A data processing method and system for anonymous person attribute abundance optimization

Abstract

The present application relates to a kind of data processing method and system for anonymous personnel attribute abundance optimization, it is related to data processing field, by obtaining first sequence to the privacy ordering of the investigation attribute set based on pre-defined privacy level table, obtain multiple groups of zero-knowledge feature value set using zero-knowledge processing data sandbox processing candidate personnel attribute characteristic value set, then according to the variance ordering of these feature value sets, obtain second sequence, combined with two sequences, sequence number is added and sorted to obtain third sequence, and finally according to third sequence, attribute abundance optimization is realized to realize personnel sorting, it solves the technical problem that it is difficult to balance screening efficiency and privacy protection when personnel screening is carried out to group in special scene, effectively balances screening efficiency and privacy protection, both improve the accuracy of personnel screening, and significantly reduce the risk of privacy leakage.

Description

Technical Field

[0001] This invention relates to the field of data processing, and in particular to a data processing method and system for optimizing the abundance of attributes of anonymous individuals. Background Technology

[0002] In today's digital age, data has become a crucial resource driving social progress and development. Many applications, such as security checks, disease screening, market research, and data mining, often require screening specific groups to identify target individuals or group characteristics. However, this process inevitably involves the processing and use of personal privacy information, raising concerns about privacy protection.

[0003] Traditional screening methods often rely on direct access to and comparison of large amounts of personal information. While this approach can improve screening accuracy to some extent, it also introduces serious risks of privacy breaches. Especially in specific scenarios, such as those involving security, public health, or sensitive commercial information, the consequences of privacy breaches can be even more severe, not only infringing on individual legal rights but also potentially threatening social stability and security. Theoretically, increasing the number of attributes used for screening can improve accuracy, as more attribute information can more comprehensively describe individual characteristics, thus more accurately identifying target individuals. However, increasing the number of attributes also means processing more personal information, which undoubtedly increases the risk of privacy breaches. Therefore, how to effectively reduce the risk of privacy breaches while ensuring screening efficiency has become a pressing technical problem to be solved. Furthermore, with the continuous improvement of data protection regulations and the increasing public awareness of privacy, how to process and screen data under compliant conditions has also become an important consideration. Summary of the Invention

[0004] This invention addresses the technical problem in existing technologies where it is difficult to balance screening efficiency and privacy protection when screening groups in special scenarios. It provides a data processing method and system for optimizing the abundance of anonymous personal attributes to solve this problem.

[0005] The technical solution of the present invention to solve the above-mentioned technical problems is as follows:

[0006] In a first aspect, the present invention provides a data processing method for optimizing the abundance of anonymous personnel attributes, comprising: sorting a set of screening attributes from smallest to largest privacy level based on a predefined privacy level table to obtain a first screening attribute sequence; sending the screening attribute set to a zero-knowledge processing data sandbox, processing multiple screening attribute monitoring feature value sets of multiple candidate personnel to obtain a first screening attribute zero-knowledge feature value set up to the Qth screening attribute zero-knowledge feature value set; and processing the first screening attribute zero-knowledge feature value set up to the Qth ... The zero-knowledge feature value variances of the feature value set are sorted in descending order to obtain a second screening attribute sequence. The first and second screening attribute sequences are summed by their corresponding attribute indices to obtain a first screening attribute indices up to the Qth screening attribute indices. The screening attribute set is then sorted in ascending order according to the sums of the first and Qth screening attribute indices to obtain a third screening attribute sequence. Attribute abundance is optimized according to the third screening attribute sequence to achieve the selection of the multiple candidate personnel.

[0007] Secondly, the present invention provides a data processing system for optimizing the abundance of anonymous personnel attributes. The system includes: a first sorting module, used to sort a set of screening attributes from smallest to largest privacy level based on a predefined privacy level table to obtain a first screening attribute sequence; a numerical processing module, used to send the screening attribute set to a zero-knowledge processing data sandbox to process multiple screening attribute monitoring feature value sets of multiple candidate personnel to obtain a first screening attribute zero-knowledge feature value set up to the Qth screening attribute zero-knowledge feature value set; and a second sorting module, used to sort the first screening attribute zero-knowledge feature value set up to the Qth screening attribute... The zero-knowledge feature value variances of the zero-knowledge feature value set are sorted in descending order to obtain a second screening attribute sequence; a summation processing module is used to sum the same attribute indices according to the first screening attribute sequence and the second screening attribute sequence to obtain the first screening attribute index sum up to the Qth screening attribute index sum; a third sorting module is used to sort the screening attribute set in ascending order according to the first screening attribute index sum up to the Qth screening attribute index sum to obtain a third screening attribute sequence; an abundance optimization module is used to optimize the attribute abundance according to the third screening attribute sequence to achieve the selection of the multiple candidate personnel.

[0008] The beneficial effects of this invention are as follows: by performing privacy sorting on the screening attribute set based on a predefined privacy level table to obtain a first sequence, using a zero-knowledge processing data sandbox to process the attribute feature value set of the candidate personnel to obtain multiple sets of zero-knowledge feature value sets, and then sorting these feature value sets according to their variance to obtain a second sequence, combining the two sequences and sorting by their sequence numbers to obtain a third sequence, and finally optimizing the attribute abundance based on the third sequence to achieve personnel sorting, this invention effectively balances screening efficiency and privacy protection, improving the accuracy of personnel screening while significantly reducing the risk of privacy leakage. Attached Figure Description

[0009] Figure 1 This is a flowchart illustrating a data processing method for optimizing the abundance of anonymous personal attributes, provided by the present invention.

[0010] Figure 2 This is a schematic diagram of the structure of a data processing system for optimizing the abundance of anonymous personal attributes, provided by the present invention.

[0011] Explanation of reference numerals in the attached figures: First sorting module 11, numerical processing module 12, second sorting module 13, summation processing module 14, third sorting module 15, abundance optimization module 16. Detailed Implementation

[0012] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0013] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0014] In the description of this invention, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed herein.

[0015] Example 1:

[0016] like Figure 1 As shown, this embodiment of the invention provides a data processing method for optimizing the abundance of anonymous personal attributes, including:

[0017] S10: Based on a predefined privacy level table, sort the set of investigation attributes from smallest to largest according to privacy requirements to obtain the first investigation attribute sequence.

[0018] For example, a predefined privacy level table assigns corresponding privacy levels to different screening attributes. The privacy level quantifies the degree of privacy involved in each screening attribute; the smaller the privacy level, the lower the privacy of the screening attribute. Based on the predefined privacy level table, for a given set of screening attributes, the privacy level corresponding to each screening attribute needs to be extracted and compared. Specifically, each screening attribute in the set is examined sequentially, its corresponding privacy level is located in the privacy level table, and recorded. Subsequently, the screening attributes in the set are rearranged according to the privacy level ranking rule from smallest to largest. For example, assuming the set of screening attributes includes three attributes: "name," "age," and "health status," in the privacy level table, "age" has a privacy level of 2, "name" has a privacy level of 1, and "health status" has a privacy level of 3. Therefore, after sorting according to privacy level from smallest to largest, the first screening attribute sequence obtained is "name," "age," and "health status." Through this series of processes, a first screening attribute sequence is finally obtained, arranged from low to high privacy level, providing an orderly screening attribute basis for subsequent privacy protection decisions or data processing operations.

[0019] S20: Send the set of investigation attributes to the zero-knowledge processing data sandbox, process the monitoring feature value sets of multiple investigation attributes of multiple candidates, and obtain the first zero-knowledge feature value set of investigation attributes up to the Qth zero-knowledge feature value set of investigation attributes.

[0020] Preferably, in the data processing workflow, the pre-organized attribute set for investigation needs to be transferred to the zero-knowledge processing data sandbox. The attribute set for investigation contains a set of multiple attribute categories to be examined; while the zero-knowledge processing data sandbox is an environment with special security mechanisms. When data is processed in this environment, the data is always encrypted and the original data content is not exposed, thereby ensuring data privacy and security.

[0021] After the set of screening attributes is submitted to the sandbox, the sandbox will perform processing operations on multiple candidate individuals separately. Each candidate individual corresponds to a set of monitoring feature values ​​for multiple screening attributes, i.e., multiple monitoring feature value sets for multiple candidate individuals. Suppose that the feature value set of candidate individual A includes features such as "income level: 8000 yuan", "education level: bachelor's degree", and "occupation type: engineer". The sandbox uses zero-knowledge proof-related methods to process these feature value sets, with the aim of extracting feature information related to the screening attributes without disclosing the original information of the specific feature values.

[0022] After sandboxing, a corresponding zero-knowledge feature value set is generated for each investigation attribute. Specifically, for the first investigation attribute, a first investigation attribute zero-knowledge feature value set is generated; for the second investigation attribute, a second investigation attribute zero-knowledge feature value set is generated, and so on, until a Q-th investigation attribute zero-knowledge feature value set is generated. These zero-knowledge feature value sets retain the key information from the original feature values ​​that can be used for subsequent analysis or judgment, while avoiding the direct exposure of the original data, providing a safe and reliable data foundation for subsequent data analysis and privacy protection decisions.

[0023] S30: Sort the investigation attribute set in descending order of the variance of the zero-knowledge feature values ​​from the first investigation attribute zero-knowledge feature value set to the Qth investigation attribute zero-knowledge feature value set to obtain the second investigation attribute sequence.

[0024] Furthermore, the zero-knowledge feature set is a set of feature values ​​extracted for each investigation attribute after zero-knowledge processing, which does not expose the original data but retains key information. After obtaining a series of zero-knowledge feature sets from the first investigation attribute to the Qth investigation attribute, variance is calculated for these zero-knowledge feature sets. Variance is a statistical indicator used to measure the dispersion of a set of data. The larger the variance, the greater the fluctuation and the higher the dispersion of the data; conversely, the smaller the variance, the smaller the fluctuation and the lower the dispersion. For each investigation attribute's corresponding zero-knowledge feature set, its variance value is calculated separately. The purpose of variance calculation is to measure the data dispersion of each investigation attribute's zero-knowledge feature set, reflecting the fluctuation of different investigation attribute data through the size of the variance. Then, the investigation attribute sets are sorted according to the order of variance from largest to smallest, providing an ordered basis for subsequent analysis and decision-making based on data dispersion characteristics, and helping to more accurately grasp the differences in characteristics of different investigation attributes at the data level. For example, suppose there are three screening attributes with corresponding variances of zero-knowledge feature value sets: the first screening attribute has a variance of 5.2, the second screening attribute has a variance of 3.8, and the third screening attribute has a variance of 7.1. After obtaining the variances of each zero-knowledge feature value set, the screening attribute sets are reordered according to the order of variance from largest to smallest, i.e., the screening attributes with larger variance values ​​are listed first, and the screening attributes with smaller variance values ​​are listed later. Following the variance values ​​in the example above, the resulting second screening attribute sequence is: third screening attribute, first screening attribute, second screening attribute. Through this process, a second screening attribute sequence arranged in descending order of zero-knowledge feature value variance is obtained, providing an ordered basis for subsequent analysis or decision-making based on data dispersion.

[0025] S40: Summing the same attribute numbers according to the first investigation attribute sequence and the second investigation attribute sequence to obtain the sum of the first investigation attribute numbers up to the Qth investigation attribute number sum.

[0026] Specifically, two investigation attribute sequences have been obtained according to different rules: the first investigation attribute sequence and the second investigation attribute sequence. The first investigation attribute sequence is obtained by sorting the investigation attribute set in ascending order of privacy based on a predefined privacy level table; while the second investigation attribute sequence is obtained by sorting the investigation attribute set in descending order of variance of the zero-knowledge feature value set of each investigation attribute. Next, the investigation attributes in the same position in these two sequences need to be summed by their ordinal numbers. Specifically, starting from the first investigation attribute in each sequence, find its corresponding ordinal number, and then add these two ordinal numbers to obtain the first investigation attribute ordinal number sum. Then, perform the same operation on the second investigation attribute to obtain the second investigation attribute ordinal number sum. And so on, until the ordinal number sum is completed for the Qth investigation attribute to obtain the Qth investigation attribute ordinal number sum. For example, suppose the first screening attribute sequence is "age, income, occupation", and the second screening attribute sequence is "occupation, age, income". Matching the screening attributes in these two sequences one-to-one, age is 1 in the first sequence and 2 in the second sequence, with a sum of 3; income is 2 in the first sequence and 3 in the second sequence, with a sum of 5; occupation is 3 in the first sequence and 1 in the second sequence, with a sum of 4. Through this process, a series of results are obtained, from the sum of the first screening attribute numbers to the sum of the Qth screening attribute numbers. The purpose is to integrate the positional information of screening attributes under different ranking rules, merging and quantifying the results of privacy-based ranking and zero-knowledge feature value variance ranking. The sum of the numbers reflects the comprehensive priority or correlation characteristics of each screening attribute under the two ranking dimensions, providing more comprehensive data support for subsequent analysis and decision-making based on multi-dimensional ranking information, and helping to more comprehensively and accurately grasp the characteristics and importance of the screening attributes.

[0027] S50: Sort the set of investigation attributes from smallest to largest according to the sum of the first investigation attribute numbers up to the sum of the Qth investigation attribute numbers, and obtain the third investigation attribute sequence.

[0028] Optionally, a series of results have been obtained, from the sum of the first investigation attribute indices to the sum of the Qth investigation attribute indices. The sum of the first to Qth investigation attribute indices is obtained by summing the investigation attribute indices at the same position in the first and second investigation attribute sequences. For example, assuming there are three investigation attributes, with indices 1, 2, and 3 in the first investigation attribute sequence and 2, 1, and 3 in the second investigation attribute sequence, then the corresponding sum of the first investigation attribute indices is 3(1 + 2), the sum of the second investigation attribute indices is 3(2 + 1), and the sum of the third investigation attribute indices is 6(3 + 3).

[0029] Next, the attribute set is reordered based on the sum of its serial numbers. Specifically, the attributes are arranged in ascending order of their sum, with attributes having smaller sums appearing first and those with larger sums appearing later. Following the example above, since the sums of the first and second attributes are both 3 and less than the sum of the third attribute (6), further differentiation can be achieved using other rules (such as the original order of the attributes in a sequence). Here, we assume priority is given to sorting by the sum of serial numbers, and in case of a tie, priority is given to attributes with smaller serial numbers in the first sequence. The resulting third attribute sequence is "Age, Income, Occupation" (assuming age, income, and occupation are the three aforementioned attributes and their serial numbers in the first sequence are 1, 2, and 3, respectively).

[0030] Through this process, a third set of investigation attributes was finally obtained, arranged in ascending order of the sum of the serial numbers, providing an orderly basis for subsequent analysis or decision-making based on the comprehensive ranking results.

[0031] S60: Optimize the attribute abundance according to the third screening attribute sequence to achieve the sorting of the multiple candidate personnel.

[0032] Specifically, attribute abundance is optimized according to the third screening attribute sequence. Attribute abundance can be understood as the richness of information contained in a screening attribute and its contribution to personnel sorting. Attribute abundance optimization aims to reassess and adjust the importance of each screening attribute in the personnel sorting process based on the third screening attribute sequence, in order to improve the accuracy and effectiveness of sorting.

[0033] Specifically, following the order of the third screening attribute sequence, starting with the first attribute in the sequence, the distribution and information difference of each screening attribute among multiple candidates are analyzed sequentially. For example, assuming the third screening attribute sequence is "age, occupation, income," the distribution of the "age" attribute among candidates is analyzed first. If the age distribution is relatively concentrated, with little difference in the number of candidates in different age groups, then the contribution of this attribute to the selection may be relatively low. If the age distribution is relatively dispersed, with significant differences in the number of people in different age groups, then the attribute abundance is high, and it should be given greater weight in the selection. Next, the "occupation" attribute is analyzed, counting the number of candidates in different occupation types. If the number of people in certain occupation types is extremely small, it may not have much distinguishing effect on the selection results, and the weight of this attribute can be appropriately reduced. Conversely, if the occupation types are widely distributed and have a significant distinguishing effect on personnel characteristics, then its weight should be increased. Finally, a similar analysis is performed on the "income" attribute.

[0034] By optimizing and adjusting the abundance of each screening attribute, the sorting model or rules are reconstructed. The optimized attribute weights are then applied to the sorting process of multiple candidates. Based on the characteristic values ​​of the candidates on each screening attribute, combined with the optimized weights, a comprehensive evaluation and ranking are performed to achieve the sorting of candidates. This makes the sorting results more in line with actual needs and improves the accuracy and rationality of the sorting.

[0035] In a preferred embodiment, based on a predefined privacy level table, the screening attribute set is sorted from smallest to largest privacy level to obtain a first screening attribute sequence, including: configuring a preset screening attribute set via a user terminal; extracting a first preset screening attribute from the preset screening attribute set, retrieving the first common recording frequency between personnel identity information and the first preset screening attribute to obtain a first exposure index, calculating a first privacy level with k as the base and the first exposure index as the exponent, where k is an integer and k≥2; up to extracting a Yth preset screening attribute from the preset screening attribute set, retrieving the first common recording frequency between personnel identity information and the Yth preset screening attribute to obtain a Yth exposure index, calculating a Yth privacy level with k as the base and the Yth exposure index as the exponent, where k is an integer and k≥2; constructing the privacy level table based on the first privacy level up to the Yth privacy level.

[0036] Optionally, in constructing a privacy level table to determine the privacy of the investigated attributes, a preset set of investigated attributes is first configured using a user terminal. This set covers multiple attributes to be examined, such as "phone number," "home address," and "consumption preferences." The user terminal refers to the terminal device or software system that initiates data investigation-related operations and configures investigated attributes. Next, attributes are extracted sequentially from the preset set of investigated attributes for processing. Taking the extraction of a first preset investigated attribute as an example, the frequency of records where the person's identity information and the first preset investigated attribute co-occur is retrieved in a database or related system. This frequency reflects the proportion of records containing both the attribute and the person's identity information in the total number of records. A specific algorithm is used to convert this frequency into a first exposure index, which quantifies the degree of exposure of the attribute in the person's information.

[0037] Specifically, the process of converting frequency into an exposure index aims to quantify the degree of exposure when screening attributes and personnel identification information co-occur. First, the co-recording frequency is determined by retrieving the number of records in the database or system where personnel identification information and a specific screening attribute co-occur, and calculating the proportion of these records to the total number of records. For example, if the total number of records is 1000, and 200 of them contain records where "phone number" and personnel identification information co-occur, then the co-recording frequency is 200 / 1000 = 0.2. The exposure index maps the frequency to a more sensitive index space through mathematical transformation. The formula is: Exposure Index = 1 / (1 - Co-recording Frequency). When the co-recording frequency is close to 1 (i.e., the attribute is almost necessarily associated with identification information), the exposure index approaches infinity, indicating extremely high exposure risk; when the co-recording frequency is close to 0 (the attribute and identification information are almost unrelated), the exposure index approaches 1, indicating extremely low exposure risk. If the common recording frequency is 1 (complete exposure), the exposure index is directly set to the preset maximum value (e.g., 1000) to avoid division by zero errors; if the frequency is 0 (no exposure), the exposure index is set to 1, indicating the lowest risk. The privacy level is calculated using the exposure index (base k, exponent of the exposure index), which further quantifies attribute privacy. A higher privacy level indicates a greater risk of attribute exposure in the data, requiring higher priority for privacy protection in subsequent processing.

[0038] Furthermore, using an integer k (k≥2) as the base and the first exposure index as the exponent, the first privacy level is calculated. For example, if k is 2 and the first exposure index is 3, then the first privacy level is 2. 3 = 8. Following the same method, continue extracting subsequent preset screening attributes from the preset screening attribute set until the Yth preset screening attribute is extracted. Similarly, retrieve the first common record frequency between the personnel identity information and the Yth preset screening attribute to obtain the Yth exposure index. Using k as the base and the Yth exposure index as the exponent, calculate the Yth privacy level. For example, if k is 2 and the Yth exposure index is 4, then the Yth privacy level is 2. 4 = 16.

[0039] After calculating the privacy level for all preset screening attributes in the preset screening attribute set, a privacy level table is constructed based on the calculated first privacy level up to the Yth privacy level. This table associates and records each preset screening attribute with its corresponding privacy level. Finally, based on this privacy level table, the screening attribute set is sorted in ascending order of privacy to obtain the first screening attribute sequence, providing a basis for subsequent privacy-based data processing or decision-making.

[0040] In a preferred embodiment, constructing the privacy level table based on the first privacy level up to the Yth privacy level includes: calculating the proportion of the same privacy level among the first privacy level up to the Yth privacy level to obtain the first privacy level reconstruction probability; when the first privacy level reconstruction probability is less than the reconstruction probability threshold, outputting the privacy level table; when the first privacy level reconstruction probability is greater than or equal to the reconstruction probability threshold: traversing the non-first preset investigation attribute set of the preset investigation attribute set, calculating the first extended common recording frequency set with the first preset investigation attribute; adding the first common recording frequency to the first extended common recording frequency set and performing mean calculation to obtain the first reconstruction exposure index, calculating the first reconstruction privacy level with k as the base and the first reconstruction exposure index as the exponent, where k is an integer, k≥2; until the Yth reconstruction privacy level is obtained; constructing the privacy level table based on the first reconstruction privacy level up to the Yth reconstruction privacy level.

[0041] Furthermore, in constructing the privacy level table, it is necessary to calculate the proportion of identical privacy levels from the first privacy level to the Yth privacy level to obtain the reconstruction probability of each privacy level. Specifically, for each privacy level, the number of times it appears among all privacy levels is counted, and then divided by the total number of privacy levels Y to obtain the reconstruction probability of that privacy level. For example, if there are 5 privacy levels, and privacy level 3 appears 2 times, then its reconstruction probability is 2 / 5 = 0.4, which is the reconstruction probability of the first privacy level (the calculation method is illustrated here using privacy level 3 as an example).

[0042] Subsequently, the reconstruction probability is compared with a preset reconstruction probability threshold. If the reconstruction probability of the first privacy level is less than the reconstruction probability threshold, it indicates that the current privacy levels are relatively dispersed and the differences in privacy risks are significant. The privacy level table at this point can be directly output, recording the privacy level corresponding to each preset screening attribute. However, when the reconstruction probability of the first privacy level is greater than or equal to the reconstruction probability threshold, it indicates that multiple preset screening attributes have the same or similar privacy levels, resulting in insufficient differentiation of privacy risks. In this case, the privacy levels need to be reconstructed.

[0043] The privacy level reconstruction process begins with the first preset screening attribute. It then iterates through the set of non-first preset screening attributes (excluding the first preset screening attribute) within the preset screening attribute set, and calculates the first extended common recording frequency set of these non-first preset screening attributes and the first preset screening attribute. This set contains the frequency information of these attribute pairs appearing together with the first preset screening attribute. Next, the initially obtained first common recording frequencies are added to the first extended common recording frequency set, and the average of all frequency values ​​in this set is calculated to obtain the first reconstruction exposure index. This index reflects the degree of exposure after considering more related attributes. Then, using an integer k (k≥2) as the base and the first reconstruction exposure index as the exponent, the first reconstruction privacy level is calculated. For example, if k is 2, the first reconstruction exposure index is 2.5, and the first reconstruction privacy level is 2. 2.5 ≈5.656.

[0044] Subsequently, following the same method, the remaining preset screening attributes are processed sequentially until the Y-th reconstructed privacy level is obtained. Finally, based on the calculated first to Y-th reconstructed privacy levels, a new privacy level table is constructed. This reconstructed table more accurately reflects the differences in privacy risks among the preset screening attributes, providing a more reliable basis for subsequent data processing or decision-making based on privacy levels. For example, suppose there are three initial preset screening attributes A, B, and C, each with a privacy level of 3, and a reconstruction probability threshold of 0.3. Since the proportion of identical privacy levels is 1 (greater than the threshold), reconstruction is required. By traversing the non-A attribute set, the frequency of co-recording with A's expansion is counted, and the mean value is calculated to obtain A's reconstruction exposure index. Then, A's reconstructed privacy level is calculated. Similarly, B and C's reconstructed privacy levels are calculated, ultimately constructing a new privacy level table.

[0045] In a preferred embodiment, constructing the privacy level table based on the first reconstructed privacy level up to the Yth reconstructed privacy level includes: calculating the proportion of identical privacy levels among the first reconstructed privacy levels up to the Yth reconstructed privacy levels to obtain a second privacy level reconstruction probability; when the second privacy level reconstruction probability is less than the reconstruction probability threshold, outputting the privacy level table; when the second privacy level reconstruction probability is greater than or equal to the reconstruction probability threshold, sending the first reconstructed privacy level up to the Yth reconstructed privacy level and the first preset investigation attribute up to the Yth preset investigation attribute to the user terminal to obtain the first reconstructed secondary privacy level up to the Yth reconstructed secondary privacy level, and constructing the privacy level table, wherein the third privacy level reconstruction probability of the first reconstructed secondary privacy level up to the Yth reconstructed secondary privacy level is less than the reconstruction probability threshold.

[0046] Specifically, when constructing the privacy level table, we first calculate the percentage of identical privacy levels obtained after the first reconstruction up to the Yth reconstruction privacy level, and then obtain the reconstruction probability of the second privacy level. For example, suppose there are five preset investigation attributes, and their privacy levels after the first reconstruction are 3, 3, 4, 3, and 5 respectively. Then, privacy level 3 appears 3 times, and the total number is 5. Therefore, the reconstruction probability of the second privacy level of privacy level 3 is 3 / 5 = 0.6. This is only to illustrate the calculation method; in practice, it is necessary to calculate the percentage of all identical privacy levels.

[0047] Next, the calculated reconstruction probability of the second privacy level is compared with the preset reconstruction probability threshold. If the reconstruction probability of the second privacy level is less than the reconstruction probability threshold, it indicates that the distribution of privacy levels meets the requirements, and different privacy levels can better distinguish the privacy risks of each preset investigation attribute. At this time, the current privacy level table can be directly output, which records the reconstructed privacy level corresponding to each preset investigation attribute.

[0048] However, when the reconstruction probability of the second privacy level is greater than or equal to the reconstruction probability threshold, it indicates that the reconstructed privacy levels still have many identical values, resulting in insufficient differentiation of privacy risks, requiring further optimization. In this case, the first reconstructed privacy levels up to the Yth reconstructed privacy level, along with the corresponding first preset screening attributes up to the Yth preset screening attributes, are sent to the user terminal. Based on more comprehensive data, professional knowledge, or other considerations, the user terminal readjusts the privacy levels of these preset screening attributes to obtain the first reconstructed secondary privacy levels up to the Yth reconstructed secondary privacy levels.

[0049] After the first and second reconstruction processes, it is ensured that the reconstruction probability of the third privacy level (i.e., the proportion of identical privacy levels in the first reconstruction, second privacy level up to the Yth reconstruction, second privacy level) is less than the reconstruction probability threshold. This allows the privacy levels in the final privacy level table to more accurately reflect the differences in privacy risks among the preset screening attributes, providing a reliable basis for subsequent data processing and security policy formulation based on privacy levels. For example, if multiple preset screening attributes still have the same privacy level after the initial reconstruction, a second adjustment can be made by sending the data to the user terminal to differentiate the privacy levels of these attributes, thereby meeting the reconstruction probability threshold requirement and completing the construction of the privacy level table.

[0050] In a preferred embodiment, the screening attribute set is sent to a zero-knowledge processing data sandbox, and multiple screening attribute monitoring feature value sets of multiple candidate personnel are processed to obtain a first screening attribute zero-knowledge feature value set up to a Qth screening attribute zero-knowledge feature value set. This includes: traversing the screening attribute set via a user terminal, configuring a partitioning step size set and a partitioning interval set; partitioning the partitioning interval set based on the partitioning step size set to obtain an interval partitioning result set; traversing the interval partitioning result set, sequentially filling the unit step size interval with Arabic numerals from one end of the interval to construct a multi-valued function set; embedding the multi-valued function set into the zero-knowledge processing data sandbox, and processing the multiple screening attribute monitoring feature value sets of the multiple candidate personnel to obtain a first screening attribute zero-knowledge feature value set up to a Qth screening attribute zero-knowledge feature value set.

[0051] For example, when the user client iterates through the set of attributes to be investigated, the client will configure a set of partitioning step sizes and a set of partitioning intervals. The partitioning step size set contains a series of step size values ​​used to partition intervals, such as {1, 2, 5}, while the partitioning interval set defines the original interval range to be partitioned, such as [0, 10]. Based on the partitioning step size set, the partitioning interval set is partitioned. For example, if the partitioning step size is 2 and the partitioning interval is [0, 10], it can be divided into multiple sub-intervals such as [0, 2), [2, 4), [4, 6), [6, 8), [8, 10], thereby obtaining a set of interval partitioning results.

[0052] Subsequently, the set of interval partitioning results is traversed, and starting from the beginning endpoint of each interval, Arabic numerals are used to fill the unit step interval sequentially. For example, for the interval [0, 2), the numbers 0 and 1 can be filled (if the unit step interval is considered to include the starting value but not the ending value, and the filling logic is illustrated here using a step of 1 as an example, the actual process may be adjusted according to specific needs). In this way, a set of multi-valued functions is constructed, and each function in the set of multi-valued functions corresponds to a rule for interval partitioning and number filling.

[0053] Next, the constructed set of multi-valued functions is embedded in a zero-knowledge processing data sandbox. The zero-knowledge processing data sandbox can process data without revealing the specific information of the original data. When processing multiple screening attribute monitoring feature sets for multiple candidate personnel, these feature sets contain the specific values ​​of the candidate personnel for each screening attribute. The zero-knowledge processing data sandbox transforms the monitoring feature values ​​of each candidate personnel for each screening attribute based on the embedded set of multi-valued functions. For example, if a candidate personnel's monitoring feature value for the age attribute is 25, and the function in the multi-valued function set for the age attribute maps the interval [20, 30) to the number 3, then the zero-knowledge feature value corresponding to the candidate personnel's age attribute after processing is 3.

[0054] Through this processing method, a corresponding set of zero-knowledge feature values ​​can be obtained for each investigation attribute, ultimately resulting in the set of zero-knowledge feature values ​​for the first investigation attribute up to the Qth investigation attribute. These sets of zero-knowledge feature values ​​provide a usable form for subsequent data analysis or decision-making while ensuring data privacy.

[0055] In a preferred embodiment, summing the indices of the same attribute according to the first investigation attribute sequence and the second investigation attribute sequence to obtain the sum of the first investigation attribute indices up to the Qth investigation attribute indices includes: weighting the indices of the first investigation attribute sequence according to a first preset weight to obtain a first investigation attribute reconstruction indices set; and weighting the indices of the second investigation attribute sequence according to a second preset weight to obtain a second investigation attribute reconstruction indices set, wherein the second preset weight is less than or equal to 0.5 times the first preset weight.

[0056] In detail, for the first and second investigation attribute sequences, a summation operation of the same attribute indices needs to be performed to obtain the summation of the first investigation attribute indices up to the Qth investigation attribute indices. First, the indices of the first investigation attribute sequence are weighted according to a first preset weight. The first preset weight is a pre-set value used to measure the importance of the indices of the first investigation attribute sequence in the summation process. For example, if the first investigation attribute sequence is [attribute A, attribute B, attribute C], and its original indices are [1, 2, 3], and the first preset weight is set to 2, then each indices are multiplied by this weight to obtain the reconstructed indices of the first investigation attributes [2, 4, 6]. This process, by assigning higher weights, highlights the influence of the first investigation attribute sequence indices in the final summation result. Furthermore, considering privacy protection requirements, this weight setting reflects that the first investigation attribute sequence may carry more critical privacy information and needs to occupy a more significant position in the calculation.

[0057] Subsequently, the indices of the second screening attribute sequence are weighted according to the second preset weight. The second preset weight is less than or equal to 0.5 times the first preset weight. This setting reflects that, under the consideration of privacy protection, the contribution of the second screening attribute sequence indices to the final summation result is relatively low. For example, if the second screening attribute sequence is also [attribute A, attribute B, attribute C], and the original indices are [1, 2, 3], and the second preset weight is set to 0.5 (satisfying the condition of being less than or equal to 0.5 times the first preset weight), then each indices are multiplied by this weight to obtain the reconstructed indices of the second screening attributes [0.5, 1, 1.5].

[0058] By setting different weights, when summing the sequence numbers of the same attribute in the future, the contribution of the two investigated attribute sequence numbers to the final result can be reasonably adjusted in combination with privacy protection requirements. This ensures that the summing result reflects the attribute relationship while fully considering privacy protection factors, providing a basis for subsequent analysis or decision-making based on the sequence summing result that meets privacy protection requirements.

[0059] In a preferred embodiment, attribute abundance optimization is performed according to the third screening attribute sequence to achieve the sorting of the multiple candidate personnel, including: extracting the first sequence screening attribute from the third screening attribute sequence; obtaining the first sequence screening attribute baseline value from the user end based on the first sequence screening attribute; sending the first sequence screening attribute baseline value to a zero-knowledge processing data sandbox to obtain the first sequence screening attribute baseline value zero-knowledge feature value; obtaining a set of candidate zero-knowledge feature values ​​for the first sequence screening attributes of the multiple candidate personnel; extracting the first-level sorting screening attributes that are consistent with the first sequence screening attribute baseline value zero-knowledge feature value from the set of candidate zero-knowledge feature values ​​for the first sequence screening attributes; when the number of the first-level sorting screening attributes is multiple, selecting the second sequence screening attribute from the third screening attribute sequence to expand the attribute abundance and then executing a loop; when the number of the first-level sorting screening attributes is 1, outputting the personnel corresponding to the first-level sorting screening attribute; when the number of the first-level sorting screening attributes is 0, updating the multiple candidate personnel and executing a loop.

[0060] Specifically, in the process of selecting multiple candidates, the first step is to optimize the abundance of attributes based on the third screening attribute sequence. Specifically, the first-order screening attribute is extracted from the third screening attribute sequence. For example, if the third screening attribute sequence is [age, income, occupation], the first-order screening attribute is "age". Then, based on the first-order screening attribute, the corresponding baseline value of the first-order screening attribute is obtained from the user's end. Assuming the user's end-to-end "age" baseline value is 30 years old, this baseline value is then sent to a zero-knowledge processing data sandbox. The zero-knowledge processing data sandbox can process the data without revealing the specific information of the original data, thereby obtaining a zero-knowledge feature value of the first-order screening attribute baseline value. This zero-knowledge feature value is the privacy-protected form of the baseline value.

[0061] Simultaneously, a set of zero-knowledge feature values ​​for the first-order screening attribute of multiple candidate personnel is obtained. For example, if there are 5 candidate personnel, their "age" zero-knowledge feature values ​​are the corresponding processed numerical sets. Then, the values ​​in the set of zero-knowledge feature values ​​are compared with the baseline zero-knowledge feature value of the first-order screening attribute, and the first-level screening attribute that matches is selected. If there are multiple first-level screening attributes, for example, if the "age" zero-knowledge feature values ​​of 3 candidate personnel match the baseline zero-knowledge feature value, then the second-order screening attribute, such as "income," is selected from the third screening attribute sequence to expand the attribute abundance. That is, the screening conditions are further refined using the "income" attribute. Then, the above cyclical process from attribute extraction to comparison and screening is repeated for these 3 candidate personnel based on the "income" attribute.

[0062] If there is only one primary sorting attribute, for example, only one candidate's "age" zero-knowledge feature value matches the baseline zero-knowledge feature value, then the candidate corresponding to that primary sorting attribute is directly output, completing the sorting. If there are zero primary sorting attributes, for example, all candidates' "age" zero-knowledge feature values ​​do not match the baseline zero-knowledge feature value, then multiple candidates are updated, for example, candidates are reselected from a broader set of candidates, and the above cyclical process from extracting the first-order sorting attribute to comparison and filtering is executed again until the sorting task is completed. Through this process, multiple candidates can be effectively sorted based on sorting attributes while ensuring privacy.

[0063] The data processing method for optimizing the abundance of anonymous personal attributes provided in this invention has at least the following technical effects:

[0064] 1. By constructing a privacy level table based on the common recording frequency of personnel identity information and investigation attributes, and introducing an expanded common recording frequency calculation and secondary reconstruction mechanism during the reconstruction process, the privacy level is dynamically adjusted. This effectively solves the problem of insufficient privacy risk differentiation caused by unreasonable distribution of privacy levels, significantly improves the accuracy and adaptability of privacy protection level classification, and ensures a more meticulous balance between data availability and privacy during data processing.

[0065] 2. The set of investigated attributes is sent to the zero-knowledge processing data sandbox. A set of multi-valued functions is constructed using the configured partition step size and interval set to process the set of investigated attribute monitoring features and generate a set of zero-knowledge features. This not only ensures the privacy and security of the original data during processing, but also transforms continuous features into discrete zero-knowledge features with privacy protection characteristics through multi-valued functions. This provides effective support for subsequent privacy-preserving data analysis and enhances the privacy and flexibility of data processing.

[0066] 3. The screening attribute set is sorted based on multiple dimensions, including privacy, eigenvalue variance, and ordinal summation, to obtain a third screening attribute sequence. This sequence is then used for attribute abundance optimization and personnel sorting. By setting different preset weights to weight the ordinal numbers and dynamically adjusting the screening attribute sequence or candidate personnel set based on the sorting results, accurate personnel sorting is achieved while ensuring privacy. This improves sorting efficiency and accuracy, providing an innovative solution for large-scale anonymized personnel data processing scenarios.

[0067] Example 2:

[0068] like Figure 2 As shown, based on the same inventive concept as the data processing method for optimizing the abundance of anonymous personal attributes provided in Embodiment 1, this embodiment of the invention also provides a data processing system for optimizing the abundance of anonymous personal attributes, the system comprising:

[0069] The first sorting module 11 is used to sort the set of investigation attributes from smallest to largest based on a predefined privacy level table to obtain the first investigation attribute sequence.

[0070] The numerical processing module 12 is used to send the screening attribute set to the zero-knowledge processing data sandbox, process the multiple screening attribute monitoring feature value sets of multiple candidates, and obtain the first screening attribute zero-knowledge feature value set up to the Qth screening attribute zero-knowledge feature value set.

[0071] The second sorting module 13 is used to sort the investigation attribute set in descending order of the variance of the zero-knowledge feature values ​​from the first investigation attribute zero-knowledge feature value set to the Qth investigation attribute zero-knowledge feature value set, thereby obtaining the second investigation attribute sequence.

[0072] The summation processing module 14 is used to sum the same attribute sequence numbers according to the first investigation attribute sequence and the second investigation attribute sequence to obtain the first investigation attribute sequence number sum up to the Qth investigation attribute sequence number sum.

[0073] The third sorting module 15 is used to sort the set of investigation attributes from smallest to largest according to the sum of the first investigation attribute numbers up to the sum of the Qth investigation attribute numbers, so as to obtain the third investigation attribute sequence.

[0074] Abundance optimization module 16 is used to optimize the attribute abundance according to the third screening attribute sequence, so as to sort out the multiple candidates.

[0075] Furthermore, the first sorting module 11 is also used to perform the following steps:

[0076] Through the user terminal, a preset set of screening attributes is configured; a first preset screening attribute is extracted from the preset screening attribute set, and the first common recording frequency of personnel identity information and the first preset screening attribute is retrieved to obtain a first exposure index. A first privacy level is calculated with k as the base and the first exposure index as the exponent, where k is an integer and k≥2; this process continues until the Yth preset screening attribute is extracted from the preset screening attribute set, and the first common recording frequency of personnel identity information and the Yth preset screening attribute is retrieved to obtain a Yth exposure index. A Yth privacy level is calculated with k as the base and the Yth exposure index as the exponent, where k is an integer and k≥2; a privacy level table is constructed based on the first privacy level up to the Yth privacy level.

[0077] Furthermore, the first sorting module 11 is also used to perform the following steps:

[0078] The proportion of identical privacy levels from the first privacy level up to the Yth privacy level is statistically analyzed to obtain the reconstruction probability of the first privacy level. When the reconstruction probability of the first privacy level is less than the reconstruction probability threshold, the privacy level table is output. When the reconstruction probability of the first privacy level is greater than or equal to the reconstruction probability threshold: the non-first preset investigation attribute set of the preset investigation attribute set is traversed, and the first extended common recording frequency set with the first preset investigation attribute is statistically analyzed. The first common recording frequency is added to the first extended common recording frequency set and the mean is calculated to obtain the first reconstruction exposure index. The first reconstruction privacy level is calculated with k as the base and the first reconstruction exposure index as the exponent, where k is an integer and k≥2. This process continues until the Yth reconstruction privacy level is obtained. The privacy level table is constructed based on the first reconstruction privacy level up to the Yth reconstruction privacy level.

[0079] Furthermore, the first sorting module 11 is also used to perform the following steps:

[0080] The proportion of identical privacy levels among the first reconstructed privacy levels up to the Yth reconstructed privacy level is statistically analyzed to obtain the reconstruction probability of the second privacy level. When the reconstruction probability of the second privacy level is less than the reconstruction probability threshold, the privacy level table is output. When the reconstruction probability of the second privacy level is greater than or equal to the reconstruction probability threshold, the first reconstructed privacy level up to the Yth reconstructed privacy level and the first preset investigation attribute up to the Yth preset investigation attribute are sent to the user terminal to obtain the first reconstructed secondary privacy level up to the Yth reconstructed secondary privacy level, and the privacy level table is constructed. The reconstruction probability of the third privacy level among the first reconstructed secondary privacy level up to the Yth reconstructed secondary privacy level is less than the reconstruction probability threshold.

[0081] Furthermore, the numerical processing module 12 is also used to perform the following steps:

[0082] Through the user terminal, the set of investigation attributes is traversed, and a set of division step sizes and a set of division intervals are configured. Based on the set of division step sizes, the set of division intervals is divided to obtain a set of interval division results. The set of interval division results is traversed, and Arabic numerals are used to fill the unit step size intervals sequentially from one end of the interval to construct a set of multi-valued functions. The set of multi-valued functions is embedded in the zero-knowledge processing data sandbox to process the monitoring feature value sets of multiple investigation attributes of the multiple candidate personnel to obtain the first zero-knowledge feature value set of the investigation attribute up to the Qth zero-knowledge feature value set of the investigation attribute.

[0083] Furthermore, the summation processing module 14 is also used to perform the following steps:

[0084] According to the first preset weight, the sequence number of the first investigation attribute sequence is weighted to obtain the first investigation attribute reconstruction sequence number set; according to the second preset weight, the sequence number of the second investigation attribute sequence is weighted to obtain the second investigation attribute reconstruction sequence number set, wherein the second preset weight is less than or equal to 0.5 times the first preset weight.

[0085] Furthermore, the abundance optimization module 16 is also used to perform the following steps:

[0086] Extract the first sequence number of the third screening attribute sequence; based on the first sequence number of the screening attribute, obtain the baseline value of the first sequence number of the screening attribute from the user end; send the baseline value of the first sequence number of the screening attribute to the zero-knowledge processing data sandbox to obtain the zero-knowledge feature value of the baseline value of the first sequence number of the screening attribute; obtain the set of candidate zero-knowledge feature values ​​of the first sequence number of the screening attributes of the multiple candidate personnel; extract the first-level sorting screening attributes that are consistent with the zero-knowledge feature value of the baseline value of the first sequence number of the screening attribute from the set of candidate zero-knowledge feature values ​​of the first sequence number of the screening attribute; when there are multiple first-level sorting screening attributes, select the second sequence number of the screening attribute from the third screening attribute sequence, expand the attribute abundance, and then execute the loop; when there is only one first-level sorting screening attribute, output the personnel corresponding to the first-level sorting screening attribute; when there are zero first-level sorting screening attributes, update the multiple candidate personnel and execute the loop.

[0087] Through the foregoing detailed description of a data processing method for optimizing the abundance of anonymous personal attributes, those skilled in the art will clearly understand that this embodiment provides a data processing system for optimizing the abundance of anonymous personal attributes. Since the system disclosed in the embodiments corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant details can be found in the method section. The above description of the disclosed embodiments enables those skilled in the art to implement or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A data processing method for optimizing the abundance of attributes of anonymous individuals, characterized in that, include: Based on a predefined privacy level table, the set of investigation attributes is sorted from smallest to largest according to privacy requirements to obtain the first investigation attribute sequence. The set of screening attributes is sent to the zero-knowledge processing data sandbox, and the monitoring feature value sets of multiple screening attributes of multiple candidates are processed to obtain the first zero-knowledge feature value set of screening attributes up to the Qth zero-knowledge feature value set of screening attributes. The investigation attribute set is sorted from largest to smallest according to the variance of the zero-knowledge feature value of the first investigation attribute zero-knowledge feature value set up to the Qth investigation attribute zero-knowledge feature value set, to obtain the second investigation attribute sequence; The first investigation attribute sequence and the second investigation attribute sequence are summed according to the same attribute sequence to obtain the sum of the first investigation attribute sequence up to the Qth investigation attribute sequence. Sort the set of investigation attributes in ascending order by summing the first investigation attribute numbers up to summing the Qth investigation attribute numbers, and obtain the third investigation attribute sequence; The attribute abundance is optimized according to the third screening attribute sequence to achieve the sorting of the multiple candidate personnel; Specifically, based on a predefined privacy level table, the set of investigated attributes is sorted from lowest to highest privacy level to obtain the first investigated attribute sequence, which includes: Configure a preset set of investigation attributes through the user client; Extract the first preset investigation attribute from the preset investigation attribute set, retrieve the first common recording frequency between the personnel identity information and the first preset investigation attribute, obtain the first exposure index, and calculate the first privacy level with k as the base and the first exposure index as the exponent, where k is an integer and k≥2; Until the Yth preset screening attribute is extracted from the preset screening attribute set, the first common recording frequency of personnel identity information and the Yth preset screening attribute is retrieved to obtain the Yth exposure index. The Yth privacy level is calculated with k as the base and the Yth exposure index as the exponent, where k is an integer and k≥2. The privacy level table is constructed based on the first privacy level up to the Yth privacy level.

2. The method as described in claim 1, characterized in that, Based on the first privacy level up to the Yth privacy level, the privacy level table is constructed, including: The proportion of identical privacy levels up to the Yth privacy level is counted to obtain the reconstruction probability of the first privacy level. When the reconstruction probability of the first privacy level is less than the reconstruction probability threshold, the privacy level table is output; When the reconstruction probability of the first privacy level is greater than or equal to the reconstruction probability threshold: Traverse the non-first preset investigation attribute set of the preset investigation attribute set, and count the first extended common record frequency set with the first preset investigation attribute; The first common recording frequency is added to the first expanded common recording frequency set and the mean is calculated to obtain the first reconstruction exposure index. The first reconstruction privacy level is calculated with k as the base and the first reconstruction exposure index as the exponent, where k is an integer and k≥2. Until the Y-th level of privacy reconstruction is obtained; The privacy level table is constructed based on the first reconstructed privacy level up to the Yth reconstructed privacy level.

3. The method as described in claim 2, characterized in that, Based on the first reconstructed privacy level up to the Yth reconstructed privacy level, construct the privacy level table, including: The probability of reconstructing the second privacy level is obtained by statistically analyzing the proportion of identical privacy levels from the first reconstructed privacy level up to the Yth reconstructed privacy level. When the reconstruction probability of the second privacy level is less than the reconstruction probability threshold, the privacy level table is output; When the reconstruction probability of the second privacy level is greater than or equal to the reconstruction probability threshold, the first reconstruction privacy level up to the Yth reconstruction privacy level and the first preset investigation attribute up to the Yth preset investigation attribute are sent to the user terminal to obtain the first reconstruction secondary privacy level up to the Yth reconstruction secondary privacy level, and the privacy level table is constructed, wherein the reconstruction probability of the third privacy level from the first reconstruction secondary privacy level up to the Yth reconstruction secondary privacy level is less than the reconstruction probability threshold.

4. The method as described in claim 1, characterized in that, The set of screening attributes is sent to a zero-knowledge processing data sandbox. Multiple screening attribute monitoring feature sets for multiple candidate personnel are processed to obtain the first set of zero-knowledge feature values ​​for screening attributes up to the Qth set of zero-knowledge feature values ​​for screening attributes, including: Through the user terminal, traverse the set of investigation attributes and configure the set of division step size and the set of division interval; Based on the set of partitioning step sizes, the set of partitioning intervals is divided to obtain a set of interval partitioning results; Traverse the set of interval division results, and fill the unit step interval with Arabic numerals starting from one end of the interval to construct a set of multi-valued functions; The set of multi-valued functions is embedded in the zero-knowledge processing data sandbox to process the monitoring feature value sets of multiple screening attributes of the multiple candidate personnel, so as to obtain the first screening attribute zero-knowledge feature value set up to the Qth screening attribute zero-knowledge feature value set.

5. The method as described in claim 1, characterized in that, Based on the first investigation attribute sequence and the second investigation attribute sequence, sum the indices of the same attribute to obtain the sum of the first investigation attribute indices up to the Qth investigation attribute indices, including: According to the first preset weight, the sequence number of the first investigation attribute sequence is weighted to obtain the first investigation attribute reconstruction sequence number set; According to the second preset weight, the sequence number of the second investigation attribute sequence is weighted to obtain the second investigation attribute reconstruction sequence number set, wherein the second preset weight is less than or equal to 0.5 times the first preset weight.

6. The method as described in claim 1, characterized in that, Attribute abundance optimization is performed according to the third screening attribute sequence to achieve the sorting of the multiple candidate personnel, including: Extract the first index of the third investigation attribute sequence; Based on the first sequence number investigation attribute, obtain the baseline value of the first sequence number investigation attribute from the user terminal; The baseline value of the first sequence number screening attribute is sent to the zero-knowledge processing data sandbox to obtain the zero-knowledge feature value of the baseline value of the first sequence number screening attribute. Obtain the first sequence number screening attribute set of zero-knowledge feature values ​​for the plurality of candidate personnel; When the first-order screening attribute is a first-level sorting and screening attribute whose zero-knowledge feature value is consistent with the baseline value of the first-order screening attribute in the candidate zero-knowledge feature value set; When there are multiple primary sorting and screening attributes, the second sequence screening attribute is selected from the third screening attribute sequence to expand the attribute abundance, and then the loop is executed. When the number of the primary sorting and screening attributes is 1, output the personnel corresponding to the primary sorting and screening attributes; When the number of the first-level sorting and screening attributes is 0, the multiple candidate personnel are updated and the loop is executed.

7. A data processing system for optimizing the abundance of attributes of anonymous individuals, characterized in that, The system is used to implement the data processing method for optimizing the abundance of anonymous personal attributes as described in any one of claims 1-6, the system comprising: The first sorting module is used to sort the set of investigation attributes from smallest to largest based on a predefined privacy level table to obtain the first investigation attribute sequence. The numerical processing module is used to send the screening attribute set to the zero-knowledge processing data sandbox, process the multiple screening attribute monitoring feature value sets of multiple candidates, and obtain the first screening attribute zero-knowledge feature value set up to the Qth screening attribute zero-knowledge feature value set. The second sorting module is used to sort the investigation attribute set in descending order of the variance of the zero-knowledge feature values ​​from the first investigation attribute zero-knowledge feature value set to the Qth investigation attribute zero-knowledge feature value set, to obtain the second investigation attribute sequence. The summation processing module is used to sum the same attribute sequence numbers according to the first investigation attribute sequence and the second investigation attribute sequence to obtain the first investigation attribute sequence number sum up to the Qth investigation attribute sequence number sum; The third sorting module is used to sort the set of investigation attributes from smallest to largest according to the sum of the first investigation attribute numbers up to the sum of the Qth investigation attribute numbers, so as to obtain the third investigation attribute sequence. The abundance optimization module is used to optimize the attribute abundance according to the third screening attribute sequence, so as to sort out the multiple candidates.

Images