Human resource data security sharing method and system

Abstract

The invention discloses a human resource data security sharing method and system, and particularly relates to the field of human resource data security, and the method comprises the steps: collecting a behavior signal sequence of a user on an identity authentication platform of human resources, the behavior signal sequence comprising a page access path, an interactive operation type, time rhythm distribution and an equipment switching event; segmenting the behavior signal sequence into continuous time sequence operation fragment groups according to a unified time sliding window mechanism; and performing feature coding processing on each time sequence operation fragment group, extracting operation frequency, rhythm change, equipment fingerprint and interface jump features, and constructing a corresponding behavior embedding vector block. By constructing a redundant field screening mechanism based on track credibility and behavior stability dynamic evolution, the mode that a traditional platform only relies on field format legality verification is broken, an attacker is effectively prevented from constructing a pseudo-legal identity to be mixed into a sharing system, and the problems of system misjudgment and real identity misrecognition caused by reverse utilization of a redundant mechanism are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of human resource data security, and more particularly to a human resource data security sharing method and system. BACKGROUND

[0002] In the process of sharing human resource data, if the platform only relies on redundant field combination for identity verification and lacks dynamic management and control of the credibility of redundant paths, it may be exploited by attackers to induce the system to make false judgments by constructing a "format legal" fake identity.

[0003] This not only causes false identities to mix into shared data, but also causes false identities to be misidentified and banned, thereby causing the redundancy mechanism to be exploited in reverse, the verification system to fail, and ultimately causing data security risks at the platform level. SUMMARY

[0004] In order to overcome the above-mentioned defects of the prior art, embodiments of the present application provide a human resource data security sharing method and system, which establishes a two-way closed loop link from behavior path to field output to verification inversion through a redundancy field generation and screening mechanism based on the combination of behavior trajectory credibility evolution and field reversible verification, to solve the problems of data forgery mixing, identity misidentification and verification system failure caused by the lack of path dynamics and source credibility modeling of the field verification mechanism in the background art.

[0005] To achieve the above object, the present application provides the following technical scheme: a human resource data security sharing method, comprising:

[0006] Collecting a behavior signal sequence of a user on a human resource identity authentication platform, the behavior signal sequence including a page access path, an interaction operation type, a time rhythm distribution and a device switching event;

[0007] Dividing the behavior signal sequence into continuous time sequence operation fragment groups according to a unified time sliding window mechanism; performing feature encoding processing on each group of time sequence operation fragment groups to extract operation frequency, rhythm change, device fingerprint and interface jump features, and constructing a corresponding behavior embedding vector block;

[0008] Taking the behavior embedding vector block as a node of a directed graph, establishing a behavior transition edge between nodes according to time sequence, and constructing a preliminary behavior path graph;

[0009] Performing node mode similarity calculation on a plurality of behavior path graphs, aggregating behavior path graphs with similar structures and uniformly rearranging them into a standardized behavior trajectory structure graph;

[0010] All standardized behavior trajectory structure maps are aggregated, combined with user identification, platform source and time period information, and a trajectory identity matrix index set is constructed to support unified retrieval of global trajectory data;

[0011] By reading the trajectory identity matrix index set, path credibility and behavior stability information are extracted, and trajectory encoding block groups, platform embedding vectors, redundancy mapping vectors and candidate field sets are generated in turn, and finally the behavior-driven redundancy field group is screened to complete the consistency mapping of trajectory to field space.

[0012] In a preferred embodiment, each trajectory path in the trajectory identity matrix index set is read, the jump frequency, behavior period and stability of each trajectory path are counted, and a trajectory credibility vector is generated for path trust evaluation;

[0013] The trajectory credibility vector is input into the path stability analyzer, combined with the trajectory entropy value and periodicity index, and a trajectory encoding block group with consistent structure is constructed; the trajectory encoding block group and the platform preference factor are fused to generate a platform embedding vector reflecting the platform context behavior characteristics; the platform embedding vector is executed redundancy mapping to complete the path feature compression and semantic transformation, and a redundancy mapping vector for field mapping is generated; the redundancy mapping vector is decoded to a structured field space, and a candidate field set satisfying the structure rule is output;

[0014] The candidate field set is subjected to stability and disturbance screening to extract the behavior-driven redundancy field group evolved from the target confidence trajectory.

[0015] In a preferred embodiment, the behavior-driven redundancy field group submitted in the identity authentication request is received, the corresponding field hash feature is extracted, and a reverse index vector is constructed as a trajectory inversion entry parameter;

[0016] The trajectory path matching the reverse index vector is retrieved in the trajectory identity matrix index set, and a trajectory candidate path set with the same redundancy mapping mode is obtained; each trajectory candidate path is input into the field simulation reconstruction module, and the reverse mapping of the trajectory encoding block group to the field hash feature is performed to generate a simulation field sequence for verification.

[0017] In a preferred embodiment, the structural consistency between the simulation field sequence and the field group to be verified is compared, the coincidence rate of the simulation path and the historical trajectory path at the redundancy mapping vector layer is calculated, and a path matching score matrix is constructed;

[0018] If there is a unique target confidence score path in the path matching score matrix, mark the behavior-driven redundant field group as a trajectory reversible generation result and pass the verification; if all candidate paths in the path matching score matrix do not reach the confidence threshold, mark the behavior-driven redundant field group as an un-mappable combination, and trigger the field structure offset identification mechanism;

[0019] In the field structure offset identification mechanism, the mapping offset value and the structure deformation feature of the un-mappable field group are extracted, combined to construct a trajectory break feature vector set, which is used for subsequent identification of potential pseudo-constructed field groups in the trajectory evolution level.

[0020] In a preferred embodiment, all binding pairs between the verified trajectory paths in the trajectory identity matrix index set and the behavior-driven redundant field group are traversed, and the corresponding trajectory path set of the field group is extracted as the main index to construct the joint distribution matrix of the redundant mapping vector in the path space, so as to statistically analyze the frequency of the same redundant field group appearing in multiple trajectory paths and its time evolution distribution trend;

[0021] The joint distribution matrix is sliced according to consecutive time windows, and the aggregation intensity of the redundant field group on the trajectory subset is evaluated in each window. Combined with the path confidence rate change rate and the trajectory entropy fluctuation coefficient, the field aggregation dynamic surface across time periods is calculated to identify the field abnormal dense growth area.

[0022] In a preferred embodiment, the abnormal area with a local gradient change rate higher than a preset threshold is extracted on the field aggregation dynamic surface, and all trajectory path nodes involved in the abnormal area are tracked to form an initial set of pseudo-constructed trajectory paths. Then, the redundant field combination in the initial set of pseudo-constructed trajectory paths is compared for coincidence rate, and the target repeated field group with similar redundant mapping vectors appearing in multiple paths is selected to form a pseudo-constructed field clustering cluster.

[0023] In a preferred embodiment, each trajectory path in the pseudo-constructed field clustering cluster is subjected to a path evolution consistency process, the jump density, node entropy fluctuation, and behavior expression offset in the platform embedding vector in the trajectory encoding block group are analyzed dimension by dimension, a path-level structure offset measurement space is constructed, and the offset vector of each path in the space is extracted to form a pseudo-constructed path offset vector set.

[0024] The pseudo-constructed path offset vector set is subjected to feature clustering, and historical path evolution trajectories are introduced as positive example samples. By comparing the behavior stability and platform adaptability differences between the positive example trajectories and the abnormal paths, an offset abnormal evolution group is divided, and the corresponding redundant field group is marked as a trajectory break risk identity group, and the behavior output result of the marked group is marked as derived from a real trajectory.

[0025] In a preferred embodiment, the redundant field group binding path in the trajectory fracture risk identity group is taken as a frozen candidate path, and a multi-level access control measure is performed on the path in the trajectory identity matrix index set, the multi-level access control measure including three freezing strategies of temporary identity authentication block, path write channel suspension and field mapping output locking to block subsequent participation in the system field verification process;

[0026] Finally, the field generation frequency, trajectory inversion failure rate, trajectory path offset gradient and coincidence index of the frozen field group in the past cycle are summarized to construct a redundant mechanism dynamic credibility monitoring report, and the redundant mechanism dynamic credibility monitoring report is pushed to the system for correction to drive the trajectory to field mapping function weight update and offset tolerance adjustment mechanism execution cycle optimization.

[0027] A human resource data security sharing system comprises an acquisition module, a segmentation module, a construction module, an aggregation module, an index module and a mapping module.

[0028] The acquisition module is configured to acquire a behavior signal sequence of a user on an identity authentication platform of human resources, the behavior signal sequence comprising a page access path, an interactive operation type, a time rhythm distribution and a device switching event.

[0029] The segmentation module is configured to segment the behavior signal sequence into continuous time sequence operation segment groups according to a uniform time sliding window mechanism, perform feature coding processing on each group of time sequence operation segment groups, extract operation frequency, rhythm change, device fingerprint and interface jump features, and construct a corresponding behavior embedding vector block.

[0030] The construction module is configured to take the behavior embedding vector block as a node of a directed graph, establish a behavior transition edge between nodes according to time sequence, and construct a preliminary behavior path graph.

[0031] The aggregation module is configured to perform node mode similarity calculation on a plurality of behavior path graphs, aggregate behavior path graphs with similar structures, and uniformly rearrange the behavior path graphs into a standardized behavior trajectory structure graph.

[0032] The index module is configured to aggregate all standardized behavior trajectory structure graphs, combine user identification, platform source and time period information, construct a trajectory identity matrix index set to support unified retrieval of global trajectory data, and perform unified retrieval of global trajectory data.

[0033] The mapping module extracts path credibility and behavior stability information by reading the trajectory identity matrix index set, and generates a trajectory code block group, a platform embedding vector, a redundant mapping vector and a candidate field set in sequence, and finally selects a behavior-driven redundant field group to complete consistent mapping of the trajectory to the field space.

[0034] The technical effects and advantages of the present application are:

[0035] 1. By constructing a redundant field screening mechanism based on the dynamic evolution of trajectory credibility and behavior stability, breaking the traditional platform's way of checking only the legality of field format, effectively preventing attackers from constructing pseudo-legal identities to mix into the sharing system, solving the problem of system misjudgment and real identity misrecognition caused by reverse utilization of the redundant mechanism;

[0036] 2. By mapping the field group mapping results to the trajectory encoding layer in reverse and constructing a path scoring matrix to judge the unique matching path, the reversible mapping closed loop from "field results" to "real behavior path" is realized, thereby providing structural level traceability guarantee for identity authentication and improving the trustworthiness and verifiability of shared field groups;

[0037] 3. By constructing a field-path joint matrix using redundant mapping vectors and path co-occurrence distribution, and introducing a time sliding window mechanism to analyze the change trend of field aggregation intensity, the non-natural growth section of the field at different time periods is effectively identified, and the abnormal behavior signs of potential batch pseudo-structure or field collusion attacks are captured in advance;

[0038] 4. Based on the trajectory offset vector, jump density, node entropy fluctuation and platform embedding bias, a structure offset space is constructed, and combined with historical positive example trajectory execution clustering comparison, the abnormal path highly deviated in behavior expression is effectively divided, and then the broken identity field possibly evolved by non-real users is identified. BRIEF DESCRIPTION OF DRAWINGS

[0039] Figure 1 The method step flowchart of the present application.

[0040] Figure 2 The behavior signal collection and trajectory generation flowchart of the present application.

[0041] Figure 3 The field inversion and reversibility verification flowchart of the present application.

[0042] Figure 4 The field anomaly aggregation and pseudo-structure clustering identification flowchart of the present application.

[0043] Figure 5 The behavior offset analysis and high-risk identity identification flowchart of the present application.

[0044] Figure 6 The system module diagram of the present application. DETAILED DESCRIPTION

[0045] With reference to the drawings of the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the scope of the present application.

[0046] With reference to the drawings of the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the scope of the present application. Figures 1-6 An embodiment of the present application is a human resource data security sharing method, comprising:

[0047] Collecting a behavior signal sequence of a user on a human resource identity authentication platform, the behavior signal sequence comprising a page access path, an interactive operation type, a time rhythm distribution and a device switching event;

[0048] Dividing the behavior signal sequence into continuous time sequence operation fragment groups according to a unified time sliding window mechanism, each group of fragments of the time sequence operation fragment groups maintaining continuity and time sequence integrity of original behavior context; performing feature coding processing on each group of time sequence operation fragment groups, extracting operation frequency, rhythm change, device fingerprint and interface jump features, and constructing a corresponding behavior embedding vector block;

[0049] Taking the behavior embedding vector block as a node of a directed graph, establishing a behavior transition edge between nodes according to time sequence, and constructing a preliminary behavior path graph;

[0050] Performing node mode similarity calculation on a plurality of behavior path graphs, aggregating behavior path graphs with similar structures and uniformly rearranging them into a standardized behavior trajectory structure graph;

[0051] Summarizing all standardized behavior trajectory structure graphs, combining user identification, platform source and time period information, constructing a trajectory identity matrix index set to support unified retrieval of global trajectory data;

[0052] By reading the trajectory identity matrix index set, extracting path credibility and behavior stability information, and sequentially generating a trajectory coding block group, a platform embedding vector, a redundancy mapping vector and a candidate field set, a behavior-driven redundancy field group is finally screened to complete the consistency mapping of the trajectory to the field space;

[0053] Among them, the structured trajectory path data is extracted from the trajectory identity matrix, and after multi-stage processing (trajectory block extraction, platform preference fusion, semantic mapping, disturbance screening), a group of "behavior-driven redundancy field groups" are obtained, which will be used for subsequent data sharing identity verification;

[0054] P m =f 扰 (f 映 (f 嵌 (f块 (M 轨 ))));

[0055] wherein P m is the behavior-driven redundancy field group, the mth field group, the unit is the field sequence set, P m represents the final data output set filtered for identity authentication; f 扰 (·) is an anti-interference elimination function, which is used to analyze whether the structural deviation output caused by abnormal trajectory path is contained in the mapped field sequence, and internally scores each field group based on the disturbance gradient and removes those with high volatility; f 映 (·) is a field semantic space mapping function, which is used to project the behavior vector after platform fusion to the field encoding space to generate preliminary field candidate values; f 嵌 (·) is a platform context embedding function, which integrates the user platform preference, operation frequency and behavior period reflected in the trajectory path, with the unit being a high-dimensional vector for semantic retention; f 块 (M 轨 ) is a trajectory block generation function, which acts on the trajectory identity matrix to extract time-continuous behavior path segments (referred to as trajectory blocks), with the unit being a graph path segment set; M 轨 is the trajectory identity matrix index set, with the unit being a path graph set combined with a user identification tuple, used for global path retrieval.

[0056] Read each trajectory path in the trajectory identity matrix index set, count the jump frequency, behavior period and stability of each trajectory path, and generate a trajectory credibility vector for path trust evaluation;

[0057] Input the trajectory credibility vector into the path stability analyzer, combine the trajectory entropy value and periodicity index, and construct a trajectory encoding block group with consistent structure representation; perform fusion operation on the trajectory encoding block group and the platform preference factor to generate a platform embedding vector reflecting the platform context behavior characteristics; perform redundancy mapping on the platform embedding vector to complete path feature compression and semantic transformation, and generate a redundancy mapping vector for field mapping; decode the redundancy mapping vector to a structured field space, and output a candidate field set that meets the structure rule;

[0058] Perform stability and disturbance screening on the candidate field set to extract the behavior-driven redundancy field group evolved from the target confidence trajectory;

[0059] It should be noted that the trajectory path is modeled for credibility, so that the most reliable path is selected for field mapping; the core is: using the jump density, rhythm consistency, platform adaptability and other factors of behavior, construct a path credibility score vector, and generate a unique semantic mapping structure through embedding fusion;

[0060]

[0061] wherein is the behavioral confidence score value of the ith trajectory path, in unit of dimensionless confidence factor, is the transition density in unit of time (transitions per second) in the path, which is derived from the number of edge connections in the graph structure; is the behavioral rhythm sequence, in unit of seconds; represents the platform ID corresponding to the path i, in unit of platform identifier; φ 跳 (·) is the transition density to score function, which is a monotonically increasing function, used to evaluate the path complexity; φ 节 (·) is the rhythm smoothness score function, which is used to measure the rhythm consistency, and the score decreases with the deviation mutation; φ 设 (·) is the platform trust function, which is used to set the confidence coefficient according to the platform identity, such as self-owned system with high score.

[0062] Receiving the behavioral driving redundancy field group submitted in the identity authentication request, extracting the corresponding field hash feature, and constructing a reverse index vector as a trajectory inversion entry parameter;

[0063] Retrieving the trajectory path matched with the reverse index vector in the trajectory identity matrix index set, obtaining the trajectory candidate path set with the same redundancy mapping mode; inputting each trajectory candidate path into the field simulation reconstruction module, performing reverse mapping of the trajectory coding block group to the field hash feature, and generating a simulation field sequence for verification.

[0064] Comparing the structural consistency between the simulation field sequence and the field group to be verified, calculating the coincidence rate of the simulation path and the historical trajectory path in the redundancy mapping vector layer, and constructing a path matching score matrix;

[0065] Judging whether there is a unique target high-confidence score path in the path matching score matrix, if there is, marking the behavioral driving redundancy field group as a trajectory reversible generation result and passing the verification; if all candidate paths in the path matching score matrix do not reach the confidence threshold, marking the behavioral driving redundancy field group as an un-mappable combination, and triggering the field structure offset identification mechanism;

[0066] In the field structure offset identification mechanism, the mapping offset value and the structure deformation feature of the un-mappable field group are extracted, combined to construct a trajectory break feature vector set, which is used for subsequent identification of potential pseudo-structure field groups at the trajectory evolution level;

[0067] Where the field group is reflected to the simulation path after remapping, compared with the historical path score, if there is a unique high confidence path, it is verified, otherwise it is considered to be fake;

[0068]

[0069] Where Score i,j is the cosine similarity score of the ith simulation path and the jth historical trajectory, unit is dimensionless ∈ [-1, 1]; is the encoding vector of the simulation trajectory, unit is is the historical trajectory vector, unit is θ 置信 is the preset confidence threshold, unit is dimensionless; indicates the score value of the ith simulation path traversing all historical trajectories j, the maximum score is selected; this maximum score is used to judge whether there is a unique credible path; Valid represents the final trajectory reversible determination result, which is a Boolean variable, including 0 or 1.

[0070] Traverse all verified trajectory paths and binding pairs between behavior-driven redundant field groups in the trajectory identity matrix index set, and extract the corresponding trajectory path set according to the field group as the main index to construct the joint distribution matrix of the redundant mapping vector in the path space, so as to count the frequency of the same redundant field group appearing in multiple trajectory paths and its time evolution distribution trend;

[0071] Slice the joint distribution matrix according to the continuous time window, and perform the aggregation intensity evaluation of the redundant field group on the trajectory subset in each window, combine the path credibility change rate and the trajectory entropy fluctuation coefficient, and calculate the field aggregation dynamic surface across the time period to identify the field abnormal dense growth area.

[0072] Extract the abnormal area with local gradient change rate higher than the preset threshold on the field aggregation dynamic surface, and track all trajectory path nodes involved in the abnormal area to form the initial set of pseudo-constructed trajectory paths, and then perform coincidence rate comparison on the redundant field combination in the initial set of pseudo-constructed trajectory paths, and select the target repeated field group that appears similar redundant mapping vector in multiple paths to form the pseudo-constructed field clustering cluster.

[0073] Perform the path evolution consistency process on each trajectory path in the pseudo-constructed field clustering cluster, perform dimension-by-dimension analysis on the jump density, node entropy fluctuation, and behavior expression offset in the platform embedding vector in the trajectory encoding block group, construct the path-level structure offset measurement space, extract the offset vector of each path in the space, and combine to generate the pseudo-constructed path offset vector set;

[0074] Perform high-dimensional feature clustering on the pseudo-constructed path offset vector set, and introduce the historical path evolution trajectory as the positive example sample. By comparing the behavior stability and platform adaptability difference between the positive example trajectory and the abnormal path, the high offset abnormal evolution group is divided, and the corresponding redundant field group in the trajectory is marked as the trajectory fracture high-risk identity group, and the behavior output result may come from the real trajectory.

[0075] The redundant field group in the trajectory fracture high-risk identity group is bound to the path as the frozen candidate path, and the multi-level access control measures are performed on these paths in the trajectory identity matrix index set. The multi-level access control measures include: temporary identity authentication blockage, path writing channel suspension, field mapping output locking three freezing strategies to block its subsequent participation in the system field verification process.

[0076] Finally, the field generation frequency, trajectory inversion failure rate, trajectory path offset gradient and coincidence index of all frozen field groups in the past period in the redundant mapping space are summarized to construct the redundant mechanism dynamic trustworthiness monitoring report, and the redundant mechanism dynamic trustworthiness monitoring report is pushed to the system for correction, driving the trajectory to field mapping function weight update and offset tolerance adjustment mechanism execution cycle optimization.

[0077] Among them, by analyzing the path distribution density of the field group at different times, whether it appears abnormal aggregation, that is, whether it appears in a large number of unrelated trajectories in a short period of time, so as to judge whether it may be used by pseudo-constructed attacks;

[0078] Where D 共 (P m ,t) represents the redundant field group P m The trajectory distribution density at time t, the unit is path number / second; Indicates the derivative with respect to time, which is used to calculate the growth trend; ε 异常 is the aggregation anomaly threshold, the unit is density change rate.

[0079] Path offset vector analysis and clustering identification, the path set suspected of pseudo-constructed is converted into a high-dimensional offset vector, and whether it comes from a real behavior trajectory is identified by clustering method. If the difference with the historical behavior is large, it is classified as a high-risk pseudo-constructed trajectory;

[0080]

[0081] Where is the offset vector of the i-th path, and the offset vector represents the difference between the current encoding and the reference encoding, the unit is Euclidean distance (encoding space unit); is the behavior encoding vector of the current suspicious trajectory, the unit is is the historical or positive example trajectory vector; Ψ 轨For the path clustering result, the path clustering result is used to output a cluster center structure combined with a class to which each path belongs; Cluster(·) represents a cluster function of a trajectory offset vector space, and the cluster function of the trajectory offset vector space is used for cluster analysis on a set of offset vectors of all pseudo-constructed paths in a high-dimensional behavior encoding difference space, so as to identify whether there is an abnormal mode clustering trend in the behavior evolution process, thereby delimiting a suspicious trajectory break identity group.

[0082] A human resource data security sharing system comprises an acquisition module, a segmentation module, a construction module, an aggregation module, an index module and a mapping module.

[0083] The acquisition module is used to acquire a behavior signal sequence of a user on an identity authentication platform of human resources, and the behavior signal sequence comprises a page access path, an interactive operation type, a time rhythm distribution and a device switching event.

[0084] The segmentation module is used to segment the behavior signal sequence into continuous time sequence operation fragment groups according to a uniform time sliding window mechanism; feature encoding processing is performed on each group of time sequence operation fragment groups, operation frequency, rhythm change, device fingerprint and interface jump features are extracted, and a corresponding behavior embedding vector block is constructed.

[0085] The construction module is used to take the behavior embedding vector block as a node of a directed graph, establish a behavior transition edge between nodes according to time sequence, and construct a preliminary behavior path graph.

[0086] The aggregation module is used to perform node mode similarity calculation on a plurality of behavior path graphs, aggregate behavior path graphs with similar structures, and uniformly rearrange the behavior path graphs into a standardized behavior trajectory structure graph.

[0087] The index module is used to summarize all standardized behavior trajectory structure graphs, combine user identification, platform source and time period information, construct a trajectory identity matrix index set to support unified retrieval of global trajectory data, and construct a trajectory identity matrix index set to support unified retrieval of global trajectory data.

[0088] The mapping module extracts path credibility and behavior stability information by reading the trajectory identity matrix index set, and generates a trajectory encoding block group, a platform embedding vector, a redundant mapping vector and a candidate field set in sequence, and finally screens a behavior-driven redundant field group to complete consistent mapping of the trajectory to the field space.

[0089] The above only describes preferred embodiments of the present application and is not used to limit the present application, and any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for securely sharing human resource data, comprising: Collect user behavior signal sequences on the human resources identity authentication platform. The behavior signal sequences include page access paths, interaction operation types, time rhythm distribution, and device switching events. Its features are: The behavioral signal sequence is divided into continuous temporal operation segments according to a unified time sliding window mechanism; feature encoding processing is performed on each temporal operation segment to extract operation frequency, rhythm changes, device fingerprint and interface jump features, and construct corresponding behavioral embedding vector blocks. The behavior embedding vector blocks are used as nodes in a directed graph. Behavior transition edges between nodes are established according to time sequence to construct a preliminary behavior path graph. Calculate the node pattern similarity of multiple behavior path graphs, aggregate behavior path graphs with similar structures, and rearrange them into a standardized behavior trajectory structure graph. By aggregating all standardized behavioral trajectory structure maps and combining user identifiers, platform sources, and time period information, a trajectory identity matrix index set is constructed to support unified retrieval of global trajectory data. By reading the trajectory identity matrix index set, path credibility and behavior stability information are extracted, and trajectory coding block groups, platform embedding vectors, redundant mapping vectors and candidate field sets are generated in sequence. Finally, behavior-driven redundant field groups are selected to complete the consistent mapping of trajectory to field space.

2. The method for securely sharing human resource data according to claim 1, characterized in that: Read each trajectory path in the trajectory identity matrix index set, count the jump frequency, behavior cycle and stability of each trajectory path, and generate a trajectory credibility vector for path trust assessment; The trajectory credibility vector is input into the path stability analyzer, and combined with the trajectory entropy value and periodicity index, a trajectory coding block group with consistent structural representation is constructed. The trajectory coding block group is fused with the platform preference factor to generate a platform embedding vector that reflects the platform context behavior characteristics. The platform embedding vector is subjected to redundant mapping to complete the path feature compression and semantic transformation, generating a redundant mapping vector oriented towards field mapping. The redundant mapping vector is decoded into a structured field space to output a set of candidate fields that satisfy the structural rules. Stability and perturbation screening are performed on the candidate field set to extract the behavior-driven redundant field group obtained from the evolution of the target confidence trajectory.

3. The method for securely sharing human resource data according to claim 2, characterized in that: Receive the behavior-driven redundant field group submitted in the identity authentication request, extract the corresponding field hash features, and construct a reverse index vector as the trajectory inversion input parameter; The trajectory path matching the inverse index vector is retrieved from the trajectory identity matrix index set to obtain a set of trajectory candidate paths with the same redundant mapping pattern. Each trajectory candidate path is input into the field simulation reconstruction module to perform inverse mapping from trajectory coding block group to field hash feature to generate a simulation field sequence for verification.

4. The method for securely sharing human resource data according to claim 3, characterized in that: Compare the structural consistency between the simulation field sequence and the field group to be verified, calculate the overlap rate between the simulation path and the historical trajectory path in the redundant mapping vector layer, and construct the path matching score matrix. Determine whether there is a unique target confidence score path in the path matching score matrix. If it exists, mark the behavior-driven redundant field group as a trajectory reversible generation result and pass the verification. If none of the candidate paths in the path matching score matrix reach the confidence threshold, mark the behavior-driven redundant field group as an unmappable combination and trigger the field structure offset recognition mechanism. In the field structure offset identification mechanism, the mapping offset values ​​and structural deformation features of unmappable field groups are extracted and combined to construct a trajectory breakage feature vector set, which is used to identify potential pseudo-structure field groups at the trajectory evolution level.

5. A method for securely sharing human resource data according to claim 4, characterized in that: Traverse all verified trajectory paths and behavior-driven redundant field groups in the trajectory identity matrix index set, extract the corresponding trajectory path set by field group as the main index, construct the joint distribution matrix of redundant mapping vector in the path space, and use this to count the frequency of the same redundant field group in multiple trajectory paths and its temporal evolution distribution trend. The joint distribution matrix is ​​sliced ​​according to continuous time windows, and the aggregation intensity of redundant field groups on the trajectory subset is evaluated in each window. Combined with the path confidence change rate and trajectory entropy fluctuation coefficient, the dynamic surface of field aggregation across time periods is calculated to identify areas of abnormally dense field growth.

6. A method for securely sharing human resource data according to claim 5, characterized in that: Anomalies with local gradient change rates exceeding a preset threshold are extracted from the dynamic surface of field aggregation. All trajectory path nodes involved in the anomalies are tracked to form an initial set of pseudo-trajectory paths. Then, the overlap rate of redundant field combinations in the initial set of pseudo-trajectory paths is compared to filter out target repeating field groups that have similar redundant mapping vectors appearing in multiple paths, forming pseudo-field clusters.

7. A method for securely sharing human resource data according to claim 6, characterized in that: Each trajectory path in the pseudo-structural field cluster undergoes a path evolution consistency process. The jump density, node entropy fluctuation, and behavioral expression offset in the platform embedding vector within the trajectory coding block group are analyzed dimension by dimension. A path-level structural offset metric space is constructed, and the offset vectors of each path in this space are extracted and combined to generate a pseudo-structural path offset vector set. Feature clustering is performed on the pseudo-path offset vector set, and historical path evolution trajectories are introduced as positive examples. By comparing the behavioral stability and platform adaptability differences between positive examples and abnormal paths, an offset abnormal evolution group is divided, and its corresponding redundant field group is marked as the trajectory breakage risk identity group, indicating that its behavioral output results come from the real trajectory.

8. A method for securely sharing human resource data according to claim 7, characterized in that: The redundant field groups in the trajectory breakage risk identity group are bound to the path as the candidate path for freezing, and multi-level access control measures are implemented on these paths in the trajectory identity matrix index set. The multi-level access control measures include three freezing strategies: temporary identity authentication blocking, path writing channel suspension, and field mapping output locking, so as to prevent them from participating in the subsequent system field verification process. Finally, the frequency of field generation, trajectory inversion failure rate, trajectory path offset gradient and their coincidence in the redundancy mapping space of all frozen field groups in the past period are summarized to construct a redundancy mechanism dynamic credibility monitoring report. The redundancy mechanism dynamic credibility monitoring report is pushed to the system for correction, driving the execution cycle optimization of trajectory to field mapping function weight update and offset tolerance adjustment mechanism.

9. A human resources data secure sharing system, comprising a collection module, a segmentation module, a construction module, an aggregation module, an indexing module, and a mapping module, characterized in that: The data acquisition module is used to collect user behavior signal sequences on the human resources identity authentication platform. The behavior signal sequences include page access paths, interaction operation types, time rhythm distribution, and device switching events. The segmentation module is used to segment the behavior signal sequence into continuous time-series operation segments according to a unified time sliding window mechanism; for each group of time-series operation segments, feature encoding processing is performed to extract operation frequency, rhythm changes, device fingerprint and interface jump features, and construct the corresponding behavior embedding vector block; The building module is used to embed behavior into vector blocks as nodes in a directed graph, establish behavior transition edges between nodes according to time sequence, and construct a preliminary behavior path graph. The aggregation module is used to perform node pattern similarity calculation on multiple behavior path graphs, aggregate behavior path graphs with similar structures, and uniformly rearrange them into a standardized behavior trajectory structure graph. The index module is used to summarize all standardized behavioral trajectory structure maps, and combine user identifiers, platform sources and time period information to build a trajectory identity matrix index set to support unified retrieval of global trajectory data; The mapping module reads the trajectory identity matrix index set, extracts path credibility and behavior stability information, and sequentially generates trajectory coding block groups, platform embedding vectors, redundant mapping vectors and candidate field sets. Finally, it selects behavior-driven redundant field groups to complete the consistent mapping of trajectory to field space.

Images