A dynamic heterogeneous weight anonymous data disassembly method and system

Abstract

The application discloses a dynamic heterogeneous weight anonymous data decomposition method, which aims at solving the large-scale anonymous data collection and statistical problems by using the dynamic data weight distribution framework of the same table different weight and the same weight different number, constructing a one-time traceless weight data collection form through an information system, realizing form data collection through an anonymous random distribution method, finally weighting and summarizing the same table data, and performing secondary same weight mean processing based on the actual recovered data to obtain the final anonymous data statistical result. The method has the advantages of dynamic decomposition, whole-process anonymity and good anti-repudiation.

Description

Technical Field

[0001] This invention belongs to the field of data statistics algorithm technology, and relates to weighted anonymous form design and data statistics methods, specifically a dynamic heterogeneous anonymous data deconstruction method and system. Background Technology

[0002] With the development of information technology and the popularization of mobile Internet, more and more administrative operations of traditional enterprises have been digitized. Informatization helps enterprises achieve high-quality development and is an important means to enhance their overall strength.

[0003] In the process of employee performance appraisal in enterprises, an important task is to conduct multi-dimensional evaluation and scoring of specific appraisees. Due to the large scale and complex structure of employees in large enterprises, the compilation, distribution and collection of scoring and evaluation test papers involve a large workload. In particular, when the evaluators have different roles and positions, different weight ratios need to be used to distinguish them, and the evaluation also needs to be anonymous. For traditional manual statistics, these problems are that the workload is large, the calculation is complicated, the timeliness is low and the error is easy to occur.

[0004] Current conventional practices mainly achieve anonymity through distributed keys or blockchain technology, followed by electronic distribution for data collection. After data collection, secondary classification of the collected data is required, and statistical analysis must be performed according to each category to achieve weighted anonymized data collection. This method has poor dynamic scalability and is not suitable for large-scale anonymized data collection and statistical scenarios, especially in cases where the organizational structure is complex, the statistical process becomes even more time-consuming. Therefore, there is a need to invent a flexible, anonymized, and orderly method for constructing and analyzing evaluation data based on different weights in order to meet the needs of constructing and statistically analyzing anonymized evaluation data with different weights. Summary of the Invention

[0005] To address the shortcomings and defects of existing technologies, the inventors designed and improved a method that solves the problem of large-scale anonymous data collection and statistics, specifically addressing the deficiencies of existing anonymous survey techniques. This method utilizes a dynamic data weighting framework with different weights for the same table and different numbers for the same weight. It constructs a one-time, traceless weighted data collection form through an information system, collects data from the form using an anonymous random distribution method, and finally weights and aggregates the data from the same table. A secondary equal-weighting average is then performed based on the actual collected data to obtain the final anonymous data statistical results. This method possesses the ability to construct large-scale anonymous data collection forms and can also perform weighted statistical aggregation of data from each form, offering advantages such as dynamic deconstruction, full-process anonymity, and strong non-repudiation.

[0006] This invention is implemented as follows: a dynamic heterogeneous anonymous data deconstruction method, comprising the following steps:

[0007] Step S1: Divide the data according to the roles in the data collection form, and divide the role data into a multi-level role category hierarchy according to the category corresponding to the role data. Construct a hierarchical role tree with the role category hierarchy as the root node. The role category hierarchy is a top-down hierarchical structure from coarse to fine. (Similar to an organization, for example, for the leadership role of a unit, it can be divided into the unit's head leader, deputy head leader, department head leader, deputy department head leader, and ordinary employee from top to bottom.)

[0008] Step S2: Based on the constructed hierarchical role tree, construct a weight tree that corresponds one-to-one with the role data, and each node of the weight tree is the total weight value of the corresponding current role.

[0009] Step S3: Based on the role tree and weight tree, generate a one-time, traceless weighted data form using the main attributes extracted from the feature vector. This form is then randomly and anonymously distributed to data collection subjects for completion, forming statistical data for the data collection form. The one-time, traceless weighted data form binds the form to a weighted code within the system. The weighted code can only be used once and will be marked in the system. This method of binding the form to the weighted code allows the code to be registered in the system. Each use of the weighted code is marked in the system, with a default usage count of 1, meaning it cannot be used again after one use. This can be implemented using, but is not limited to, QR codes. The system mentioned here refers to a data storage area used in this embodiment to store coded data, implemented based on a software algorithm program.

[0010] Step S4: Analyze and statistically process the data collection forms, and aggregate the data in the same table: Traverse each table, extract the role information, weight information, and total data information of each table, and construct a statistical data structure using these three types of information; The total data information is used to calculate the total data for the score, and the actual score of the current data form is obtained by multiplying the total data by the weight.

[0011] Step S5: Perform a second weighted average on the statistical data structure and sum the results within the same table to obtain the statistical results of the anonymized data with different weights. Since there are differences in the actual number of people in each role, but each role category has the maximum weight value, a second weighted average is required. That is, for each role category, a second weighted average is performed on the actual number of form data under that role category based on the total weight value of that role category, thereby obtaining the weighted score result for that role category.

[0012] Furthermore, the method for constructing the role tree in step S1 is as follows: first classify the roles according to their categories, then count the role categories at the same level, classify them again, until all roles are classified to the finest granularity, and then construct the role tree from top to bottom, with the highest-level role as the root node, and the leaf nodes of each root node as secondary roles, until all specific roles are added to the role tree and become terminal nodes; wherein, each root node is a common feature division of the role type, which is refined from top to bottom, and the terminal node of the role tree is the finest division of the role.

[0013] Furthermore, the method for constructing the weight tree in step S1 is as follows: Based on the constructed role tree, find the terminal leaf node of the role tree, assign a weight value to each leaf node in turn, and recursively calculate the total weight value of each root node at each level from bottom to top. The rule is that the weight of each root node is the sum of the weights of all leaf nodes. The weight represents the total weight of that type of role. Since the actual number of people under each type of role is different, the total weight is not a simple sum of the weights of the specific people under that type of role. For the same role with the same weight, the weight value of each node on the role tree is the average of the child nodes of that node.

[0014] Furthermore, the method for generating the one-time seamless weighted data form is as follows: based on the data form, role tree, and weight tree, a set of four-tuple feature vectors is extracted:

[0015] Vector n =(tableID) n ,roleID n ,value n ,random_timeSeq n )

[0016] In the above vectors, n represents the number of all terminal leaf nodes in the role tree; tableID is the identifier id of the data collection form, used to distinguish different data forms; roleID is the role identifier id, used to distinguish the role name; value is the weight value, used to calculate the weight value of this vector group; random_timeSeq is a time-based random sequence value, used to generate the identifier of a one-time data form. This identifier can only be used once each time data is collected, thus preventing duplicate data collection. At the same time, a randomization algorithm is used to make the generated identifier anonymous. To prevent data manipulation and unauthorized vote manipulation, although data collection is anonymous, each anonymous credential must be registered in the system. Through the above four-tuple feature vectors, each system-authorized, unforgeable, anonymous weighted data collection form can be distributed to anonymous voters for data collection.

[0017] Furthermore, the aggregation of data within the same table includes the following steps:

[0018] Step a) Construct an empty multi-branch statistical tree for each different table to calculate the heterogeneous weight data statistics, and set the height of the multi-branch statistical tree to 3; Step b) For each completed data table, extract three features: role information, weight information, and data information; Step c) Based on the features extracted from each table, search the statistical tree, matching sequentially in the order of role-weight. If no matching (role, weight) node is found, create the corresponding node; otherwise, insert the data information from the features into the next lower level of the retrieved stage. The leaf nodes of the statistical tree are all data nodes; Step d) Repeat the above three steps until the data of all tables has been added to the statistical tree.

[0019] Furthermore, the secondary equal-weighting average processing includes: performing calculations on data under the same role and weight to ensure that the data does not exceed the upper limit of the data under that role and weight, and the calculation method includes, but is not limited to, calculating the average.

[0020] In another aspect, the present invention provides a dynamic heterogeneous anonymized data deconstruction system, comprising: a dynamic heterogeneous anonymized data construction module for constructing a role tree and a weight tree, constructing one-time anonymized tags and form data binding, and constructing a weighted anonymized data form for one-time data collection; a dynamic heterogeneous anonymized data distribution module for data distribution, generating pre-classified weighted anonymized data forms for various types of one-time data collection, and distributing them to each role; and a dynamic heterogeneous anonymized data statistics module for collecting and summarizing the filling results of the distributed dynamic heterogeneous data forms, identifying the role and weight of each data sample, constructing a statistical tree, performing secondary mean processing on data samples with the same role weight, and finally summing all the results to obtain the statistical results of the dynamic heterogeneous anonymized data.

[0021] Furthermore, the dynamic heterogeneous anonymous data construction module can also divide the data collection object according to its role, divide it into multiple levels according to the role category, construct a hierarchical role tree with the role category level as the root node; and construct a one-to-one corresponding weight tree according to the role tree.

[0022] Furthermore, the dynamic heterogeneous anonymized data statistics module can also aggregate data from the same table, traverse each table, extract three types of information—role, weight, and data—from each table, and construct a statistical data structure based on these three types of information; it also performs secondary equal-weighting average processing, sums the statistics from the same table, and obtains the statistical results of heterogeneous anonymized data.

[0023] The working principle and beneficial effects of this invention are as follows: This invention provides a dynamic heterogeneous weighted anonymous data deconstruction method. In the data collection form construction stage, a form data role tree is first constructed, followed by a form data weight tree. The role tree and weight tree are then bound to the data collection form. An appropriate number of one-time, traceless, weighted data collection forms are pre-generated and anonymously and randomly distributed. The distribution is based on roles, but the specific individuals within those roles are unknown. For example, if the distribution target is middle-level managers, n voting forms for middle-level managers will be generated. The distribution will target middle-level managers, not ordinary employees, but it is impossible to track which middle-level manager voted for which data. Anonymity means that after distribution, the ballot forms cannot be traced to who filled them out, thus protecting the anonymity of the filler. In the data collection form analysis and statistics stage, data from the same table is first aggregated, and then a secondary equal-weighted averaging process based on the actual collected data is performed according to the role tree constructed in the construction stage, ultimately obtaining the statistical results of the dynamic heterogeneous weighted anonymous data. This dynamic heterogeneous weighted anonymized data deconstruction method is an innovative data collection approach. In large-scale anonymized data collection and statistics, it utilizes a dynamic data weighting framework with different weights within the same table and different numbers for the same weight. It constructs a one-time, traceless weighted data collection form through an information system, collects data from the form using an anonymous random distribution method, and finally weights and aggregates the data within the same table. A secondary equal-weighting average is then performed based on the actual collected data to obtain the final anonymized statistical results. This method possesses the ability to construct large-scale anonymized data collection forms and to statistically aggregate data from each form according to weights. It has the advantages of dynamic deconstruction, full-process anonymity, and good non-repudiation. Attached Figure Description

[0024] Figure 1 This is a flowchart of the dynamic heterogeneous anonymized data deconstruction method of the present invention;

[0025] Figure 2 These are schematic diagrams of the role tree of the present invention in three embodiments;

[0026] Figure 3 This is a schematic diagram of the weighted tree of the present invention in Example 3. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments and the accompanying drawings. It should be understood that these descriptions are merely exemplary and not intended to limit the scope of the invention. Furthermore, descriptions of well-known structures and techniques are omitted in the following description to avoid unnecessarily obscuring the concept of the invention.

[0028] Example 1: A method for constructing dynamic heterogeneous anonymized data. This method mainly consists of three stages: dynamic heterogeneous anonymized data construction, dynamic heterogeneous anonymized data distribution, and dynamic heterogeneous anonymized data statistics.

[0029] The construction of dynamic heterogeneous anonymous data is a preparatory stage: this stage requires the completion of role tree construction, weight tree construction, one-time anonymous label construction, and form data binding. After processing in this stage, a weighted anonymous data form for one-time data collection will be constructed.

[0030] Dynamic heterogeneous anonymous data distribution phase: This phase involves data distribution. Since the data collection targets are divided according to roles, the weight of each role and the actual number of people in each role are dynamic and uncertain. Therefore, this phase requires the pre-classification of various one-time data collection forms generated in the preparation phase, and then offline distribution to the personnel in each role.

[0031] Dynamic heterogeneous anonymized data statistics stage: In this stage, the results of the distributed dynamic heterogeneous data forms need to be collected and summarized. Since it is impossible to locate the destination of each data form in the anonymized state, the summary statistics need to be based on the actual number of data samples collected. This stage will identify the role and weight of each data sample, then construct a statistical tree, perform secondary mean processing on data samples with the same role and weight, and finally summarize and sum all the data to obtain the statistical results of dynamic heterogeneous anonymized data.

[0032] like Figure 1 As shown, the method includes the following steps:

[0033] Step 1: In the data collection form construction stage, divide the data collection objects according to their roles, divide them into multiple levels according to the role categories, and construct a hierarchical role tree with the role category level as the root node.

[0034] Step 2: In the data collection form construction stage, construct a weight tree one by one according to the role tree constructed in Step 1;

[0035] Step 3: In the data collection form construction stage, based on the results of Step 1 and Step 2, a one-time, traceless, weighted data form is generated and randomly and anonymously distributed to the data collection subjects for completion.

[0036] Step 4: In the data collection form analysis and statistics stage, aggregate the data in the same table. The aggregation method is to traverse each table, extract the three types of information of each table: role, weight and data, and use these three types of information to construct the statistical data structure.

[0037] Step 5: In the data collection form analysis and statistics stage, the statistical data structure in Step 4 is subjected to secondary equal-weighted mean processing, and then the same table is statistically summed to obtain the statistical results of the anonymized data with different weights.

[0038] Example 2

[0039] For example, a company wants to conduct an anonymous annual performance evaluation for three employees, Zhang San, Li Si, and Wang Wu. The company's role organizational structure is represented by the table `table_role`, the evaluation questionnaire by the table `table_question`, and the weight table `table_priority` corresponding to each role.

[0040] table_role:

[0041] table_role

[0042]

[0043] As shown in the roles above, the organizational structure of a certain unit is mainly divided into first-level role A (highest level), first-level role B (secondary level), and first-level role C. The second-level roles under the first-level roles are further subdivisions of the current first-level roles, and the third-level roles under the second-level roles are further subdivisions of the second-level roles. The role level structure in this example is a three-level structure, and the third-level roles represent terminal roles. Each role has specific weight rules.

[0044] table_question:

[0045] table_question

[0046]

[0047] This table contains the contents of a questionnaire. For ease of illustration, only one question is used as an example.

[0048] table_priority:

[0049] table_priority

[0050]

[0051] In the weighting rules mentioned above, the weight value is the sum of the actual number of people scoring that type of role. For example, when there are 2 people in A1, the actual weight of each person (A11, A12) is 30% / 2 = 15%, and the actual weight of other roles follows the same principle.

[0052] Based on step 1, in the data collection form construction stage, the data collection objects are divided according to their roles. These roles are categorized into multiple levels, and a hierarchical role tree is constructed using the role category level as the root node. The constructed role tree is as follows: Figure 2 As shown:

[0053] Then, based on step 2, in the data collection form construction phase, and according to the role tree constructed in step 1, a weight tree is constructed one-to-one. The constructed weight tree is as follows: Figure 3 As shown:

[0054] In this example, it is assumed that all leaf nodes have one person. In actual use, the actual weight of voters with the same terminal role is the average of the total role weights.

[0055] Then, according to step 3, in the data collection form construction stage, based on the results of steps 1 and 2, a one-time, anonymous, weighted data form is generated and randomly and anonymously distributed to the data collection subjects for completion. In this example, for ease of explanation, each type of form has only one actual participant, but in actual use, multiple people are involved, and the anonymity principle is not limited by this. According to this step, the generated anonymous voting form is transformed into a four-tuple feature vector.

[0056] Vector n =(tableID) n ,roleID n ,value n ,random_timeSeq n )

[0057] Specifically:

[0058] vector1=(table_question,A11,15%,10001)

[0059] vector2=(table_question,A12,15%,10003)

[0060] vector3=(table_question,A21,20%,10002)

[0061] vector4=(table_question,B11,15%,10012)

[0062] vector5=(table_question,B21,2.5%,10006)

[0063] vector6=(table_question,B22,2.5%,10007)

[0064] vector7=(table_question,C11,30%,10009)

[0065] The example above represents seven registered data collection tickets. Each ticket can only be voted on and scored once, thus preventing vote-rigging.

[0066] According to step 4, in the data collection form analysis and statistics stage, the data in the same table is aggregated. The aggregation method is to traverse each table and extract three types of information for each table: role, weight, and data. Based on these three types of information, a statistical data structure is constructed. Assuming that after the above 7 votes are collected, the score corresponding to each vote is as shown in the table below:

[0067]

[0068] Finally, according to step 5, in the data collection form analysis and statistics stage, the statistical data structure in step 4 is subjected to a second equal-weighted mean transformation, and then the same-table statistical summation is performed to obtain the statistical results of the heterogeneous anonymized data. The specific process is as follows:

[0069] Assume Zhang Sanqi's evaluation data is as follows:

[0070]

[0071]

[0072] This result is related to the feature vector.

[0073] vector1=(table_question,A11,15%,10001)

[0074] vector2=(table_question,A12,15%,10003)

[0075] vector3=(table_question,A21,20%,10002)

[0076] vector4=(table_question,B11,15%,10012)

[0077] vector5=(table_question,B21,2.5%,10006)

[0078] vector6=(table_question,B22,2.5%,10007)

[0079] vector7=(table_question,C11,30%,10009)

[0080] After combining, the score is calculated as follows:

[0081] Zhang San's test score = 80*0.15 + 80*0.15 + 80*0.2 + 80*0.15 + 80*0.025 +

[0082] 80 * 0.025 + 80 * 0.3 = 80 points

[0083] Similarly, we can calculate that Li Si gets 90 points and Wang Wu gets 100 points.

[0084] This method enables the construction and statistical analysis of anonymous heterogeneous data. In actual use, the role organizational structure, weights, and questionnaire binding relationships will be more complex, but this method can achieve unified processing, enabling the collection and statistics of massive amounts of anonymous data. It has the advantages of dynamic deconstruction, full-process anonymity, and good non-repudiation.

[0085] It should be understood that the specific embodiments described above are merely illustrative or explanatory of the principles of the invention and do not constitute a limitation thereof. Therefore, any modifications, equivalent substitutions, improvements, etc., made without departing from the spirit and scope of the invention should be included within the protection scope of the invention. Furthermore, the appended claims are intended to cover all variations and modifications falling within the scope and boundaries of the appended claims, or equivalent forms of such scope and boundaries.

Claims

1. A method for dynamic heterogeneous anonymized data deconstruction, characterized in that, Includes the following steps: Step S1: Divide the data according to the roles in the data collection form, divide the role data into multi-level role category hierarchy according to the category corresponding to the role data, and construct a hierarchical role tree with the role category hierarchy as the root node; The character categories are hierarchically structured from coarse to fine, and from top to bottom. The method for constructing the role tree is as follows: first, classify roles according to their categories, then count the role categories under the same level, classify again, until all roles are classified to the finest granularity, and then construct the role tree from top to bottom, with the highest-level role as the root node, and the leaf nodes of each root node as secondary roles, until all specific roles are added to the role tree and become terminal nodes; wherein, each root node is a common feature division of the role category, which is refined from top to bottom, and the terminal node of the role tree is the finest division of the role. Step S2: Based on the constructed hierarchical role tree, construct a weight tree that corresponds one-to-one with the role data, and each node in the weight tree is the total weight value of the corresponding current role; the construction method of the weight tree is as follows: based on the constructed role tree, find the terminal leaf node of the role tree, assign a weight value to each leaf node in turn, and recursively calculate the total weight value of each root node at each level from bottom to top. The rule is that the weight of each root node is the sum of the weights of all leaf nodes; where the weight value is the total weight of the role of this type, and the weight value of each node of the same role in the role tree with the same weight is the average of the child nodes of that node; Step S3: Based on the role tree and weight tree, a one-time, traceless weighted data form is generated using attributes extracted from the feature vector. This form is then randomly and anonymously distributed to data collection subjects for completion, forming the statistical data of the data collection form. The one-time, traceless weighted data form binds the form to a weighted code within the system. This weighted code can only be used once and will not be reused, and a marker will be generated in the system. Step S4: Analyze and statistically process the data collection forms, and aggregate the data in the same table: Traverse each table, extract the role information, weight information, and total data information of each table, and construct a statistical data structure using these three types of information; The total data information is used to calculate the total data for the score, and the actual score of the current data form is obtained by multiplying the total data by the weight. Step S5: Perform secondary equal-weighting average processing on the statistical data structure and sum the results of the same table to obtain the statistical results of the anonymized data with different weights.

2. The dynamic heterogeneous anonymized data deconstruction method according to claim 1, characterized in that, The method for generating the one-time, seamless, weighted data form is as follows: Based on the data form, role tree, and weight tree, a set of four-tuple feature vectors is extracted: In the above vector, n is the number of all terminal leaf nodes in the role tree; tableID is the identifier id of the data collection form, used to distinguish different data forms; roleID is the role identifier id, used to distinguish the name of the role; value is the weight value, used to calculate the weight value of this vector group; random_timeSeq is a time-based random sequence value, used to generate the identifier of a one-time data form. This identifier can only be used once each time data is collected, thereby preventing duplicate data collection. At the same time, a randomization algorithm is used to make the generated identifier anonymous.

3. The dynamic heterogeneous anonymized data deconstruction method according to claim 1, characterized in that, The aggregation of data from the same table includes the following steps: Step a). Construct an empty multi-branch statistical tree for each different table to calculate the statistics of heterogeneous data, and set the height of the multi-branch statistical tree; Step b). For each completed data table, extract three features: role information, weight information, and data information. Step c). Based on the features extracted from each table, a search is performed on the statistical tree, matching sequentially in the order of role-weight. If no matching role or weight node is found, the corresponding node is created; otherwise, the data information in the features is inserted into the next level of the retrieved stage. All leaf nodes of the statistical tree are data nodes. Step d). Repeat the above three steps until all data from all tables has been added to the statistics tree.

4. The dynamic heterogeneous anonymized data deconstruction method according to claim 1, characterized in that, The aforementioned secondary equal-weighting average processing includes: performing calculations on data under the same role and weight to ensure that the data does not exceed the upper limit of the data under that role and weight. The calculation methods include, but are not limited to, calculating the average.

5. A dynamic heterogeneous anonymized data destruction system, used to implement the dynamic heterogeneous anonymized data destruction method according to any one of claims 1-4, characterized in that, include: Dynamic heterogeneous anonymized data construction module: used to build role trees and weight trees, build one-time anonymous labels and form data binding, and construct weighted anonymous data forms for one-time data collection; Dynamic heterogeneous anonymous data distribution module: Used for data distribution, it generates various types of weighted anonymous data forms from one-time data collection, pre-categorizes them, and distributes them to each role as dynamic heterogeneous data forms. The Dynamic Heterogeneous Anonymous Data Statistics Module is used to collect and summarize the results of the distributed dynamic heterogeneous data forms. It identifies the role and weight of each data sample, constructs a statistical tree, performs secondary mean processing on data samples with the same role and weight, and finally summarizes and sums all the results to obtain the statistical results of the dynamic heterogeneous anonymized data.

6. The dynamic heterogeneous anonymized data deconstruction system according to claim 5, characterized in that, The dynamic heterogeneous anonymous data construction module can also be divided according to the role of the data collection object, and divided into multiple levels according to the role category, and construct a hierarchical role tree with the role category level as the root node. It can also construct a one-to-one weighted tree based on the role tree.

7. The dynamic heterogeneous anonymized data analysis system according to claim 5, wherein the dynamic heterogeneous anonymized data statistics module can also aggregate data in the same table, traverse each table, extract the role, weight and data of each table, and construct a statistical data structure based on the three types of information; and perform secondary equal weighting average processing, summate the statistics of the same table, and obtain the statistical results of heterogeneous anonymized data.

Images