Supervision submission system management method, device and equipment based on knowledge graph
By constructing a knowledge graph network and a front-end interactive interface, the problem of scattered information storage in the management of regulatory reporting systems has been solved, enabling efficient and accurate querying and analysis, and meeting users' needs for comprehensive understanding.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-27
- Publication Date
- 2026-04-14
AI Technical Summary
Existing technologies result in fragmented information storage in regulatory reporting systems, leading to complex, inefficient, and inaccurate queries that fail to meet users' needs for comprehensive understanding.
A knowledge graph network is constructed to manage regulatory reporting information through structured storage of nodes and relation edges. Standardized interactive function interfaces are provided on the front-end interface to receive user requests and execute them based on the knowledge graph network, enabling relation reasoning, path analysis, and entity attribute query.
It improves the efficiency and accuracy of the regulatory reporting system, reduces operational complexity, and meets users' needs for comprehensive understanding.
Smart Images
Figure CN121858644A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of financial regulatory technology, and in particular to a method, apparatus and equipment for managing a regulatory reporting system based on knowledge graphs. Background Technology
[0002] In today's digital age, various industries face the challenge of managing and processing massive amounts of information. Especially in the area of regulatory reporting systems, with the continuous refinement of regulations and the increasing complexity of business processes, how to efficiently integrate, analyze, and utilize relevant information has become a key challenge for improving regulatory effectiveness and ensuring compliant operations.
[0003] Currently, in the management of regulatory reporting systems, some existing technologies employ traditional database storage methods to manage regulatory reporting-related information. This information is scattered across different table structures, linked by simple primary and foreign key relationships. When users need to query certain information within the regulatory reporting system, such as the relationships between different entities, multi-table joins or complex recursive query algorithms are often required. This approach is not only complex and inefficient, but also lacks a deep understanding and reasoning ability regarding semantic relationships. It cannot mine potential relationships between entities based on business logic and domain knowledge, resulting in low query accuracy and failing to meet users' needs for a comprehensive understanding of the regulatory reporting system. Summary of the Invention
[0004] In view of this, this application provides a method, apparatus and equipment for managing regulatory reporting systems based on knowledge graphs, which can improve the query efficiency and accuracy of regulatory reporting systems and meet users' needs for a comprehensive understanding of regulatory reporting systems.
[0005] According to the first aspect of this application, a regulatory reporting system management method based on knowledge graphs is provided, comprising: Obtain the knowledge graph network constructed for the regulatory reporting system; Construct a front-end interactive interface, and provide interactive function interfaces for the knowledge graph network in the front-end interactive interface; Receives an interaction request input by the user through the interaction function interface, the interaction request including at least one of a relationship reasoning request, a path analysis request, and an entity attribute query request; The interaction request is executed based on the knowledge graph network, and the interaction execution result for the regulatory reporting system is output.
[0006] According to a second aspect of this application, a knowledge graph-based regulatory reporting system management device is provided, comprising: The acquisition module is used to acquire the knowledge graph network constructed for the regulatory reporting system; A building module is used to build a front-end interactive interface, which provides interactive function interfaces for the knowledge graph network. The receiving module is used to receive interactive requests input by the user through the interactive function interface, wherein the interactive requests include at least one of relation reasoning requests, path analysis requests, and entity attribute query requests; The execution module is used to execute the interaction request based on the knowledge graph network and output the interaction execution result for the regulatory reporting system.
[0007] According to a third aspect of this application, a storage medium is provided on which a computer program is stored, which, when executed by a processor, implements the above-described knowledge graph-based regulatory reporting system management method.
[0008] According to a fourth aspect of this application, an electronic device is provided, including a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor executes the program to implement the aforementioned knowledge graph-based regulatory reporting system management method.
[0009] By employing the aforementioned technical solutions, the knowledge graph-based regulatory reporting system management method, apparatus, and equipment provided in this application, through the construction of a knowledge graph network, structurally stores regulatory reporting system information in the form of nodes and relational edges, effectively solving the problem of complex queries caused by the scattered storage of information in existing technologies. An interactive interface for the knowledge graph network is provided on the front-end interface. This interface receives user-inputted interaction requests and executes them based on the knowledge graph network, determining the interaction execution result for the regulatory reporting system. Complex regulatory reporting system association analysis can be transformed into standardized command operations. Users do not need to master database joins or graph algorithm knowledge; they can trigger the semantic computing capabilities of the knowledge graph simply through commands such as relational reasoning, path analysis, and entity attribute queries. The graph structure storage of the knowledge graph allows direct traversal of the belonging and affiliation edges between entities during command execution, avoiding the complex operations of traditional multi-table joins in databases. This improves the response speed and accuracy of relational reasoning, path analysis, and entity attribute queries, completely solving the problems of low query efficiency and fragmented results caused by complex operations and lack of semantic reasoning capabilities in existing technologies. Ultimately, this meets the user's need for a comprehensive understanding of the regulatory reporting system.
[0010] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description
[0011] Figure 1This illustration shows a flowchart of a knowledge graph-based regulatory reporting system management method provided in an embodiment of this application. Figure 2 A flowchart illustrating a knowledge graph-based regulatory reporting system management method is shown in another embodiment of this application. Figure 3 This illustration shows a structural diagram of a knowledge graph-based regulatory reporting system management device provided in an embodiment of this application. Figure 4 This illustration shows a schematic diagram of another knowledge graph-based regulatory reporting system management device provided in an embodiment of this application. Detailed Implementation
[0012] The present application will be described in detail below with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in the embodiments of the present application can be combined with each other.
[0013] Currently, in the management of regulatory reporting systems, some existing technologies employ traditional database storage methods to manage regulatory reporting-related information. This information is scattered across different table structures, linked by simple primary and foreign key relationships. When users need to query certain information within the regulatory reporting system, such as the relationships between different entities, multi-table joins or complex recursive query algorithms are often required. This approach is not only complex and inefficient, but also lacks a deep understanding and reasoning ability regarding semantic relationships. It cannot mine potential relationships between entities based on business logic and domain knowledge, resulting in low query accuracy and failing to meet users' needs for a comprehensive understanding of the regulatory reporting system.
[0014] To address the aforementioned technical problems, embodiments of the present invention provide a regulatory reporting system management method based on knowledge graphs, such as... Figure 1 As shown, the method includes: Step 110: Obtain the knowledge graph network constructed for the regulatory reporting system.
[0015] In specific application scenarios, modeling can be performed on regulatory reporting documents, interface tables, and related asset materials to define key entity types such as regulatory agencies, business departments, reporting systems, interface tables, and fields, as well as relationship types such as attribution relationships (e.g., issuing, containing) and peer relationships (e.g., belonging to the same regulatory agency, belonging to the same business theme). Then, OCR technology and regular expressions are used to extract entity names, attribute values, and other information from the documents, and each entity is assigned a unique UID code. The entity, attribute, and relationship data are imported into graph databases such as Neo4j, and finally a knowledge graph network containing multiple nodes and relationship items is generated. This network stores the hierarchical associations and implicit relationships of the regulatory reporting system in a structured and semantic form.
[0016] For the embodiments disclosed herein, the knowledge graph network constructed for the regulatory reporting system can be directly invoked to execute the subsequent steps of the regulatory reporting system management process.
[0017] By generating a knowledge graph network corresponding to the regulatory reporting system, the information of the regulatory reporting system, which is traditionally stored in different table structures, can be transformed into a structured expression of nodes and relational edges in the knowledge graph network, thus completely solving the problem of information fragmentation. In this embodiment, the knowledge graph network constructed for the regulatory reporting system can be invoked. Through explicit attribution relationships (such as which agency issued the system and which fields it contains) and implicit affiliation relationships (such as the association of systems under the same regulatory agency), the complex multi-table query requirements can be simplified to graph path search, improving the efficiency of cross-system entity relation queries. The semantic modeling capability of the knowledge graph supports rule-based relational reasoning and path analysis, automatically mining potential relationships between entities. Simultaneously, the knowledge graph network provides a unified data foundation for subsequent interactive queries and visualization, enabling "one-click" acquisition of the entire system information chain, significantly reducing the operational complexity of regulatory system management, and meeting users' needs for comprehensive information, relevance, and efficient querying.
[0018] Step 120: Build a front-end interactive interface and provide interactive function interfaces for knowledge graph networks in the front-end interactive interface.
[0019] In this embodiment of the disclosure, front-end technologies such as HTML, CSS, and JavaScript can be used to design a visual user interface, and the complex functions of the knowledge graph network (such as relation reasoning, path analysis, and entity attribute query) can be encapsulated into standardized interfaces. Users can input search keywords or select command types (relation reasoning request, path analysis request, entity attribute query request) through interactive components such as filter boxes and command buttons in the interface. These operation requests are transmitted to the backend through the interactive function interface, triggering the knowledge graph network to execute the corresponding command, and the results are fed back to the front-end interface for display in the form of visual graphs, attribute lists, etc.
[0020] Through a visual interactive interface and standardized functional interfaces, users do not need to master professional knowledge such as graph database operation and algorithm writing. They can achieve in-depth analysis of regulatory reporting systems with just simple clicks and input operations. For example, they can query the implicit relationships between different systems or the complete regulatory chain of a certain indicator with one click, enabling non-technical personnel to obtain information efficiently. At the same time, compared with the text query results of traditional databases, this interactive information presentation is more three-dimensional and comprehensive, which can greatly improve the efficiency of obtaining regulatory system information and the analysis experience, and meet users' needs for comprehensive and convenient query of regulatory reporting systems.
[0021] Step 130: Receive the interaction request input by the user through the interactive function interface.
[0022] The interaction requests include, but are not limited to, relational reasoning requests, path analysis requests, and entity attribute query requests. In the following embodiments of this application, the interaction requests include relational reasoning requests, path analysis requests, and entity attribute query requests as examples to illustrate the technical solutions in this application, but this does not constitute a specific limitation. The relational reasoning command is used to trigger the mining of implicit relationships between entities in a knowledge graph network. It requires specifying multiple entity nodes, the known relationship types between multiple entity nodes, and the new relationship type to be generated through reasoning. It analyzes the implicit entity relationships between multiple entity nodes through predefined rules (such as rules for affiliated regulatory agencies). The path analysis command is used to query the relationship path between two entities in the knowledge graph. It requires specifying the source and end entities of the path (such as "system X" and "indicator Y"). Specifically, it can traverse the feasible node paths between the two through breadth-first search (BFS) or depth-first search (DFS) algorithms, and supports outputting the shortest path or full path mode. The entity attribute query command is used to retrieve entities and their associated information in the knowledge graph. After the user enters the search keywords (such as "liquidity coverage ratio"), the semantic similarity is calculated through the edit distance algorithm, the matching entity nodes are located, and their associated relationship nodes (including belonging relationship edges and same-origin relationship edges) are traversed in reverse. For example, when querying an indicator, its affiliated system, issuing agency, and cross-system same-origin indicators are displayed simultaneously.
[0023] In this embodiment of the disclosure, the front-end interactive interface can capture user actions in real time through a standardized interface, transforming user-triggered operations such as relational reasoning, path analysis, or entity attribute queries into structured requests containing parameters such as instruction type and entity identifier. For example, when a user selects two institutional nodes and triggers the "Relationship Reasoning" button in the front-end interactive interface, the interactive function interface encapsulates the first instruction information, such as the source entity node ID and the target entity node ID, into a JSON-formatted interactive request and transmits it to the back-end service. Accordingly, the back-end service receives the corresponding interactive request as a relational reasoning request. If the user enters "capital adequacy ratio" and clicks query, the interface generates an entity attribute query request using the search keywords as parameters. Accordingly, the back-end service receives the corresponding interactive request as an entity attribute query request, and can then parse and execute the corresponding operation based on the knowledge graph network.
[0024] By defining a standardized interactive instruction system, the complex analytical needs of regulatory reporting systems are transformed into machine-executable instruction sets, completely solving the operational pain points of "manually writing complex SQL statements" or "debugging multi-layer recursive algorithms" in traditional database queries. This allows users to complete in-depth correlation analysis without mastering technical details, significantly reducing the technical threshold for regulatory system management.
[0025] Step 140: Execute the interaction request based on the knowledge graph network and output the interaction execution result for the regulatory reporting system.
[0026] In this embodiment of the disclosure, after receiving relationship reasoning, path analysis, or entity attribute query requests from the front-end interactive function interface, the back-end service can utilize the storage and computing capabilities of the knowledge graph network for processing. For relationship reasoning requests, the back-end service can search for implicit entity associations between source entity nodes and target entity entities in the knowledge graph network according to predefined rules; for path analysis requests, the back-end service can calculate feasible paths between entities using graph traversal algorithms; for entity attribute query instructions, the back-end service can locate entities based on semantic similarity matching and traverse their associated nodes in reverse. Finally, the processing results (such as relationship edge data, path sequences, and entity attribute lists) are transformed into visual graphs, structured tables, etc., and returned to the front-end interface for display through the interactive function interface, enabling users to perform in-depth queries and analyses of regulatory reporting system information.
[0027] In summary, the knowledge graph-based regulatory reporting system management method provided by this invention effectively solves the problem of complex queries caused by the scattered storage of information in existing technologies by constructing a knowledge graph network and storing regulatory reporting system information in the form of nodes and relational edges. An interactive interface for the knowledge graph network is provided in the front-end interface. This interface receives user-inputted interaction requests and executes them based on the knowledge graph network, determining the interaction execution result for the regulatory reporting system. Complex regulatory reporting system association analysis can be transformed into standardized command operations. Users do not need to master database joins or graph algorithm knowledge; they can trigger the semantic computing capabilities of the knowledge graph simply through commands such as relational reasoning, path analysis, and entity attribute queries. The graph structure storage of the knowledge graph allows direct traversal of the belonging and affiliation edges between entities during command execution, avoiding the complex operations of traditional multi-table joins in databases. This improves the response speed and accuracy of relational reasoning, path analysis, and entity attribute queries, completely solving the problems of low query efficiency and fragmented results caused by complex operations and lack of semantic reasoning capabilities in existing technologies. Ultimately, this meets the user's need for a comprehensive understanding of the regulatory reporting system.
[0028] Furthermore, as a refinement and extension of the specific implementation methods of the above embodiments, and to fully illustrate the implementation methods of this embodiment, this embodiment also provides another regulatory reporting system management method based on knowledge graphs, such as... Figure 2 As shown, the method includes: Step 210: Obtain the knowledge graph network constructed for the regulatory reporting system.
[0029] For the specific implementation process of the embodiments disclosed herein, please refer to the relevant description in step 110 of the embodiment, which will not be repeated here.
[0030] Step 220: Build a front-end interactive interface and provide interactive function interfaces for knowledge graph networks in the front-end interactive interface.
[0031] For the specific implementation process of the embodiments disclosed herein, please refer to the relevant description in step 120 of the embodiments, which will not be repeated here.
[0032] Step 230: Receive an interaction request input by the user through the interactive function interface. The interaction request includes at least one of the following: relational reasoning request, path analysis request, and entity attribute query request.
[0033] For the specific implementation process of the embodiments disclosed herein, please refer to the relevant description in step 130 of the embodiment, which will not be repeated here.
[0034] Step 240: Execute the interaction request based on the knowledge graph network and output the interaction execution result for the regulatory reporting system.
[0035] In one possible implementation of this disclosure, when the interaction request includes a relationship reasoning request, step 240 may include: determining first instruction information carried by the relationship reasoning request, wherein the first instruction information includes at least predefined reasoning rules, the reasoning rules include a local topology in the knowledge graph network, the local network structure defines multiple entity nodes and known relationship types between the multiple entity nodes, the reasoning rules also include a new relationship type to be reasoned and generated, the new relationship type being a relationship type that does not yet exist among the multiple entity nodes; searching for entity association data matching the local topology according to the knowledge graph network traversal; and performing relationship reasoning on the matching entity association data based on the reasoning rules to generate new relationship type data between the multiple entity nodes.
[0036] The relation reasoning request is an interactive instruction initiated by the user through the front-end interface to deduce new relationships based on known entity relationships in the knowledge graph network. Its core is to mine implicit connections between entities that are not explicitly labeled through predefined rules. The first instruction information is a set of key parameters carried in the relation reasoning request, which is the basic input for triggering and executing relation reasoning. Its core content is the predefined reasoning rules, used to clarify the basis, scope, and goal of the reasoning. The predefined reasoning rules are a pre-set logical framework to guide relation reasoning, containing two core elements: the existing local topology in the knowledge graph (as a premise for reasoning) and the new relation type to be generated (as the goal of reasoning), serving as a logical bridge connecting known and new relations. The local topology is the existing local entity relationship network form in the knowledge graph network, which needs to clearly define the multiple entity nodes (such as regulatory agencies, policy documents, indicators, etc.) and the known relation types between these nodes (such as affiliation, hierarchy, business association, etc.), and can be represented as star, chain, ring, etc. The specific structure includes: Known relation types are the existing and explicitly labeled association categories between entity nodes in the local topology, including but not limited to hierarchical relationships (such as issuing, containing), hierarchical relationships (such as superior-subordinate system), and business association relationships (such as system-applicable business scenario), which serve as the basis for inferring new relations; New relation types to be inferred are association categories that do not yet exist between multiple entity nodes in the local topology and need to be generated through inference. These relation types have a logical deductive connection with known relation types and are the target output of relation inference; Entity association data are specific entity and relation data in the knowledge graph network that match the local topology, i.e., actual data instances that conform to the entity node types, known relation types, and network forms defined by the local topology; New relation type data are structured information generated through relation inference to represent the new relation types to be inferred, including entity node identifiers, relation type tags, and related attributes connected by the new relation, which supplements and expands the entity association relationships in the knowledge graph network.
[0037] The generation of new relation types based on rule-based reasoning can automatically mine implicit relations that originally required manual sorting (such as inferring the affiliation relationship between multiple entities and the same source entity), making the relation network of the knowledge graph more complete, providing richer association information for subsequent query and analysis operations, significantly reducing manual costs, improving the accuracy of relation mining, and reducing omissions caused by human factors.
[0038] In specific application scenarios, after performing relational reasoning on matched entity association data based on predefined reasoning rules to generate new relational data between multiple entity nodes, the knowledge graph network can be updated based on the newly generated relational data. This transforms previously implicit connections into structured information that can be directly queried and traversed within the graph, thereby improving the comprehensive coverage of regulatory reporting system relationships by the knowledge graph. Accordingly, the implementation steps may further include: updating the knowledge graph network based on the newly generated relational data; specifically, updating the knowledge graph network based on the newly generated relational data may include: generating new relational edges between multiple entity nodes based on the new relational data; and using a graph database incremental update algorithm to write the edge data of the new relational edges into the knowledge graph network, updating the entity association relationships between multiple entity nodes. Among them, the graph database incremental update algorithm is an algorithm used in graph databases (such as Neo4j and JanusGraph) to dynamically add or modify data. It supports inserting new nodes, edges or attribute data into the graph without affecting the existing graph structure, ensuring the atomicity and consistency of data updates. Entity association is the connection established between entity nodes in the knowledge graph network through relation edges, including belonging relations (such as descending and containing) and peer relations (same-level association), which are used to represent the semantic relationship between entities.
[0039] In another possible implementation of this disclosure, when the interaction request includes a path analysis request, step 240 may include: determining the second instruction information carried in the path analysis request, wherein the second instruction information includes at least a path source entity node and a path end entity node; based on a knowledge graph network, using a graph traversal algorithm to traverse and find feasible node paths between the path source entity node and the path end entity node, and recording the path steps of the feasible node paths, wherein the feasible node paths include ownership relationship edges and same-ownership relationship edges between entities; and outputting the path query results according to a preset mode.
[0040] The path analysis request is initiated by the user through the front-end interface to query the relationship path between two entity nodes in the knowledge graph. It requires specifying the source and end nodes of the path analysis. The source entity node is the starting entity node of the path analysis, such as a regulatory agency or top-level policy document, serving as the starting point for the relationship path search. The end entity node is the target entity node of the path analysis, such as specific indicators or interface tables, serving as the ending point for the relationship path search. Graph traversal algorithms are algorithms used to search for nodes and edges in the graph structure. Common algorithms include Breadth-First Search (BFS) and Depth-First Search (DFS), which are used in path analysis to traverse the knowledge graph network to find the source... Feasible node paths between the end and end nodes; feasible node paths are all possible connection paths between the source and end entity nodes calculated by a graph traversal algorithm. The paths include ownership relationship edges (such as issuing, containing) and same-affiliation edges (such as belonging to the same regulatory agency or the same business theme); the number of path steps is the number of nodes traversed in each feasible node path, used to measure the length of the path, and used as a filtering criterion in the shortest path mode; the preset mode is a pre-defined rule for outputting path query results, which may include the shortest path mode (filtering the path with the fewest path steps) and the full path mode (outputting all feasible paths and sorting them in ascending order of path steps).
[0041] A structured path analysis process effectively addresses the challenge of quickly locating entity relationship chains in traditional regulatory management. Compared to traditional databases that require complex recursive queries or manual tracing of relationship chains, knowledge graph-based path analysis improves the efficiency of cross-level and cross-regulatory relationship queries, shortening response time. The graph traversal algorithm, combined with preset output modes, supports flexible path display requirements. The shortest path mode quickly locates core related links, while the full path mode comprehensively presents all feasible relationship chains, enabling regulators to fully grasp the scope of regulatory influence.
[0042] Accordingly, when outputting path query results according to a preset mode, the implementation steps may include: when the preset mode is the shortest path mode, filtering the shortest node path with the fewest corresponding path steps from the feasible node paths and outputting it; when the preset mode is the full path mode, traversing all feasible node paths and outputting a complete set of node paths containing path length, node sequence and relationship type, with feasible node paths arranged in order of path steps in the complete set of node paths.
[0043] In another possible implementation of this disclosure, when the interaction request includes an entity attribute query request, step 240 may include: receiving search keywords entered by the user in a preset filter box; calculating the semantic similarity between the search keywords and the description text of each entity node in the knowledge graph network using an edit distance algorithm; determining the search entity node based on the comparison result of the semantic similarity with a preset threshold; traversing the related relationship nodes of the search entity node in reverse based on the knowledge graph network, with the traversal range including belonging relationship edges and same-belonging relationship edges, centered on the search entity node; and outputting the search entity node and related relationship nodes.
[0044] The entity attribute query request is an interactive command initiated by the user through the front-end interface to retrieve specific entities and their associated information in the knowledge graph. Users need to enter search keywords in a preset filter box. The edit distance algorithm is an algorithm for calculating the similarity between two strings. It determines similarity by comparing the number of character replacement, insertion, and deletion operations. For example, the smaller the edit distance between "capital adequacy ratio" and "core capital adequacy ratio," the higher the similarity. Semantic similarity is the degree of semantic similarity between the search keywords and the entity node description text. It is quantified as a value between 0 and 1 using the edit distance algorithm; a higher value indicates greater semantic similarity. The preset threshold is a critical value (e.g., 0.7) used to filter semantic similarity results; only similar results are considered. Only entity nodes with a similarity score exceeding the threshold will be identified as search targets to avoid over-matching. Search entity nodes are knowledge graph entity nodes that match the search keywords, selected based on semantic similarity, such as indicator nodes matched when "capital adequacy ratio" is entered. Reverse traversal is the process of searching along the relational edges (including attribution and related edges) in the knowledge graph from the search entity node as the starting point, for example, searching for the system and regulatory agency to which the indicator node belongs from the reverse traversal. Related nodes are other nodes that have a direct or indirect relationship with the search entity node, found through reverse traversal, including attribution nodes (such as system and regulatory agency) and related nodes (such as systems with the same source indicator or system belonging to the same business theme).
[0045] By using semantic-driven entity retrieval and reverse traversal mechanisms, the query scope can be expanded from a single entity to a complete network of relationships, significantly improving the accuracy and comprehensiveness of querying regulatory reporting system information.
[0046] Step 250: Construct a user behavior analysis model, collect and analyze users' search history, entity viewing records and click behavior on related nodes in the front-end interactive interface, and generate user behavior sequences.
[0047] The user behavior analysis model is an analytical framework built on data mining and machine learning techniques. It is used to collect, process, and understand user actions on the front-end interactive interface to reveal user preferences and needs. The retrieval history consists of the search keywords entered by the user on the front-end interactive interface and their corresponding query records, reflecting the direction of information actively searched by the user (such as "non-performing loan rate calculation method"). The entity viewing record consists of the details of entity nodes that the user has browsed in the knowledge graph (such as policy documents and indicator definition pages), recording the user's in-depth access behavior to specific information. The relationship node click behavior consists of the user's click operations on relationship edges or related nodes in the knowledge graph (such as other policies with the same affiliation relationship, or lower-level indicators with the same affiliation relationship), reflecting the user's need to explore the relationships between entities. The user behavior sequence is a set of user operation records arranged in chronological order, including behavior type (retrieval / viewing / clicking), operation object (entity node / relationship edge), and timestamp, used to describe the complete trajectory of the user's interaction with the system.
[0048] In this embodiment of the disclosure, the user's search history (such as keywords such as "capital adequacy ratio"), entity viewing records (such as the viewed details page of the "Bank Capital Management Measures"), and click behavior on related nodes can be monitored in real time (such as expanding the interface tables and indicators contained in the system after clicking on the system node). These behavioral data are then structured according to the order of operation time to generate a user behavior sequence with timestamps (such as "searching keywords → viewing system nodes → clicking related indicator nodes"). This process collects raw data through data tracking technology or log analysis tools, and then transforms it into an ordered behavioral sequence that can be used for subsequent analysis after cleaning and classification, thereby constructing a dynamic profile of the user's interaction with the knowledge graph of regulatory reporting systems.
[0049] By leveraging user behavior analysis models and behavior sequence generation, the intelligence level and user experience of knowledge graph interactions can be significantly improved. On one hand, analysis of retrieval history and entity viewing records can accurately capture the regulatory areas users are interested in (e.g., frequent searches for credit-related indicators suggest a focus on credit regulation), laying the foundation for subsequent personalized recommendations. Compared to traditional indiscriminate display, this improves the accuracy of information delivery. On the other hand, tracking click behavior on related nodes can uncover users' interests in implicit relationships (e.g., multiple clicks on cross-institutional indicator paths indicate interest in inter-institutional synergy), thereby optimizing the relationship display logic of the knowledge graph and improving the efficiency of presenting complex relationships. Furthermore, the generation of user behavior sequences transforms fragmented operations into structured data, supporting the mining of potential user needs based on sequence patterns (e.g., proactively recommending similar regulations after discovering high-frequency behavior patterns of "retrieval-viewing-favoriting"), providing regulators with more business-oriented intelligent assistance and driving a shift from passive querying to proactive service.
[0050] Step 260: Using collaborative filtering algorithm and knowledge graph embedding technology, user behavior sequences are mapped to semantic paths in the knowledge graph network. Based on the semantic paths, the correlation between user behavior and the regulatory reporting system corresponding to each entity node is calculated, and a recommendation list of regulatory reporting systems for users is generated according to the correlation.
[0051] In this embodiment of the disclosure, a collaborative filtering algorithm can be used to analyze user behavior sequences (such as search history and node click records) to identify user groups with similar behavioral patterns to the target user. Simultaneously, knowledge graph embedding technology is used to map the operation objects (such as search keywords and viewed entity nodes) in the user behavior sequence to actual semantic paths in the knowledge graph network (such as the association link of "regulatory agency → system → indicator"). Next, based on the generated semantic paths, the correlation between the regulatory reporting system corresponding to each entity node and the user behavior is calculated, quantifying the degree of correlation through indicators such as path length and node access frequency. Finally, a personalized list of recommended regulatory reporting systems is generated based on the correlation from high to low, pushing relevant policy documents, indicator explanations, and other content that the user may be interested in.
[0052] By integrating collaborative filtering and knowledge graph embedding technologies, the accuracy and practicality of regulatory reporting system recommendations can be significantly improved. Traditional keyword-based recommendations are susceptible to semantic ambiguity, while knowledge graph embedding technology transforms user behavior into structured semantic paths, making the recommendations more aligned with business logic and thus improving accuracy. Collaborative filtering algorithms leverage the similarity of group behavior to uncover potential user needs. Furthermore, the combination of correlation calculation and path features allows for dynamic adjustment of recommendation priorities, avoiding interference from irrelevant information. Ultimately, the generated personalized recommendation list reduces the time cost for users to obtain information, effectively addressing the efficiency problem of regulators when faced with massive amounts of regulations, and driving the upgrade of knowledge graphs from passive querying to proactive service.
[0053] In summary, the technical solution of this application, by constructing a knowledge graph network and combining it with a front-end interactive request execution mechanism, can fundamentally solve the problems of information dispersion, complex querying, and lack of semantic reasoning capabilities in regulatory reporting system management under traditional database storage methods. Specifically, the knowledge graph network stores entities such as regulatory agencies, systems, and interface tables, along with their affiliation / related relationships, in a structured manner using nodes and relational edges. This transforms multi-table joins in traditional databases into graph path searches, improving the efficiency of cross-system entity association queries. Relationship reasoning requests automatically mine implicit entity associations between target entities based on predefined rules (such as associations with source indicators), dynamically completing relational edges. This solves the time-consuming problem of manually sorting relationships in existing technologies, improving the efficiency of cross-system compliance analysis. Path analysis requests generate full or shortest paths between entities through graph traversal algorithms, transforming complex recursive queries into visual topology graph displays. Entity attribute queries combined with edit distance algorithms achieve semantic-level fuzzy matching, improving query accuracy. The front-end interactive interface encapsulates the complex operations of knowledge graphs into a visual command interface, allowing users to complete hierarchical association analysis through click-based interaction. This improves the efficiency and accuracy of querying regulatory reporting systems, meeting users' needs for a comprehensive understanding of these systems.
[0054] Furthermore, as Figure 1 and Figure 2 The specific implementation of the method shown in this embodiment provides a knowledge graph-based regulatory reporting system management device, such as... Figure 3 As shown, the device includes: an acquisition module 31, a construction module 32, a receiving module 33, and an execution module 34.
[0055] Module 31 can be used to acquire the knowledge graph network constructed for the regulatory reporting system; Module 32 can be used to build a front-end interactive interface, providing interactive function interfaces for knowledge graph networks in the front-end interactive interface; The receiving module 33 can be used to receive interactive requests input by the user through the interactive function interface. The interactive requests include at least one of relation reasoning requests, path analysis requests, and entity attribute query requests. The execution module 34 can be used to execute interactive requests based on the knowledge graph network and output the interactive execution results for the regulatory reporting system.
[0056] In some embodiments of this application, when the interaction request is a relational reasoning request, the execution module 34 can be specifically used to determine the first instruction information carried by the relational reasoning request. The first instruction information includes at least a predefined reasoning rule. The reasoning rule includes a local topology in the knowledge graph network. The local network structure defines multiple entity nodes and known relation types between multiple entity nodes. The reasoning rule also includes a new relation type to be reasoned and generated. The new relation type is a relation type that does not yet exist among the multiple entity nodes. The knowledge graph network is traversed to find entity association data that matches the local topology. Based on the inference rules, relational inference is performed on the matched entity association data to generate new relational data between multiple entity nodes.
[0057] In some embodiments of this application, such as Figure 4 As shown, the device also includes: an update module 35; Update module 35 can be used to update the knowledge graph network based on new relation type data generated by reasoning; Correspondingly, update module 35 can be used to generate new relational edges between multiple entity nodes based on new relational type data; and to write the edge data of the new relational edges into the knowledge graph network using a graph database incremental update algorithm to update the entity association relationship between multiple entity nodes.
[0058] In some embodiments of this application, when the interaction request is a path analysis request, the execution module 34 can be used to determine the second instruction information carried by the path analysis request. The second instruction information includes at least the source entity node and the end entity node of the path. Based on the knowledge graph network, a graph traversal algorithm is used to traverse and find feasible node paths between the source entity node and the end entity node of the path, and the path steps of the feasible node paths are recorded. The feasible node paths include the belonging relationship edges and the same relationship edges between entities. The path query results are output according to a preset mode.
[0059] Correspondingly, when outputting path query results according to the preset mode, the execution module 34 can be used to filter and output the shortest node path with the fewest corresponding path steps among the feasible node paths when the preset mode is the shortest path mode; when the preset mode is the full path mode, it traverses all feasible node paths and outputs a complete set of node paths containing path length, node sequence and relationship type, and the feasible node paths are arranged in order of path steps in the complete set of node paths.
[0060] In some embodiments of this application, when the interaction request is an entity attribute query request, the execution module 34 can be specifically used to receive the search keywords entered by the user in the preset filter box, calculate the semantic similarity between the search keywords and the description text of each entity node in the knowledge graph network using the edit distance algorithm; determine the search entity node based on the comparison result of the semantic similarity with the preset threshold; traverse the related relationship nodes of the search entity node in reverse based on the knowledge graph network with the search entity node as the center, and the traversal range includes the belonging relationship edge and the same belonging relationship edge; and output the search entity node and the related relationship nodes.
[0061] In some embodiments of this application, such as Figure 4 As shown, the device also includes: a generation module 36; The generation module 36 can be used to build a user behavior analysis model, collect and analyze users' search history, entity viewing records and click behavior on related nodes in the front-end interactive interface, and generate user behavior sequences; using collaborative filtering algorithm and knowledge graph embedding technology, the user behavior sequences are mapped to semantic paths in the knowledge graph network, and the correlation between user behavior and the regulatory reporting system corresponding to each entity node is calculated based on the semantic path, and a list of recommended regulatory reporting systems for users is generated according to the correlation.
[0062] It should be noted that other corresponding descriptions of the functional units involved in the knowledge graph-based regulatory reporting system management device provided in this embodiment can be found in [reference needed]. Figure 1 and Figure 2 The corresponding descriptions in [the document] will not be repeated here.
[0063] Based on the above, Figure 1 and Figure 2 Accordingly, this embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements the above-described method. Figure 1 and Figure 2 The example shown is a knowledge graph-based regulatory reporting system management method.
[0064] Based on this understanding, the technical solution of this application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as CD-ROM, USB flash drive, mobile hard drive, etc.) and includes several instructions to cause an electronic device (such as personal computer, server, or network device, etc.) to execute the methods of various implementation scenarios of this application.
[0065] Based on the above, Figure 1 and Figure 2 The method shown, and Figure 3 and Figure 4To achieve the above objectives, the present application also provides an electronic device, specifically a personal computer, tablet computer, server, or other network device, as shown in the virtual device embodiment. This device includes a storage medium and a processor; the storage medium stores a computer program; the processor executes the computer program to achieve the above-described objectives. Figure 1 and Figure 2 The example shown is a knowledge graph-based regulatory reporting system management method.
[0066] Optionally, the aforementioned physical devices may also include a user interface, a network interface, a camera, radio frequency (RF) circuitry, sensors, audio circuitry, a Wi-Fi module, etc. The user interface may include a display screen, input units such as a keyboard, etc., and optional user interfaces may also include USB interfaces, card reader interfaces, etc. The network interface may optionally include standard wired interfaces, wireless interfaces (such as Wi-Fi interfaces), etc.
[0067] Those skilled in the art will understand that the physical device structure provided in this embodiment does not constitute a limitation on the physical device, and may include more or fewer components, or combine certain components, or have different component arrangements.
[0068] The storage medium may also include an operating system and a network communication module. The operating system is a program that manages the hardware and software resources of the aforementioned physical device, supporting the operation of information processing programs and other software and / or programs. The network communication module is used to enable communication between the various components within the storage medium, as well as communication with other hardware and software in the information processing physical device.
[0069] Through the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware platform, or it can be implemented by hardware.
[0070] This invention, by constructing a knowledge graph network and combining it with a front-end interactive request execution mechanism, fundamentally solves the problems of information dispersion, complex queries, and lack of semantic reasoning capabilities in regulatory reporting system management under traditional database storage methods. Specifically, the knowledge graph network stores entities such as regulatory agencies, systems, and interface tables, along with their affiliation / related relationships, in a structured manner using nodes and relational edges. This transforms multi-table joins in traditional databases into graph path searches, improving the efficiency of cross-system entity association queries. Relationship reasoning requests automatically mine implicit entity associations between target entities based on predefined rules (such as associations with source indicators), dynamically completing relational edges. This solves the time-consuming problem of manually sorting relationships in existing technologies, improving the efficiency of cross-system compliance analysis. Path analysis requests generate full or shortest paths between entities through graph traversal algorithms, transforming complex recursive queries into visual topology graph displays. Entity attribute queries, combined with edit distance algorithms, achieve semantic-level fuzzy matching, improving query accuracy. The front-end interactive interface encapsulates the complex operations of knowledge graphs into a visual command interface, allowing users to complete hierarchical association analysis through click-based interaction. This improves the efficiency and accuracy of querying regulatory reporting systems, meeting users' needs for a comprehensive understanding of these systems.
[0071] Those skilled in the art will understand that the accompanying drawings are merely schematic diagrams of a preferred embodiment, and the modules or processes shown in the drawings are not necessarily essential for implementing this application. Those skilled in the art will understand that the modules in the apparatus of the embodiment can be distributed within the apparatus of the embodiment as described, or can be modified to be located in one or more apparatuses different from this embodiment. The modules of the above-described embodiment can be combined into one module, or further divided into multiple sub-modules.
[0072] The serial numbers in this application are for descriptive purposes only and do not represent the superiority or inferiority of any particular implementation scenario. The above disclosures are merely a few specific implementation scenarios of this application; however, this application is not limited thereto, and any variations conceived by those skilled in the art should fall within the protection scope of this application.
Claims
1. A management method for a regulatory reporting system based on knowledge graphs, characterized in that, include: Obtain the knowledge graph network constructed for the regulatory reporting system; Construct a front-end interactive interface, and provide interactive function interfaces for the knowledge graph network in the front-end interactive interface; Receives an interaction request input by the user through the interaction function interface, the interaction request including at least one of a relationship reasoning request, a path analysis request, and an entity attribute query request; The interaction request is executed based on the knowledge graph network, and the interaction execution result for the regulatory reporting system is output.
2. The method according to claim 1, characterized in that, When the interaction request is the relational reasoning request, the execution of the interaction request based on the knowledge graph network, and the output of the interaction execution result for the regulatory reporting system, include: Determine the first instruction information carried by the relation reasoning request. The first instruction information includes at least a predefined reasoning rule. The reasoning rule contains a local topology structure in the knowledge graph network. The local network structure defines multiple entity nodes and known relation types between the multiple entity nodes. The reasoning rule also contains a new relation type to be reasoned and generated. The new relation type is a relation type that does not yet exist in the multiple entity nodes. The knowledge graph network is traversed to find entity association data that matches the local topology; Based on the inference rules, relational inference is performed on the matched entity association data to generate new relational data between the multiple entity nodes.
3. The method according to claim 2, characterized in that, The method further includes: The knowledge graph network is updated based on the new relation type data generated through reasoning. The updating of the knowledge graph network based on the new relation type data generated through reasoning includes: Based on the new relationship type data, new relationship edges are generated between the multiple entity nodes; An incremental update algorithm for graph databases is used to write the edge data of the new relation edges into the knowledge graph network, thereby updating the entity association relationships between the multiple entity nodes.
4. The method according to claim 1, characterized in that, When the interaction request is the path analysis request, the step of executing the interaction request based on the knowledge graph network and outputting the interaction execution result for the regulatory reporting system includes: The second instruction information carried in the path analysis request is determined, and the second instruction information includes at least the path source entity node and the path end entity node; Based on the knowledge graph network, a graph traversal algorithm is used to traverse and find feasible node paths between the source entity node and the end entity node of the path, and the path steps of the feasible node path are recorded. The feasible node path includes the belonging relationship edge and the same relationship edge between entities. Output the path query results according to the preset mode.
5. The method according to claim 4, characterized in that, The output of path query results according to the preset mode includes: When the preset mode is the shortest path mode, the shortest node path with the fewest steps is selected from the feasible node paths and output; When the preset mode is full path mode, all feasible node paths are traversed, and a complete set of node paths containing path length, node sequence and relationship type is output. The feasible node paths are arranged in order of path steps in the complete set of node paths.
6. The method according to claim 1, characterized in that, When the interaction request is an entity attribute query request, the step of executing the interaction request based on the knowledge graph network and outputting the interaction execution result for the regulatory reporting system includes: The system receives search keywords entered by the user in a preset filter box and uses an edit distance algorithm to calculate the semantic similarity between the search keywords and the description text of each entity node in the knowledge graph network. Based on the comparison result between the semantic similarity and the preset threshold, the retrieval entity node is determined; Centered on the retrieved entity node, the association nodes of the retrieved entity node are traversed in reverse based on the knowledge graph network. The traversal range includes belonging relationship edges and same-belonging relationship edges. Output the retrieved entity node and the associated relationship node.
7. The method according to claim 6, characterized in that, The method further includes: Construct a user behavior analysis model, collect and analyze the user's search history, entity viewing records and click behavior on the related nodes on the front-end interactive interface, and generate user behavior sequences; By employing collaborative filtering algorithms and knowledge graph embedding technology, the user behavior sequence is mapped to a semantic path in the knowledge graph network. Based on the semantic path, the correlation between user behavior and the regulatory reporting system corresponding to each entity node is calculated, and a recommendation list of regulatory reporting systems for the user is generated according to the correlation.
8. A knowledge graph-based regulatory reporting system management device, characterized in that, include: The acquisition module is used to acquire the knowledge graph network constructed for the regulatory reporting system; A building module is used to build a front-end interactive interface, which provides interactive function interfaces for the knowledge graph network. The receiving module is used to receive interactive requests input by the user through the interactive function interface, wherein the interactive requests include at least one of relation reasoning requests, path analysis requests, and entity attribute query requests; The execution module is used to execute the interaction request based on the knowledge graph network and output the interaction execution result for the regulatory reporting system.
9. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 7.
10. An electronic device comprising a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 7.