Power grid data-based responsibility affiliation identification method and system

Abstract

The invention discloses a responsibility attribution identification method and system based on power grid data, and the method comprises the steps: collecting a system user table and a responsibility identification table based on a power grid system database, and carrying out the field standardization preprocessing of the system user table and the responsibility identification table; structural features, semantic features and business rule features are extracted from the preprocessed system user table and responsibility recognition table, a dynamic bidirectional A * algorithm is adopted to construct an association path between the system user table and the responsibility recognition table, and data owner access logic is obtained; the method comprises the steps of collecting system audit logs in real time, segmenting the system audit logs according to a time window to calculate time sequence behavior similarity, preferentially selecting system audit log fields with the time sequence behavior similarity higher than a threshold value to perform user table field matching, and obtaining a data owner liability list through user table field matching and data owner access logic. According to the method, the problems of low efficiency, poor accuracy and incapability of dynamically tracking data change responsibility of manual affirmation of power grid data responsibility are solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of data management, and particularly relates to a method and system for identifying the responsibility attribution based on power grid data. Background Art

[0002] With the deepening of digital transformation, the importance of data management has become more prominent, and the application of emerging technologies has provided more tools and methods for data management. The progress of technology has promoted the development of the big data industry, and the technologies in various links, including data collection, cleaning, storage, analysis, and visualization, have been continuously innovated. The technological development in the data responsibility attribution industry is manifested in the continuous advancement of technological innovation, the rapid growth of the data management market for AI, the wide application of cryptography technology, the gradual development of the DataOps system, the popularization of the engineering construction thinking, the increasing importance of the business data map, the obvious differentiation of industry differences, and the deep integration with digital transformation. These development trends indicate that the data management industry is moving towards a more intelligent, automated, and business-driven direction.

[0003] However, there are still some problems in the existing data management technologies: First, the division of data management responsibilities is unclear. After the informatization construction of many enterprises, the data management responsibilities are concentrated in a certain department. As a result, when data quality problems occur, the responsibility investigation often focuses on this department. However, the definition, entry, and use of data involve multiple links, and the collaborative role of each department is crucial in data governance. The actual responsibilities should be clearly divided and jointly borne within the entire enterprise. Second, data management is disconnected from business, and the internal driving force of data management needs to be improved. Management lag and lack of evaluation have become the biggest bottleneck restricting data management capabilities. Third, the data management technology system is insufficient. Although technological progress has enhanced the operability of data management, the dependence on manual work for data responsibility recognition is still relatively strong, and automation and real-time performance still need to be improved.

[0004] Chinese Patent with Publication No. CN112507374A discloses a data responsibility management system, including a responsibility management module for allocating personnel for the creation, access, acquisition, and use permissions of data; a data governance module for constructing a data management mechanism and managing the creation, access, acquisition, and use of data; a first governance module for supervising the consistency between the enterprise IT infrastructure and IT processes; a second governance module for displaying data; and a warehouse governance module for performing data interaction with the first governance module and supporting the data delivery of the data governance module through the IT infrastructure. This invention uses static permission allocation and relies on manual work, unable to perform automated associated responsibilities, and is prone to responsibility mismatches due to departmental adjustments. Summary of the Invention

[0005] The object of the present invention is to provide a method and system for identifying the responsibility attribution based on grid data, so as to solve the problems of low efficiency, poor accuracy in the manual determination of the responsibility attribution of grid data and the inability to dynamically track the responsibility for data changes.

[0006] The technical solution of the present invention is as follows:

[0007] On the one hand, the present invention provides a method for identifying the responsibility attribution based on grid data, including the following steps:

[0008] Collect the system user table and the responsibility recognition table based on the grid system database, and perform field standardization preprocessing on the metadata and detailed data of the system user table and the responsibility recognition table. The system user table is used to store user information, and the responsibility recognition table is used to store the responsibility attribution.

[0009] Extract the structural features, semantic features and business rule features from the preprocessed system user table and responsibility recognition table, and use the dynamic bidirectional A * algorithm to construct the association path between the system user table and the responsibility recognition table, and obtain the data owner data extraction logic including the data owner identifier, field association relationship and association path.

[0010] Collect the system audit logs in real time, split the system audit logs according to the time window to calculate the time series behavior similarity, preferentially select the system audit log fields with the time series behavior similarity higher than the threshold for user table field matching, and obtain the data owner responsibility list through the user table field matching and the data owner data extraction logic.

[0011] Preferably, extracting the structural features, semantic features and business rule features from the preprocessed system user table and responsibility recognition table specifically includes:

[0012] Calculate the structural features based on the type matching degree:

[0013]

[0014] In the formula, S struct is the type matching degree.

[0015] Convert the system user table and the responsibility recognition table into vector representations through the word vector model, and calculate the semantic feature S sem for the vector representations through the cosine similarity.

[0016] Calculate the business rule features based on the business rule matching degree:

[0017]

[0018] In the formula, S biz is the business rule matching degree.

[0019] Preferably, use the dynamic bidirectional A *The specific method for constructing the association path between the user table and the responsibility table in the algorithm is as follows:

[0020] Quantify the extracted structural features, semantic features, and business rule features as the edge weights of the association path. The edge weights are defined as:

[0021] W edge = λ1·S struct + λ2·S sem ·S biz

[0022] In the formula, W edge is the edge weight of the association path; λ1 is the structural feature weight; λ2 is the composite weight of semantics and rules.

[0023] Define the heuristic function as:

[0024]

[0025] In the formula, h(n) is the heuristic function; α is the weight gap term coefficient; W max is the maximum weight value of a single edge in the explored path; W current is the sum of the weights of all edges on the path from the starting point to the current node; D remaining is the remaining estimated number of nodes in the path; β is the rule dynamic term coefficient; is the rate of change of the business rule matching degree over time.

[0026] Start a forward search for the association path from the user table fields and expand nodes according to the priority of h(n).

[0027] Start a backward search for the association path from the responsibility table fields and expand nodes according to the priority of 1 / h(n).

[0028] When the edge weight error of the bidirectional search path is less than the set threshold, it is determined as a valid association path.

[0029] Conduct a Monte Carlo tree search on the candidate paths, simulate N path selections, and retain the optimal solution of the association path.

[0030] The data owner data retrieval logic includes the data owner identifier, field association relationship, and association path.

[0031] Preferably, the edge weight error of the bidirectional search path is defined as:

[0032] ΔW = |W forward + W backward - W meet |

[0033] In the formula, ΔW is the edge weight error of the bidirectional search path; W forwardis the cumulative weight of the forward search to the encounter node; W backward is the cumulative weight of the reverse search to the encounter node; W meet is the theoretical weight of the encounter edge.

[0034] Preferably, the system audit logs are segmented by time window to calculate the temporal behavior similarity, specifically: the system audit logs are segmented by time window, and the temporal behavior similarity is calculated for the field changes within each time window;

[0035] The calculation of the temporal behavior similarity is expressed as:

[0036] Similarity(A,B)=ω1·cos(vec(A),vec(B))+ω2·TS(A,B)

[0037] In the formula, Similarity(A,B) is the temporal behavior similarity between field A and field B; ω1 is the weight coefficient of the literal feature; cos() is the cosine similarity; vec(A) and vec(B) are the vector representations of field A and field B respectively; ω2 is the weight coefficient of the temporal behavior feature; TS(A,B) is the temporal similarity between field A and field B.

[0038] The temporal similarity is defined as:

[0039]

[0040] In the formula, T is the total number of time windows, t = 1, 2, …, T; ω t is the time decay weight; is the operation vector of field A within time window t; is the operation vector of field B within time window t.

[0041] Preferably, the data owner responsibility list is obtained through user table field matching and data owner data fetching logic, specifically:

[0042] Perform user table field matching to map the system audit logs to the user table, and use the association path between the user table and the responsibility recognition table in the data owner data fetching logic to obtain the association path between the system audit logs and the responsibility recognition table, and output the data owner responsibility list including the operation responsible person identifier, the operation data of the audit logs, and the association path.

[0043] On the other hand, the present invention provides a responsibility attribution recognition system based on power grid data, including a data preprocessing module, a feature extraction and association path construction module, and a responsibility list generation module.

[0044] A data preprocessing module, which is used to collect the system user table and the responsibility table based on the power grid system database, and perform field standardization preprocessing on the metadata and detailed data of the system user table and the responsibility table. The system user table is used to store user information, and the responsibility table is used to store the responsibility attribution.

[0045] A feature extraction and association path construction module, which is used to extract structural features, semantic features, and business rule features from the preprocessed system user table and responsibility table, and use the dynamic bidirectional A * algorithm to construct the association path between the system user table and the responsibility table, and obtain the data owner data fetching logic including the data owner identifier, field association relationship, and association path.

[0046] A responsibility list generation module, which is used to collect the system audit logs in real time, split the system audit logs according to the time window to calculate the temporal behavior similarity, preferentially select the system audit log fields with temporal behavior similarity higher than the threshold for user table field matching, and obtain the data owner responsibility list through user table field matching and the data owner data fetching logic.

[0047] On the other hand, the present invention also provides an electronic device, which includes: a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, it implements the responsibility attribution recognition method based on power grid data according to any embodiment of the present invention.

[0048] On the other hand, the present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the responsibility attribution recognition method based on power grid data according to any embodiment of the present invention.

[0049] Compared with the prior art, the present invention has the following technical effects:

[0050] 1. Through the dynamic bidirectional A * algorithm combined with multi-modal features (structure, semantics, business rules), the present invention automatically constructs the optimal association path between the user table and the responsibility table, realizes the automatic recognition of the full amount of field relationships between the system user table and the responsibility table, can quickly and accurately locate the association relationship between data, provides a clear data owner traceability path for users, and greatly improves the efficiency of data owner responsibility management.

[0051] 2. Through the analysis of audit logs based on temporal behavior similarity (TS), the present invention can capture high-frequency key operations in real time, preferentially match the responsible persons to output the data owner responsibility list, and at the same time deeply mine the audit log information, extract the page form fields in the log by intelligent means, and automatically match the relationship between the page form fields and the fields of the responsibility recognition form, providing a more refined data owner responsibility list for users and ensuring the timeliness and accuracy of responsibility tracking and audit analysis in the power grid system. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 is the overall flowchart of the method for identifying responsibility attribution based on power grid data according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0053] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with specific embodiments of the present application and with reference to the accompanying drawings.

[0054] Embodiment 1

[0055] This embodiment provides a method for identifying responsibility attribution based on power grid data. Referring to Figure 1 as shown, the method includes the following steps:

[0056] Collect the system user table and the responsibility recognition table from the power grid system database, and perform field standardization preprocessing on the metadata and detailed data of the system user table and the responsibility recognition table. The system user table is used to store user information, and the responsibility recognition table is used to store responsibility attribution. Specifically, the system user table is an independent entity table for storing user identity information, and the fields include user ID, department, role, etc. The responsibility recognition table is a relational table for recording data responsibility attribution and depends on the foreign key of the user table. The metadata is data that describes data and includes table name, field name, field type, field description, etc. The field standardization preprocessing is to unify the format of field descriptions.

[0057] Extract the structural features, semantic features, and business rule features from the preprocessed system user table and responsibility recognition table. The structural features are the physical attributes of the fields (such as type, length, and whether it is null); the semantic features are the implicit meanings of the field names and descriptions; the business rule features are the data governance rules defined by the power system organization. Use the dynamic two-way A * algorithm to construct the association path between the system user table and the responsibility recognition table, and obtain the data owner data fetching logic including the data owner identifier, field association relationship, and association path.

[0058] As a preferred implementation manner of this embodiment, the extraction of the structural features, semantic features, and business rule features from the preprocessed system user table and responsibility recognition table is specifically as follows:

[0059] Calculate structural features based on type matching degree:

[0060]

[0061] In the formula, S struct is the type matching degree.

[0062] Convert the system user table and the responsibility table into vector representations through the word vector model, and calculate the semantic feature S through the cosine similarity of the vector representations sem . The selection of the word vector model is not restricted here and can be selected according to the actual application scenario, such as using the Word2Vec or BERT model.

[0063] Calculate business rule features based on business rule matching degree:

[0064]

[0065] In the formula, S biz is the business rule matching degree. The number of rule hits is the number of times the business rule is satisfied in the data. The following is a simple example: If a business rule is defined as "the department code (DeptCode) in the user table must be mapped to the responsible department (RespDept) in the responsibility table", then the number of hits is the number of records that simultaneously meet the following conditions: the DeptCode field in the user table is not empty; the RespDept field in the responsibility table is not empty; the values of the two fields match logically.

[0066] As a preferred implementation method of this embodiment, the dynamic bidirectional A * algorithm is used to construct the association path between the system user table and the responsibility table. Specifically:

[0067] Quantify the extracted structural features, semantic features, and business rule features as the edge weights of the association path. The edge weights are defined as:

[0068] W edge = λ1·S struct + λ2·S sem ·S biz

[0069] In the formula, W edge is the edge weight of the association path; λ1 is the structural feature weight, calculated through historical data (such as the mean of the structural features of the correct path / the mean of the wrong path); λ2 is the composite weight of semantics and rules.

[0070] λ2 is used to nonlinearly strengthen the influence of business rules, and its calculation formula is expressed as:

[0071]

[0072] Define the heuristic function as:

[0073]

[0074] Wherein, h(n) is the heuristic function; α is the weight gap term coefficient; W max is the maximum weight value of a single edge in the explored path; W current is the sum of the weights of all edges on the path from the starting point to the current node; D remaining is the remaining estimated number of nodes in the path, obtained by calculating through the metadata level; β is the rule dynamic term coefficient; is the rate of change of the business rule matching degree with time, obtained by calculating through the audit log sliding window (such as taking the slope of the number of business rule hits in the audit log in the most recent 1 hour).

[0075] Search for the associated path forward from the user table field, and expand the nodes according to the priority of h(n).

[0076] Search for the associated path backward from the liability recognition table field, and expand the nodes according to the priority of 1 / h(n).

[0077] When the path edge weight error of the bidirectional search is less than the set threshold, it is determined as a valid associated path.

[0078] Perform Monte Carlo tree search on the candidate path, simulate N path selections, and retain the optimal solution of the associated path.

[0079] The data owner data acquisition logic includes the data owner identifier (used to clarify the responsible person (i.e., the data owner) of each data record, usually represented by the user ID, user name or other unique identifier), the field association relationship (clarify which fields are used to determine the data owner) and the associated path (clarify how to find the data owner through field association).

[0080] As a preferred implementation manner of this embodiment, the path edge weight error of the bidirectional search is defined as:

[0081] ΔW = |W forward + W backward - W meet |

[0082] Wherein, ΔW is the path edge weight error of the bidirectional search; W forward is the cumulative weight of the forward search to the meeting node; W backward is the cumulative weight of the backward search to the meeting node; W meet is the theoretical weight of the meeting edge, and the theoretical weight is calculated in advance for each pair of fields that may be associated through the metadata.

[0083] Specifically, the implementation steps of the Monte Carlo tree are specifically as follows:

[0084] Starting from the root node of the candidate path, select the child node with the highest weight; expand the unexplored edges; randomly simulate N path selections and record the proportion of successfully reaching the target; update the node weights according to the simulation results; retain the optimal solution of the associated path as the main associated path and the sub-optimal solution as the alternative associated path.

[0085] Real-time collect system audit logs, split the system audit logs by time window to calculate the temporal behavior similarity, and preferentially select the system audit log fields with temporal behavior similarity higher than the threshold for user table field matching to preferably ensure its responsibility traceability, and obtain the data owner responsibility list through user table field matching and the data owner data fetching logic. Specifically, the audit logs include recorded page form data, including but not limited to: user information fields, such as user ID, user name, user role, user department, etc. Operation information fields, such as operation time, operation type, operation object, operation result, etc. Form field information, such as form ID, form name, form field name, old field value, new field value, field type, etc. System environment information, such as IP address, device information, session ID, etc. Business context information, such as business module, business process ID, business rules, etc. Security-related information, such as permission level, operation approval status, operation log ID, etc.

[0086] As a preferred implementation manner of this embodiment, splitting the system audit logs by time window to calculate the temporal behavior similarity is specifically: splitting the system audit logs by time window and calculating the temporal behavior similarity for the field changes within each time window;

[0087] The calculation of the temporal behavior similarity is expressed as:

[0088] Similarity(A,B)=ω1·cos(vec(A),vec(B))+ω2·TS(A,B)

[0089] In the formula, Similarity(A,B) is the temporal behavior similarity between field A and field B; ω1 is the weight coefficient of the literal feature, determined by grid search; cos() is the cosine similarity; vec(A) and vec(B) are the vector representations of field A and field B respectively; ω2 is the weight coefficient of the temporal behavior feature; TS(A,B) is the temporal similarity between field A and field B.

[0090] The temporal similarity is defined as:

[0091]

[0092] In the formula, T is the total number of time windows, t = 1, 2, …, T; ω t is the time decay weight, the closer the operation is, the higher ω t is, ω t =e-ηt , where η is the attenuation rate, which can be set according to actual needs, such as set to 0.1; is the operation vector of field A within the time window t; is the operation vector of field B within the time window t.

[0093] As a preferred implementation manner of this embodiment, the specific method for obtaining the data owner responsibility list through user table field matching and data owner data fetching logic is as follows:

[0094] Perform user table field matching to map the system audit log to the user table, use the association path between the user table and the responsibility recognition table in the data owner data fetching logic to obtain the association path between the system audit log and the responsibility recognition table, perform dynamic responsibility matching, and output the data owner responsibility list including the operation responsible person identifier, the operation data of the audit log, and the association path. Further, the change impact analysis can be performed on the page form information of the audit log, comparing the old value and the new value of the field, and marking sensitive changes.

[0095] Embodiment Two

[0096] Correspondingly, this embodiment provides a responsibility attribution recognition system based on power grid data. The system is used to implement the responsibility attribution recognition method based on power grid data as described in Embodiment One of the present invention, and includes a data preprocessing module, a feature extraction and association path construction module, and a responsibility list generation module.

[0097] The data preprocessing module is used to collect the system user table and the responsibility recognition table based on the power grid system database, and perform field standardization preprocessing on the metadata and detailed data of the system user table and the responsibility recognition table. The system user table is used to store user information, and the responsibility recognition table is used to store responsibility attribution.

[0098] The feature extraction and association path construction module is used to extract structural features, semantic features, and business rule features from the preprocessed system user table and responsibility recognition table, and use the dynamic bidirectional A * algorithm to construct the association path between the system user table and the responsibility recognition table, and obtain the data owner data fetching logic including the data owner identifier, field association relationship, and association path.

[0099] The responsibility list generation module is used to collect the system audit log in real time, split the system audit log by time window to calculate the temporal behavior similarity, preferentially select the system audit log fields with temporal behavior similarity higher than the threshold for user table field matching, and obtain the data owner responsibility list through user table field matching and data owner data fetching logic.

[0100] Embodiment Three

[0101] This embodiment provides an electronic device, which includes: a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, it implements the method for identifying the responsibility attribution based on grid data as described in Embodiment 1 of the present invention.

[0102] Embodiment 4

[0103] This embodiment provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the method for identifying the responsibility attribution based on grid data as described in Embodiment 1 of the present invention.

[0104] In the embodiments of the present application, "at least one" means one or more, and "a plurality" means two or more. "And / or" describes the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent the cases of A existing alone, A and B existing simultaneously, and B existing alone. Where A and B can be singular or plural. The character " / " generally indicates that the associated objects before and after are in an "or" relationship. "At least one of the following" and its similar expressions refer to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b, and c can represent: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, and c can be single or multiple.

[0105] Those of ordinary skill in the art can realize that the units and algorithm steps described in the embodiments disclosed herein can be implemented by a combination of electronic hardware, computer software, and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. A professional technician can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present application.

[0106] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be described in detail here.

[0107] In several embodiments provided by the present application, if any function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM), random access memories (RAM), magnetic disks, or optical discs.

[0108] The above are only the embodiments of the present invention, and thus do not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall be equally included in the patent protection scope of the present invention.

Claims

1. A method for identifying liability attribution based on power grid data, characterized in that, The following steps are involved: Based on the power grid system database, a system user table and a responsibility table are collected, and field standardization preprocessing is performed on the metadata and detailed data of the system user table and the responsibility table. The system user table is used to store user information, and the responsibility table is used to store responsibility attribution. Extract structural features, semantic features, and business rule features from the preprocessed system user table and responsibility table, and use the dynamic bidirectional A * algorithm to construct the association path between the system user table and the responsibility table, and obtain the data owner data extraction logic including the data owner identifier, field association relationship, and association path; Collect system audit logs in real time, segment them by time window to calculate the similarity of temporal behavior, prioritize system audit log fields with temporal behavior similarity higher than the threshold for user table field matching, and obtain the data owner responsibility list through user table field matching and data owner data retrieval logic.

2. The method for identifying liability attribution based on power grid data according to claim 1, wherein The structural features, semantic features, and business rule features extracted from the pre-processed system user table and responsibility table are as follows: Calculate structural features based on type matching: Where S struct is the type matching degree; Convert the system user table and the responsibility recognition table into vector representations through the word vector model, and calculate the semantic feature S for the vector representations through cosine similarity sem ; Calculate business rule characteristics based on business rule matching degree: In the formula, S biz The business rule matching degree.

3. The method for identifying responsibility based on power grid data according to claim 2, characterized in that: Adopt dynamic two-way A * The specific method for constructing the association path between the system user table and the responsibility recognition table by the algorithm is as follows: The extracted structural features, semantic features, and business rule features are quantified into edge weights of the associated paths, where the edge weights are defined as: W edge = λ1·S struct + λ2·S sem ·S biz Where, W edge is the edge weight of the associated path; λ1 is the structural feature weight; λ2 is the composite weight of semantics and rules; The heuristic function is defined as: where h(n) is the heuristic function; α is the weight difference term coefficient; W max is the maximum weight value of a single edge in the explored path; W current is the sum of the weights of all edges on the path from the starting point to the current node; D remaining is the remaining estimated number of nodes in the path; β is the coefficient of the rule dynamics term; is the rate of change of the business rule matching degree over time; Starting from the user table field, forward search the associated path and expand the node according to the h(n) priority; Reversely search the associated path from the fields in the responsibility table and expand the nodes according to the priority of 1 / h(n); When the path edge weight error of the bidirectional search is less than the set threshold, it is determined to be a valid association path; Perform Monte Carlo tree search on candidate paths, simulate N path selections, and retain the optimal solution of the associated path; The output data master retrieval logic includes data master identification, field association relationship and association path.

4. The method for identifying liability attribution based on power grid data according to claim 1, wherein The path edge weight error of the bidirectional search is defined as: ΔW = |W forward + W backward - W meet | Where ΔW is the path edge weight error of bidirectional search; W forward is the cumulative weight of the forward search to the meeting node; W backward is the cumulative weight for reverse searching to the encounter node; W meet is the theoretical weight of the encounter edge.

5. The method for identifying responsibility based on power grid data according to claim 1, characterized in that: Segmenting the system audit log by time window to calculate the temporal behavior similarity is as follows: Segmenting the system audit log by time window and calculating the temporal behavior similarity for the field changes within each time window; The temporal behavior similarity calculation is expressed as: Similarity(A,B)=ω1·cos(vec(A),vec(B))+ω2·TS(A,B) Where Similarity(A,B) is the temporal behavior similarity between field A and field B; ω1 is the weight coefficient of the literal feature; cos() is the cosine similarity; vec(A) and vec(B) are the vector representations of fields A and B respectively; ω2 is the weight coefficient of the temporal behavior feature; TS(A,B) is the temporal similarity of fields A and B; The temporal similarity is defined as: Wherein, T is the total number of time windows, and t = 1, 2, …, T; ω t is the time decay weight; is the operation vector of field A within the time window t; is the operation vector of field B within the time window t.

6. The method for identifying liability attribution based on power grid data according to claim 1, characterized in that The data owner responsibility list obtained through user table field matching and data owner data retrieval logic is as follows: Perform user table field matching to map the system audit log to the user table, and use the association path between the user table and the responsibility table in the data master retrieval logic to obtain the association path between the system audit log and the responsibility table. Output the data master responsibility list including the operation responsible person identification, audit log operation data and association path.

7. A responsibility identification system based on power grid data, characterized in that: The system is used to implement the responsibility attribution identification method based on power grid data according to any one of claims 1 to 6, comprising a data preprocessing module, a feature extraction and association path construction module, and a responsibility list generation module; A data preprocessing module, which is used to collect the system user table and the responsibility table based on the power grid system database, and perform field standardization preprocessing on the metadata and detailed data of the system user table and the responsibility table. The system user table is used to store user information, and the responsibility table is used to store responsibility attribution; Feature extraction and association path construction module, which is used to extract structural features, semantic features and business rule features from the preprocessed system user table and responsibility recognition table, and adopts a dynamic bidirectional A * algorithm to construct the association path between the system user table and the responsibility recognition table, and obtain the data owner data extraction logic including the data owner identifier, field association relationship and association path; A responsibility list generation module, which is used to collect system audit logs in real time, split the system audit logs according to a time window to calculate the temporal behavior similarity, preferentially select the system audit log fields with a temporal behavior similarity higher than a threshold for user table field matching, and obtain a data owner responsibility list through user table field matching and data owner data fetching logic.

8. An electronic device, the electronic device comprising: A memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor implements the responsibility attribution recognition method based on power grid data according to any one of claims 1 to 6 when executing the computer program.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the responsibility attribution recognition method based on power grid data according to any one of claims 1 to 6.

Images