a plurality of anonymous values,
determining a total anonymous value representing a sum of the plurality of anonymous values based on the reference record; and
determining a combination of the plurality of anonymous values within a range of the sum anonymous value when the total anonymous value is greater than the number of the plurality of anonymous values; and
and determining the plurality of anonymous values according to one of the combinations of the plurality of anonymous values using the unit price per case determined based on the one or more reference records.
According to claim 8,
The step of determining the plurality of anonymous values
determining a first predicted value for each of the plurality of anonymous values using a unit price per case determined based on the one or more reference records;
determining a second prediction value for each of the plurality of anonymous values using a difference between the total anonymous value and the number of the plurality of anonymous values; and
And determining a prediction value for each of the plurality of anonymous values using the first prediction value and the second prediction value.
In a device for determining a predicted value for an anonymous value,
a receiving unit that obtains a first record including an anonymized anonymized value and an identifiable first identification value and a plurality of second records composed of second identification values from original data according to a predetermined de-identification level; and
Based on a comparison result between the first identification value and the second identification value, one or more reference records corresponding to the first record among the plurality of second records are determined;
A processor for determining a predicted value for the anonymous value within a range of the non-identification level by using the one or more reference records.
According to claim 10,
The processor
The device of determining the prediction value based on at least one of the presence or absence of higher data for the first record, a statistical analysis result for the one or more reference records, and the number of the first records.
According to claim 10,
The processor
If there is no upper data for the first record, determining the predicted value using a unit price per case determined based on the one or more reference records.
According to claim 10,
The processor
When there is upper data for the first record and there are a plurality of anonymous values,
Determine a total anonymous value representing the sum of the plurality of anonymous values based on the reference record;
When the total anonymous value and the number of the plurality of anonymous values are the same, determining each of the plurality of anonymous values as 1.
According to claim 10,
The processor
When there is upper data for the first record and there are a plurality of anonymous values,
Determine a total anonymous value representing the sum of the plurality of anonymous values based on the reference record;
When the sum anonymous value is greater than the number of the plurality of anonymous values, determining a combination of the plurality of anonymous values within a range of the sum anonymous value;
The device of determining the plurality of anonymous values according to any one of the combinations of the plurality of anonymous values using the unit price per case determined based on the one or more reference records.
A computer-readable recording medium recording a program for executing the method of any one of claims 1 to 9 in a computer.

Images