Association identification method and device, equipment, storage medium and program product

Abstract

The embodiment of the invention provides an association identification method and device, equipment, a storage medium and a program product, and relates to the field of big data and financial science and technology. The method comprises the steps of obtaining business data of a target object; based on the business data, a semantic baseline of a preset element corresponding to the target object is constructed, and the semantic baseline comprises a relation type set, a role type set and a constraint rule set; generating a path template based on the semantic baseline; based on the path template, an initial path meeting the semantic baseline is obtained through mining in a preset knowledge graph, a target path of the target object is obtained through closed-loop consistency verification, and the target path is used for representing the incidence relation corresponding to the target object. According to the method, path search is prevented from being separated from compliance semantics, and the error recognition rate is reduced. A logic closed loop of path search and compliance verification is ensured, limitations on semantic expression, path search and evidence chain recheck are solved, and the accuracy and traceability of associated transaction identification are remarkably improved.

Description

Technical Field

[0001] This application relates to the fields of big data and financial technology, and in particular to a method, apparatus, device, storage medium and program product for association identification. Background Technology

[0002] As business operations become increasingly complex, multi-source information intertwines to form intricate, multi-hop relationship networks, making it difficult for traditional methods to penetrate the surface and uncover hidden related-party transactions. Furthermore, regulatory agencies require companies to disclose the compliance of related-party transactions, but existing systems rely on human experience to determine preset factors, making it difficult to quantify whether a company is compliant.

[0003] Existing technologies for transaction monitoring based on thresholds or preset rules have limited ability to identify correlations. General graph algorithm-driven graph database solutions are prone to false positives. Furthermore, lightweight ontology and rule engine-based local matching schemes suffer from incomplete evidence chains, making it difficult to form a compliant evidence chain. Therefore, current methods for regulating corporate affiliates have limitations in regulatory semantic expression, path search constraints, and the ability to structure evidence chains. Summary of the Invention

[0004] This application provides a method, apparatus, device, storage medium, and program product for identifying related parties, in order to solve the technical problems of limitations in existing supervision of corporate related parties in terms of regulatory semantic expression, path search constraints, and evidence chain structuring capabilities.

[0005] Firstly, this application provides an association identification method, including:

[0006] Obtain the business data of the target object;

[0007] Based on business data, construct a semantic baseline for the preset elements corresponding to the target object. The semantic baseline includes a set of relationship types, a set of role types, and a set of constraint rules.

[0008] Generate path templates based on semantic baselines;

[0009] Based on the path template, an initial path that meets the semantic baseline is mined from the preset knowledge graph. Through closed-loop consistency verification, the target path of the target object is obtained. The target path is used to represent the association relationship corresponding to the target object.

[0010] Secondly, this application provides an association identification device, comprising:

[0011] The acquisition module is used to acquire business data of the target object;

[0012] The construction module is used to build a semantic baseline for the preset elements corresponding to the target object based on business data. The semantic baseline includes a set of relationship types, a set of role types, and a set of constraint rules.

[0013] The generation module is used to generate path templates based on the semantic baseline;

[0014] The mining module is used to mine initial paths that meet the semantic baseline in a preset knowledge graph based on path templates. Through closed-loop consistency verification, the target path of the target object is obtained. The target path is used to represent the association relationship corresponding to the target object.

[0015] Thirdly, this application provides an electronic device, including: a processor and a memory communicatively connected to the processor;

[0016] The memory stores the instructions that the computer executes;

[0017] The processor executes computer-executable instructions stored in memory to implement any of the methods of the first aspect.

[0018] Fourthly, this application provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the method of any one of the first aspects.

[0019] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the method of any one of the first aspects.

[0020] The association identification method, apparatus, device, storage medium, and program product provided in this application construct a semantic baseline of preset elements, unifying fragmented business data with preset elements to form executable constraint rules. Subsequently, a path template is generated based on the semantic baseline, limiting the rules for path composition, preventing path search from deviating from compliant semantics, compressing the search space, and reducing the false identification rate. When performing path mining in the knowledge graph, only compliant paths that satisfy the path template are searched, and false associations are eliminated through closed-loop consistency verification. Using preset elements as a benchmark, a logical closed loop between path search and compliance verification is ensured, ultimately outputting the target path, which is the association relationship. This solves the limitations of existing technologies in semantic expression, path search, and evidence chain verification, significantly improving the accuracy and traceability of associated transaction identification. Attached Figure Description

[0021] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0022] Figure 1 A diagram illustrating methods for identifying relationships in a financial regulatory scenario;

[0023] Figure 2 A flowchart illustrating an association identification method provided in an embodiment of this application;

[0024] Figure 3 A flowchart illustrating a method for discovering related party transaction relationships based on knowledge graph reasoning and path mining, provided for an embodiment of this application;

[0025] Figure 4 This is a schematic diagram of the structure of an association identification device provided in an embodiment of this application;

[0026] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0027] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0028] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0029] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, storage, use, processing, transmission, provision, disclosure, and application of the relevant data all comply with the relevant laws, regulations, and standards of the relevant countries and regions, have taken necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation access points for users to choose to authorize or refuse.

[0030] Furthermore, the technical solution involved in this application, which involves big data analysis of user information (including but not limited to personal biometrics, identity data, consumption data, asset data, electronic terminal operation data, etc.) and the use of artificial intelligence technology for automated decision-making, and makes decisions that have a significant impact on personal rights based on the results of automated decision-making, provides users with corresponding operation entry points for users to choose to agree to or reject the results of automated decision-making; if the user chooses to reject, the process will proceed to the expert decision-making process.

[0031] It should be noted that the association identification method, apparatus, equipment, storage medium and program products provided in this application can be used in the fields of big data and fintech, as well as in any other fields. The application fields of the association identification method, apparatus, equipment, storage medium and program products in this application are not limited.

[0032] The specific application scenario of this application is in the financial supervision scenario, where multi-source information is intertwined to form a complex multi-hop relationship network, and it is necessary to identify the relationship between related parties. Figure 1 This is a diagram illustrating methods for identifying relationships in a financial regulatory scenario, such as... Figure 1 As shown, existing methods for identifying relationships in financial regulatory scenarios can be categorized into three types: 1. Threshold- and rule-based regulatory systems: These rely on simple rules such as amount and frequency for early warning, but cannot express the complex roles and relationship constraints in regulatory clauses, resulting in limited ability to identify relationships. 2. Graph database solutions driven by general graph algorithms: These mine topological connectivity through algorithms such as shortest path and random walk, but lack semantic constraints on relationship types and subject roles, easily leading to false positives, such as misjudging paths with no actual connection as having a relationship. 3. Local matching solutions using lightweight ontology and rule engines: These cover local segments through pattern matching, but cannot construct temporal closed loops and lack start and end subject binding and key node verification, resulting in an incomplete evidence chain and difficulty in forming verifiable compliance evidence.

[0033] Therefore, existing methods for identifying relationships suffer from the following common shortcomings: Misalignment between preset element semantics and knowledge graph elements: Preset elements are not transformed into executable constraints, causing path search to deviate from compliant semantics. Lack of type and order restrictions in path search: General graph algorithms only focus on topological connectivity and cannot limit temporal sequence and role consistency. Unstructured evidence: Rule hit descriptions and source credentials are not bound together, making results difficult to verify and providing insufficient support for handling guidelines.

[0034] The related party identification method, device, equipment, storage medium, and program products provided in this application, by constructing a regulatory semantic-driven semantic baseline and path template, unify the modeling of fragmented business data and compliance judgment elements, and realize the automatic identification, compliance review, and generation of structured evidence chains of related party transactions based on type constraint path mining and closed-loop consistency verification, aim to solve the above-mentioned technical problems of the prior art.

[0035] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0036] Figure 2 This is a flowchart illustrating an association identification method provided in an embodiment of this application, as shown below. Figure 2 As shown, the method includes:

[0037] S201. Obtain the business data of the target object.

[0038] In one example, business data may include: preset relationships, cash flow, rule text, etc., which are not limited in this application.

[0039] S202. Based on business data, construct a semantic baseline for the preset elements corresponding to the target object.

[0040] In this embodiment of the application, the semantic baseline includes a set of relation types, a set of role types, and a set of constraint rules;

[0041] In one example, preset elements may include relational elements defined in preset rules, such as control, voting rights, and cash flow. For instance, "control" in the preset element could be defined as a relation type Control, with directionality A→B (A controls B), and setting participation boundaries for role type Role A→Role B. The semantic baseline includes executable constraints that formalize the preset elements into a knowledge graph, including a set of relation types, a set of role types, and rules such as connection constraints and sequence constraints. For example, the set of relation types defined in the semantic baseline might include Control, Shareholding, and Transaction, and connection constraints might limit allowed connections from Control to Shareholding.

[0042] S203. Generate a path template based on the semantic baseline.

[0043] In one example, a path template may include a structured template for guiding knowledge graph path searches, containing allowed relation sequences, key node types, and start and end subject binding information. For example, the path template [Control→Shareholding→Transaction] restricts the evidence chain to include control relationships and binds the start and end subjects to roles A and B.

[0044] S204. Based on the path template, an initial path that meets the semantic baseline is mined from the preset knowledge graph. The target path of the target object is obtained through closed-loop consistency verification.

[0045] In this embodiment, the target path is used to characterize the association relationship corresponding to the target object.

[0046] In one example, the initial path that meets the semantic baseline can be mined from the preset knowledge graph through type-constrained path mining, which includes: performing path search based on the semantic constraints of the path template in the preset knowledge graph, matching only the initial path that meets the requirements of relation sequence, key node type and order, and determining the initial path that meets the requirements of relation sequence, key node type and order as the target path of the target object.

[0047] In one implementation scenario, taking the discovery of related-party transaction relationships as an example, Figure 3 A flowchart illustrating a method for discovering related-party transaction relationships based on knowledge graph reasoning and path mining, provided in this application embodiment, is shown below. Figure 3 As shown, the methods for discovering related-party transaction relationships based on knowledge graph reasoning and path mining include:

[0048] S1. Compliance semantics and graph ontology definition: Based on regulatory guidelines, define relationship types and role types such as control, shareholding, voting rights, cross-appointment, guarantee, transaction, bill, and capital flow in the knowledge graph, form a semantic baseline, and output the set of relationship types and the set of role types;

[0049] S2. Ontology rule reasoning identifies suspected association types and subject pairs; taking the semantic baseline output from S1 as input, it performs rule reasoning and outputs a set of subject pairs with association type labels and a rule hit description.

[0050] S3. Semantic constraint generation from association type to path template: Taking the set of subject pairs and association type labels output by S2 as input, on the semantic baseline of S1, each association type is mapped to the allowed relation sequence and key node type and the order requirements are set. At the same time, the start and end subject pairs are bound to the corresponding templates, and the set of path templates with start and end subject pairs is output. This step transforms the regulatory semantics into executable constraints.

[0051] S4. Type-constrained path mining performs evidence search; taking the path template set output by S3 as input, type-constrained path mining is performed one by one in the knowledge graph, and the output is a set of candidate transaction chains that satisfy the template.

[0052] S5, Transaction Closed-Loop Consistency Verification and Compliance Alignment: Taking the candidate transaction chain set output by S4 as input, verify the integrity and role consistency of the transaction, capital flow, and guarantee chain, eliminate paths that do not constitute closed-loop evidence, and output a set of compliant evidence chains;

[0053] S6. Structured Representation of Evidence Chains and Audit Admissibility Labeling: Taking the set of compliant evidence chains output from S5 as input, generate evidence chain objects pointing to entities, relationships, transactions, and amounts, along with association type tags and rule hit descriptions, and output a set of evidence chain objects.

[0054] S7. Risk Identification and Handling Guidelines Generation: Using the evidence chain object set output from S6 as input, risk identification conclusions and handling suggestions for related party transactions are generated for auditing and risk control use.

[0055] Specifically, in step S1, a semantic baseline is constructed for subsequent reasoning and path mining. Using the subjects and relationships in a pre-defined knowledge graph as the carrier, the judgment elements of regulatory criteria are solidified into executable connection constraints and sequence constraints. The pre-defined knowledge graph uses a graph as its basic data structure, represented as shown in the formula. ,in Representing a knowledge graph, Represents a set of subjects. This represents the set of relation edges. The semantic baseline, serving as a unified constraint carrier, is represented as... ,in Indicates the semantic baseline, Represents a set of character types. Represents a set of relation types. Represents the set of connection constraints. Represents a set of order constraints. The role position requirement mapping indicates the relationship type. This represents the set of factors used to determine regulatory criteria. This represents a set of related type labels. The type binding between the subject and the relationship is expressed using a type mapping function. and ,in A mapping function from subject to role type. This is a mapping function from relation to relation type.

[0056] In setting roles and relationship types, the core participants and key relationships in the transaction evidence chain are selected based on regulatory semantics, forming the basic elements of the semantic baseline. Among these, It must include at least the subject types that meet the role consistency verification; It must include at least the relationship types that satisfy the chain of evidence identification. The role and position requirements of the relationship must be mapped. Description, in which The mapping requirement is used to define the position and direction of each relationship type in the chain of evidence, such as from the payer to the payee in a cash flow, or from one party to another in a contract.

[0057] At the constraint level, connectivity constraints and order constraints are directly imported from regulatory guidelines and explicitly defined on the semantic baseline. (Connectivity constraint set) This represents a pair of relation types that are allowed to be connected adjacently in the chain of evidence, used to define the connection methods such as connecting a transaction after a contract, connecting a flow of funds after a transaction, and connecting a flow of funds with a guarantee; sequence constraint set. This represents the allowed sequence of relation types, used to limit the order in which relationships such as contract → transaction → cash flow → guarantee appear. To ensure the interpretability of subsequent reasoning and verification, this is a set of regulatory criteria for determining factors. The system includes criteria for determining consistency such as the amount of money involved, chronological order, role consistency, and chain integrity, and maintains consistency with the aforementioned constraints; it also includes a set of related type tags. This is used to provide readable association type labels for subject pairs and evidence chain objects in subsequent steps, ensuring consistency in the binding from rule hit descriptions to path templates to transaction evidence paths. The output of step S1 is the semantic baseline. It is referenced as a unified input in subsequent steps to ensure that ontology rule reasoning, semantic constraint generation from associated types to path templates, type constraint path mining, transaction closed-loop consistency verification and compliance alignment, and evidence chain structured expression and audit admissibility annotation are all performed under the same semantic caliber.

[0058] In step S2, based on the semantic baseline output in step S1 With knowledge graph Perform ontology rule reasoning to generate a set of subject pairs with associated type labels and rule hit descriptions. Specifically, establish a rule set. Each of these rules The rule header corresponds to an associated type tag, which is mapped Given that the rule body consists of a sequence of relation types constitute, From the set of relation types; for each relation type in the rule body Mapping based on role position requirements in the semantic baseline Set start and end character type ,in The subjects and relations in the knowledge graph are respectively mapped through type mapping functions. and The type is fixed for rule body instantiation and constraint verification; the output of step S2 includes a set of subject pairs. And a rule hit description that corresponds one-to-one with the subject, used to maintain consistent referencing in subsequent steps.

[0059] To ensure that reasoning triggers have computable and verifiable criteria, a reasoning trigger determination function is defined, which unifies the instantiation of the relation type sequence of the rule body in the knowledge graph with the semantic baseline constraint to a computable value.

[0060] Let a certain rule A sequence of relation types is instantiated in a knowledge graph as ,in Represents the main node, Represent a relation edge, and satisfy... ,definition

[0061]

[0062] in This indicates the degree to which the instantiation satisfies the semantic baseline. This is an indicator function that takes the value 1 when the condition is true and 0 otherwise. This is a set of connection constraints used to limit allowed connection pairs of the adjacency relationship type; This is a set of order constraints used to limit the allowed order of sequences of relation types; The role position requirement for the relationship type is used to verify the direction and position of each relationship edge on the main role type; and They are type-bound for the subject and the relation, respectively. Only when At that time, the instantiation triggers the rule header and generates the subject pair.

[0063] In practice, the rule set is read one by one. For each rule In knowledge graphs The facts are enumerated to satisfy Relational type serial instantiation and calculate .like Then the starting and ending subjects will be... Recorded as the main body And based on the mapping Assign a corresponding association type label to the subject pair; simultaneously generate a rule hit description for the subject pair. The rule hit description is an ordered record, which includes at least a rule identifier, a rule text summary, a list of instantiated subject nodes and relation edges, and is arranged according to the relation type sequence. The order of recording the role type of each subject Type identifier for each relation To ensure the connection constraints with the semantic baseline With sequence constraints Maintain consistent citations.

[0064] Furthermore, step S2 will combine each subject pair Related type tags The rules and their matching descriptions establish a one-to-one correspondence, and are grouped according to the associated type tags to form subject-to-set pairs. The output includes the rule hit description. The output is directly referenced as input to step S3 to map the association type to allowed relation sequences and order requirements on the semantic baseline and bind start and end subject pairs, ensuring that the process from rule reasoning to path templates, then to type constraint path mining and transaction closure consistency verification is always performed under the same semantic caliber.

[0065] In step S3, based on the semantic baseline output in step S1 The set of principal pairs output by step S2 And the corresponding associated type tags, generating semantic constraints from associated types to path templates and completing the start and end subject binding, thereby outputting a set of path templates with start and end subject pairs for subsequent type-constrained path mining to be executed directly. Therefore, the path template set is defined as follows: any of the path templates Represented as a quadruple ,in Indicates a related type tag, Describes the allowed sequence of relations and Indicates the relation type, Represents a set of key node types. Indicates the start and end subject binding information and Represents the subject; where the mapping from subject to role type is as follows: The mapping from relation to relation type is The mapping from relationship type to role position requirement is as follows: To implement regulatory semantics at the template construction layer, a mapping between tags and sequence requirements is defined. This is used to select the allowed sequence of relation types based on the association type label; and to define the mapping from label to key node type. Used to determine the set of elements according to regulatory standards. Select the subject role type that must appear in the chain of evidence. Further, define the function to retrieve the subject's associated type label. This is used to read the associated type label for each subject pair from the output of step S2.

[0066] To ensure the path template is executable on the semantic baseline, a template consistency judgment function is constructed to centrally verify sequence requirements, connection constraints, role positions, and key node coverage. The start and end subject binding information is also included in the verification, ensuring seamless integration between template generation and subsequent matching. Let... and The length is For each have and At the same time, set and This is the association type label for this subject pair.

[0067] definition:

[0068]

[0069] in The value represents the consistency of the template on the semantic baseline. A value of 1 indicates that the allowed relation sequence, order requirements, connection constraints, key node types, and start and end subject binding information of the path template are consistent under the same semantic caliber. For indicator functions; Represents the set of allowed sequences as determined by the association type label; Connection constraints that indicate adjacency relationships; Give the first The starting and ending position requirements of each relationship type in the chain of evidence; The main character types are the starting point and the ending point, respectively. For a set of key node types, it is required that the role types at the sequence positions be covered within the union of those types.

[0070] In the specific implementation, firstly, for each subject pair output in step S2... Read its associated type label , and by Extract the set of candidate allowed relation sequences that match the label, where each candidate sequence comes from the set of order constraints in the semantic baseline. Secondly, according to Determine the set of key node types and map them to role position requirements based on relationship types. With the set of connection constraints Construct a path template quadruple for each candidate sequence. ,in , For this candidate sequence, , Next, calculate the path template for each completed path. Only when When it is included in the path template set And will be by , and The summarized connection constraints and order requirements are stored as semantic constraints of the path template to ensure that subsequent path mining is performed only within the allowed connection range and order on the semantic baseline. Finally, The paths are merged according to the associated type labels, and a set of path templates with start and end subject pairs is output. This set is used to directly perform type constraint path mining in step S4 and generate a set of candidate transaction chains that meet the templates.

[0071] In step S4, based on the path template set output in step S3 With the knowledge graph defined in step S1 and semantic baseline Type constraint path mining is performed to complete the evidence search. Type constraint path mining revolves around each path template. Expand, among which For association type tags, For allowed relation sequences and , A set of key node types, For the start and end of the subject binding information and The set of connection constraints based on the semantic baseline. Sequence constraint set Mapping from relationship type to role position requirement In the knowledge graph, starting from the main body To the end of the main body Enumeration and Aligned path instances are generated, and role positions, adjacent connections, and key node coverage are verified on each instance to generate candidate transaction chains that meet the template. To ensure consistency in subsequent steps, the output of type-constrained path mining uses a set of candidate transaction chains as a structured carrier, denoted as . and maintain the path template in the collection elements. References and their associated type tags .

[0072] Choose any path template Aligned path instances are represented as ,in Represents the main node, Represents a relation edge, and satisfies relation type binding. Binding to the main type .

[0073] Define the satisfaction function of path instances with respect to templates and semantic baselines as follows:

[0074]

[0075] in This indicates the degree of satisfaction of the path instance with the semantic baseline and the path template. A value of 1 indicates that the path instance satisfies the requirements of start and end subject binding, relation type sequence, role position requirements, adjacent connection constraints, order requirements, and key node type coverage. This is an indicator function that takes the value 1 when the condition is true and 0 otherwise. This represents the set of main role types on the path instance, used to verify the type coverage of key nodes; In For association type tags, For allowed relation sequences, A set of key node types, Binding information to the starting and ending entities; For the first The role position requirements for each relationship type are verified based on the subject's role type. Correspondence verification; and These are used to define the connection method of adjacent relationship types and the order of occurrence of the overall relationship type sequence, respectively. Based on this determination function, the candidate transaction chain set is defined as follows: The collection elements retain the binding relationship between path instances and corresponding path templates in the form of tuples, and are accompanied by association type tags. To serve subsequent verification and expression.

[0076] During execution, the system reads the path template set one by one. Path templates in Based on its start and end subject binding information As a search boundary, in knowledge graphs Upper edge allowed relation sequence Enumerated path instances During the enumeration process, the requirements are first processed in order. First, determine the position of the fixed relation type, then the join constraint. Constrain the allowed connections for adjacent relationship pairs, and then map them based on role and position requirements. Verify the start and end subject role types of each relation edge, and use... Set of main character types along the path Perform key node type coverage verification. Calculate the following for each enumerated path instance: The path instance is associated only when its value is 1. The bindings are included in the candidate transaction chain set. And record the associated type label in the collection elements. The output includes a list of main nodes and relational edges for each path instance, along with their corresponding positions in the semantic baseline constraints. The final output is a set of candidate transaction chains. As input to step S5, it is directly referenced for transaction loop consistency verification and compliance alignment, ensuring consistent references and one-to-one correspondences at the terminology and character level from path template to evidence search to verification stage.

[0077] In step S5, the candidate transaction chain set output in step S4 is used. As input, based on semantic baseline For each candidate transaction chain, perform transaction loop consistency verification and compliance alignment, eliminate paths that do not constitute closed-loop evidence, and output a set of compliant evidence chains. Elements in the candidate transaction chain set are represented in binary form as follows: ,in For path instances, Represents the main node, Represents a relation edge, and satisfies relation type binding. Binding to the main type ; For path templates, where Indicates a related type tag, For allowed relation sequences and , A set of key node types, Information is bound to the starting and ending entities.

[0078] Define relation type identifier element Represent the types of contracts, transactions, bills, cash flows, and guarantees; and define numerical functions for amounts. Represents the monetary value of the relation edge, identifying the function. Indicates the currency of the relation edge, timestamp function This represents the time record of the relation edge, where This represents the set of currencies. Further define the set of type location indices for path instances: , , , , To establish a verification window linking amounts between transactions and fund flows, for each transaction index... Define its subsequent transaction limits Define the set of fund flow indexes associated with the transaction. and the set of invoice indexes associated with the transaction. .

[0079] To address the core criteria for aligning closed-loop consistency and compliance, a unified verification function is defined. This function performs integrated calculations on connection constraints, sequence constraints, role position requirements, chain integrity, consistency of monetary amounts, and temporal order to obtain an executable judgment value.

[0080]

[0081] in This indicates whether the candidate transaction chain constitutes closed-loop evidence and completes compliance alignment under the semantic baseline and path template requirements. A value of 1 indicates that all verifications have been passed. This is an indicator function that takes the value 1 when the condition is true and 0 otherwise. The mapping from associated type labels to the set of required relation types is determined based on regulatory criteria. Specify the required relationship types for different association type labels (at least including contracts, transactions, and cash flows; if required by regulatory requirements, include bills and guarantees). For the set of connection constraints, restrict the allowed connection pairs of the adjacency relationship type; For a set of order constraints, it limits the order in which relation types appear; This is a mapping from relationship type to role position requirement, used to verify the start and end subject role types of each relationship edge; For monetary value functions, For currency identification functions, This is a timestamp function.

[0082] By reading the candidate transaction chain set one by one elements in According to the path template Allowed relation sequences in Key node type set Information bound to the start and end entities and at the semantic baseline Calculation under constraints .like Then the path instance is identified. This constitutes closed-loop evidence and has been aligned for compliance; it should be compared with the corresponding path template. Binding relationships and association type tags Keep them together; if If the path fails to pass verification, it is removed. Finally, all verified path instances are merged according to their association type label, and a set of compliant evidence chains is output, denoted as [example chain]. , It is directly referenced in subsequent steps for the structured expression of the chain of evidence and the annotation of audit admissibility, ensuring that the semantics and one-to-one correspondence are consistent at the terminology and character level from path templates, evidence search to closed-loop consistency and compliance alignment.

[0083] In step S6, using the set of compliant evidence chains output in step S5 as input, each path instance that has passed the transaction loop consistency verification and compliance alignment is given a structured expression of the evidence chain and an audit admissibility label, thereby generating evidence chain objects and outputting a set of evidence chain objects. The set of compliant evidence chains is represented as follows: ,in Represents a path instance. Represents the main node, Representing relation edges, the knowledge graph is... ; Indicates a path template. Indicates a related type tag, Describes the allowed sequence of relations and Indicates the relation type, Represents a set of key node types. Indicates the start and end subject binding information; semantic baseline is represented as ,in A collection of character types For a set of relation types, To connect the set of constraints, For a set of order constraints, This is a mapping from relationship type to role position requirements. This is a set of factors for determining regulatory criteria. For a collection of related type tags; the main type binding function is... The relationship type binding function is Define the relation type identifier according to the notation in step S5. These represent the types of transactions, fund flows, and guarantee relationships, respectively; numerical function for monetary amounts. Represents the monetary value of the relation edge, currency function Indicates the currency of the relation edge, timestamp function This represents the time record of the relation edge, where This is a set of currencies. To reuse windows and indices from S5, a set of indices is defined. , , , , Define transaction boundaries Fund Flow Index Set index set In addition, define the credential source function. Map the relation edges to the set of source credential identifiers, where Define the source credential identifier set; define the rule hit description and retrieval function. ,in For the subject to the set, The set of text describing the rule hit, and Obtain the rule hit description corresponding to the start and end subject binding information from step S2.

[0084] To further bind the four types of elements—"subject object, relation object, transaction object, and amount-targeting object"—to a single, referable evidence chain object with "association type label, rule hit description, and audit admissibility annotation," this step defines an evidence chain object generation function. This function outputs an object value for each compliance path instance, thus forming a collection of evidence chain objects. The core construction formula is:

[0085]

[0086] in Represents the chain of evidence object. Represents a collection of main objects. Represents a collection of relational objects. Represents a set of transaction objects. This indicates that the amount points to a collection of objects. For association type tags, For rule hit explanation, For audit admissibility marking. The chain of evidence object set is defined as follows: .

[0087] main object collection Derived from the main node on the path, take The main object , As the main node identifier, As a character type, This is an indicator function; a value of 1 indicates that the main character type belongs to the set of key node types. Otherwise, it is 0. (Set of relational objects) Among them, the relational objects , For relation types, For direction, time, amount, and currency, respectively, are determined by... , , Given, and defined and If and only if Otherwise, take an empty value. ,in This indicates that the attribute does not exist. (Collection of transaction objects) Among them, the trading partners , For the transaction relationship side, If the set is not empty, otherwise This represents the index of the nearest contract edge corresponding to the transaction. This represents the set of ticket edges associated with a transaction. This represents the set of fund flow edges associated with a transaction. This represents the set of collateral edges within the transaction window. The amount points to the set of objects. The amount points to the object. Record the payer, payee, amount, and currency separately.

[0088] Audit Admissibility Marking The elements used to solidify auditable review are defined as follows:

[0089]

[0090] Among them, source certificate citation ,and

[0091] ;

[0092] Regulatory Comparison Vector: The results were obtained through item-by-item calculation, among which

[0093] Indicates the comparison terms of connection constraints. Indicates the order constraint comparison item. This indicates the comparison items required for character position. Indicates the chain integrity comparison item,

[0094] This indicates that the amount points to a consistency comparison item.

[0095] This indicates the time sequence and the comparison items between the guarantee timeline.

[0096] This indicates the alignment and comparison items between the bill and the transaction; the result of the order and role verification is... Consistency of Amount Record ,in This indicates the difference between the transaction amount and the amount recorded in the fund flow window; path template reference. Used to trace back the allowed relationship sequence and the binding information of the start and end subjects. (The above) This is a mapping from associated type labels to a set of required relationship types, derived from the set of regulatory criteria for judgment. Definition; This is an indicator function; its value is 1 if the condition is true, and 0 otherwise.

[0097] Based on the above construction, the system... Each element in calculate The obtained evidence chain objects are then tagged according to their association type. Perform merging and output a set of evidence chain objects. The collection maintains structured references to subject objects, relational objects, transaction objects, and amount-pointing objects at the object level, and establishes a one-to-one correspondence with associated type tags, rule hit descriptions, and audit admissibility annotations, thereby ensuring that subsequent audits and regulatory reviews can be consistently traced back along the "path template - path instance - object - annotation" link.

[0098] In step S7, using the set of evidence chain objects output in step S6 as input, risk identification and handling guidelines are generated based on the associated type tags, rule hit descriptions, and audit admissibility annotations already bound to the evidence chain objects. Preliminary risk assessment and classification: When the chain integrity comparison item or sequence constraint comparison item is not met, it is identified as a missing evidence chain or temporal anomaly; when the role position requirement comparison item is not met, it is identified as a conflict of principal roles; when there are discrepancies in the amount consistency records or inconsistencies in currency, it is identified as an abnormal amount pointing; when the bill and transaction alignment comparison item is not met, it is identified as a bill anomaly; when the guarantee temporal comparison item is not met, it is identified as an abnormal guarantee liability implementation. Based on this, combined with the associated type tags and rule hit descriptions of the evidence chain objects, the risk types are further refined and categorized, maintaining a one-to-one correspondence with the path template references to ensure the retrospective nature of risk identification.

[0099] For cases involving missing evidence chains or chronological anomalies, the system provides audit recommendations for supplementing source documents and reconstructing relationship sequences. For cases involving conflicting subject roles, the system provides guidance on subject identification correction and role change review. For cases involving abnormal monetary amounts, the system provides guidance on suspension, verification, and internal reporting. For cases involving anomalies, the system provides guidance on authenticity verification, issuance, and endorsement process review. All of the above handling guidelines use source document references within the evidence chain as the location basis and path template references as the review path, ensuring that auditing and risk control can quickly locate specific relationship edges and subject objects at the execution level.

[0100] The association identification method provided in this embodiment constructs a semantic baseline of preset elements, unifies fragmented business data with preset elements into a unified model, and forms executable constraint rules. Subsequently, a path template is generated based on the semantic baseline, limiting the rules for path composition, preventing path search from deviating from compliant semantics, compressing the search space, and reducing the false identification rate. When performing path mining in the knowledge graph, only compliant paths that satisfy the path template are searched, and false associations are eliminated through closed-loop consistency verification. Using preset elements as a benchmark, a logical closed loop between path search and compliance verification is ensured, ultimately outputting the target path, which is the association relationship. This method overcomes the limitations of existing technologies in semantic expression, path search, and evidence chain verification, significantly improving the accuracy and traceability of associated transaction identification.

[0101] Optionally, the constraint rule set includes a connection constraint set and a sequence constraint set. Based on business data, a semantic baseline is constructed for the preset elements corresponding to the target object, including: obtaining the relationship type set and role type set corresponding to the target object; determining the connection constraint set and sequence constraint set based on the preset elements corresponding to the target object; and integrating the relationship type set, role type set, connection constraint rule set, sequence constraint set, and the association type label corresponding to the target object to generate a semantic baseline.

[0102] In one example, the set of connection constraints may include: settings for limiting allowed connection pairs between adjacent relation types. The set of order constraints may include: settings for limiting the order in which relation types appear. The set of association type labels may include: settings for identifying the type of association.

[0103] For example, in the step of constructing a semantic baseline, a set of relationship types and a set of role types are defined, and a set of connection constraints and a set of order constraints are set based on the preset element definitions. The semantic baseline is then generated, integrating relationship types, role types, connection constraints, order constraints, and association type tags to provide a unified semantic constraint framework for subsequent path template generation and path mining.

[0104] By defining sets of relationship types and role types, and combining connection constraints and sequence constraints, refined modeling of regulatory elements was achieved. Connection constraints prevent the generation of non-compliant paths. The set of sequence constraints ensures the temporal closure of the evidence chain, avoiding misjudgments caused by fund circulation or interference from accounts with the same name. The generation of semantic baselines provides a unified constraint benchmark for path templates and path mining, significantly improving the accuracy and compliance of association identification.

[0105] Optionally, after generating the path template based on the semantic baseline, the method further includes: constructing a rule set based on the set of relation types and the set of role types; performing rule reasoning in a preset knowledge graph based on the rule set to determine whether the rule set satisfies the connection constraints and order constraints of the semantic baseline; and generating subject pairs and rule hit descriptions corresponding to the rule set when the rule set satisfies the connection constraints and order constraints of the semantic baseline. The rule hit descriptions include rule identifiers, rule text summaries, and a list of instantiated entities and relations.

[0106] In one example, a rule hit description may include: a record of the facts that triggered the rule and a summary of the rule text. A rule set is constructed based on the set of relation types and the set of role types in the semantic baseline, and rule reasoning is performed in a predefined knowledge graph. If the rule body satisfies the connection and order constraints of the semantic baseline, a rule header is generated, and a subject pair is instantiated. Simultaneously, a rule hit description is generated for each subject pair, recording the entity and relation instance that triggered the rule.

[0107] By generating a set of subject pairs labeled with association types through rule-based reasoning, the process of identifying suspected associations becomes transparent. It provides traceable evidence of associations, significantly reduces false positives, and enhances the interpretability of the results through structured expressions explaining rule hits.

[0108] Optionally, the target path of the target object can be obtained through closed-loop consistency verification, including: performing structured processing on the initial path that has passed the closed-loop consistency verification to obtain the structured initial path; binding the structured initial path with the associated type tag, rule hit statement and path template reference to obtain the target path of the target object.

[0109] In one example, the target object may include a structured data object consisting of a subject object, a relation object, a transaction object, and an amount pointer object. The path instances that pass the closed-loop consistency check are then processed in a structured manner. Simultaneously, association type tags, rule hit descriptions, and path template references are bound to obtain the target path of the target object.

[0110] By using structured representations, the chain of compliance evidence is transformed into a verifiable structured object. Staff can verify the compliance of the chain of evidence layer by layer. This significantly improves the interpretability and traceability of the results, avoiding compliance blind spots caused by fragmented evidence chains.

[0111] Optionally, construct a semantic baseline for the preset elements corresponding to the target object, including: parsing the preset file and extracting the preset elements and constraint rule set from the preset file; and / or updating the relationship type set, role type set, and constraint rule set in the semantic baseline, and retaining the historical preset file.

[0112] In one example, natural language processing techniques may include: parsing a preset file using named entity recognition and relation extraction techniques; using natural language processing techniques to parse the preset file and extract preset elements and constraint rule sets; dynamically updating the relation type set, role type set, and constraint rule set in the semantic baseline, and retaining historical versions of the preset file to support backtracking.

[0113] By using natural language processing technology to parse preset files in real time, the semantic baseline dynamically adapts to changes in the preset files. This significantly improves the system's timeliness and compliance, avoiding compliance errors caused by outdated rules.

[0114] Optionally, before constructing the semantic baseline of the preset elements corresponding to the target object, the method further includes: parsing the preset file through natural language processing and optical character recognition to extract unstructured data from the preset file; and mapping the unstructured data to the corresponding relationship type of the preset knowledge graph.

[0115] In one example, text information can be extracted from a preset text using optical character recognition (OCR) technology. The preset text is parsed using natural language processing, and unstructured data such as amounts and timestamps are extracted using OCR. This unstructured data is then mapped to a predefined knowledge graph's corresponding relationship type, and attributes such as amount, currency, and timestamp are added.

[0116] By integrating multimodal data and enhancing semantics, the construction of the evidence chain becomes more comprehensive. This significantly reduces compliance blind spots caused by data fragmentation and improves the completeness and accuracy of association identification.

[0117] Optionally, the method further includes: matching the initial path based on subject pairs using local path templates; and / or, performing path search in the newly added or modified parts of the preset knowledge graph to mine the initial path.

[0118] In one example, local path template matching may include performing path template matching within the scope of the start-end subject pairs of the association. When performing type-constrained path mining, path search is triggered only for newly added or modified portions of the preset knowledge graph, and local path template matching is performed within the scope of the associated start-end subject pairs. Simultaneously, validated and compliant initial paths are cached to avoid redundant calculations.

[0119] Incremental path mining significantly improves processing efficiency. In monitoring business data flows with frequent changes, the system can quickly respond to new transactions and generate target paths without recalculating the entire knowledge graph. This significantly optimizes resource consumption, enabling efficient and real-time identification of related transactions.

[0120] Optionally, the method also includes: determining the causal dependency between subject pairs based on a preset causal graph model; filtering out initial paths with inconsistent causality based on the causal dependency, and marking the initial paths with inconsistent causality as abnormal paths.

[0121] In one example, closed-loop verification enhanced by causal reasoning may include: analyzing causal dependencies using a causal graph model; identifying causal inconsistencies by analyzing causal dependencies using a causal graph model; and displaying the path structure of the initial path, rule hit descriptions, and admissibility annotations using an interactive interface (such as a dynamic path graph or rule hit tree) to support layer-by-layer verification by staff.

[0122] By performing reasoning based on causal graphs, abnormal paths with inconsistent causal relationships are marked, significantly improving efficiency and the credibility of results, reducing subjective bias in manual review, and directly supporting compliance decisions and risk management.

[0123] Figure 4 This is a schematic diagram of the structure of an association identification device provided in an embodiment of this application, as shown below. Figure 4 As shown, the association identification device 40 provided in this embodiment includes:

[0124] Module 401 is used to acquire business data of the target object;

[0125] The construction module 402 is used to construct a semantic baseline of preset elements corresponding to the target object based on business data. The semantic baseline includes a set of relationship types, a set of role types, and a set of constraint rules.

[0126] Generation module 403 is used to generate path templates based on semantic baselines;

[0127] The mining module 404 is used to mine the initial path that meets the semantic baseline in the preset knowledge graph based on the path template, and obtain the target path of the target object through closed-loop consistency verification. The target path is used to represent the association relationship corresponding to the target object.

[0128] In one possible implementation, the constraint rule set includes a connection constraint set and a sequence constraint set. The construction module 402 is specifically used to: obtain the relationship type set and role type set corresponding to the target object; determine the connection constraint set and sequence constraint set based on the preset elements corresponding to the target object; and integrate the relationship type set, role type set, connection constraint rule set, sequence constraint set, and association type label corresponding to the target object to generate a semantic baseline.

[0129] In one possible implementation, the association identification device is further specifically used to: construct a rule set based on a set of relationship types and a set of role types; perform rule reasoning in a preset knowledge graph based on the rule set to determine whether the rule set satisfies the connection constraints and order constraints of the semantic baseline; and when the rule set satisfies the connection constraints and order constraints of the semantic baseline, generate subject pairs and rule hit descriptions corresponding to the rule set, wherein the rule hit descriptions include rule identifiers, rule text summaries, and instantiated entity and relationship lists.

[0130] In one possible implementation, the mining module 404 is specifically used to: perform structured processing on the initial path that has passed the closed-loop consistency check to obtain the structured initial path; bind the structured initial path with the associated type label, rule hit statement and path template reference to obtain the target path of the target object.

[0131] In one possible implementation, the construction module 402 is further specifically used to: parse the preset file, extract the preset elements and constraint rule set in the preset file; and / or, update the relationship type set, role type set and constraint rule set in the semantic baseline, and retain the historical preset file.

[0132] In one possible implementation, the association recognition device is further specifically used to: parse a preset file through natural language processing and optical character recognition, extract unstructured data from the preset file, and map the unstructured data to the corresponding relationship type of a preset knowledge graph.

[0133] In one possible implementation, the association identification device is further specifically used to: match the initial path based on subject pairs using local path templates; and / or to perform path search to mine the initial path in the newly added or modified parts of the preset knowledge graph.

[0134] In one possible implementation, the association identification device is further specifically used to: determine the causal dependency relationship between subject pairs based on a preset causal graph model; based on the causal dependency relationship, filter out initial paths with inconsistent causality, and mark the initial paths with inconsistent causality as abnormal paths.

[0135] The association identification device provided in this embodiment can execute the method provided in the above method embodiment. Its implementation principle and technical effect are similar, and will not be described in detail here.

[0136] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 5 As shown, the electronic device 50 may include a memory 501 and a processor 502. Optionally, the electronic device may also include a transceiver 503, wherein the memory 501 and the processor 502 communicate with each other; for example, the memory 501, the processor 502 and the transceiver 503 may communicate via a communication bus 504, the memory 501 is used to store a computer program, and the processor 502 executes the computer program to implement the method of the above embodiments.

[0137] Optionally, the aforementioned processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps in the method embodiments disclosed in this application can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0138] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the methods in any of the above method embodiments.

[0139] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the methods in any of the above method embodiments.

[0140] All or part of the steps in the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a readable memory. When the program is executed, it performs the steps of the above method embodiments; and the aforementioned memory (storage medium) includes: read-only memory (ROM), RAM, flash memory, hard disk, solid-state drive, magnetic tape, floppy disk, optical disk, and any combination thereof.

[0141] This application describes embodiments with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processing unit of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processing unit of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0142] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0143] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0144] Obviously, those skilled in the art can make various modifications and variations to the embodiments of this application without departing from the spirit and scope of this application. Therefore, if these modifications and variations to the embodiments of this application fall within the scope of the claims of this application and their equivalents, this application also intends to include these modifications and variations.

[0145] In this application, the term "comprising" and its variations can refer to non-limiting inclusion; the term "or" and its variations can refer to "and / or". The terms "first", "second", etc., in this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. In this application, "multiple" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. The character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0146] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily essential to this application.

[0147] It should be further noted that although the steps in the flowchart are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowchart may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0148] It should be understood that the above-described device embodiments are merely illustrative, and the device of this application can also be implemented in other ways. For example, the division of units / modules in the above embodiments is only a logical functional division, and there may be other division methods in actual implementation. For example, multiple units, modules, or components may be combined, or integrated into another system, or some features may be ignored or not executed.

[0149] Furthermore, unless otherwise specified, the functional units / modules in the various embodiments of this application can be integrated into one unit / module, or each unit / module can exist physically separately, or two or more units / modules can be integrated together. The integrated units / modules described above can be implemented in hardware or as software program modules.

[0150] When integrated units / modules are implemented in hardware, the hardware can be digital circuits, analog circuits, etc. The physical implementation of the hardware structure includes, but is not limited to, transistors, memristors, etc. Unless otherwise specified, the processor can be any suitable hardware processor, such as a CPU, GPU, FPGA, DSP, and ASIC, etc. Unless otherwise specified, the storage unit can be any suitable magnetic or magneto-optical storage medium, such as Resistive Random Access Memory (RRAM), Dynamic Random Access Memory (DRAM), Static Random Access Memory (SRAM), Enhanced Dynamic Random Access Memory (EDRAM), High-Bandwidth Memory (HBM), Hybrid Memory Cube (HMC), etc.

[0151] If the integrated unit / module is implemented as a software program module and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned memory includes various media capable of storing program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.

[0152] In the above embodiments, the descriptions of each embodiment have their own emphasis. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments. The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification.

[0153] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the following claims.

[0154] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A method of association recognition, characterized by, The method comprises: acquiring business data of a target object; based on the business data, constructing a semantic baseline of a preset element corresponding to the target object, the semantic baseline comprising a relationship type set, a role type set and a constraint rule set; based on the semantic baseline, generating a path template; based on the path template, mining an initial path in a preset knowledge graph that satisfies the semantic baseline, and through closed-loop consistency checking, acquiring a target path of the target object, the target path being used to represent an associated relationship corresponding to the target object.

2. The method of claim 1, wherein, The constraint rule set comprises a connection constraint set and a sequence constraint set, and the constructing of the semantic baseline of the preset element corresponding to the target object based on the business data comprises: acquiring a relationship type set and a role type set corresponding to the target object; determining the connection constraint set and the sequence constraint set based on the preset element corresponding to the target object; integrating the relationship type set, the role type set, the connection constraint rule set, the sequence constraint set and an associated type label corresponding to the target object to generate the semantic baseline.

3. The method of claim 2, wherein, After the generating of the path template based on the semantic baseline, the method further comprises: based on the relationship type set and the role type set, constructing a rule set; based on the rule set, performing rule reasoning in the preset knowledge graph to determine whether the rule set satisfies the connection constraint and the sequence constraint of the semantic baseline; when the rule set satisfies the connection constraint and the sequence constraint of the semantic baseline, generating a subject pair corresponding to the rule set and a rule hit description, the rule hit description comprising a rule identifier, a rule text summary and a list of instantiated entities and relationships.

4. The method of claim 3, wherein, The acquiring of the target path of the target object through closed-loop consistency checking comprises: performing structured processing on the initial path that passes through the closed-loop consistency checking to obtain a structured processed initial path; binding the structured processed initial path with the associated type label, the rule hit description and a path template reference to obtain the target path of the target object.

5. The method of claim 1, wherein, The constructing of the semantic baseline of the preset element corresponding to the target object comprises: analyzing a preset file to extract a preset element and a constraint rule set in the preset file; and / or, updating the relationship type set, the role type set and the constraint rule set in the semantic baseline, and retaining a historical preset file.

6. The method of claim 5, wherein, Before the constructing of the semantic baseline of the preset element corresponding to the target object, the method further comprises: analyzing the preset file through natural language processing and optical character recognition to extract unstructured data in the preset file; mapping the unstructured data to a corresponding relationship type of the preset knowledge graph.

7. The method of claim 3, wherein, The method further comprises: based on the subject pair, matching the initial path through a local path template; and / or, performing path search on the initial path by mining in a newly added or modified part of the preset knowledge graph.

8. The method of claim 4, wherein, The method further comprises: based on a preset causal graph model, determining a causal dependency relationship between the subject pair; Based on the causal dependency, an initial path with inconsistent causality is screened out, and the initial path with inconsistent causality is marked as an abnormal path.

9. An association identifying apparatus characterized by comprising: The device comprises: An acquisition module is configured to acquire service data of a target object. A construction module is configured to construct a semantic baseline of a preset element corresponding to the target object based on the service data, the semantic baseline comprising a relationship type set, a role type set, and a constraint rule set. A generation module is configured to generate a path template based on the semantic baseline. An excavation module is configured to excavate an initial path satisfying the semantic baseline from a preset knowledge graph based on the path template, and acquire a target path of the target object through closed-loop consistency verification, the target path being used to represent an associated relationship corresponding to the target object.

10. An electronic device, comprising: Comprise: A processor, and a memory in communication connection with the processor; The memory stores computer execution instructions; The processor executes the computer execution instructions stored in the memory to implement the method of any one of claims 1 to 8.

11. A computer readable storage medium, characterized in that, The computer readable storage medium stores computer execution instructions, and the computer execution instructions are executed by the processor to implement the method of any one of claims 1 to 8.

12. A computer program product, characterised in that, The computer program is executed by the processor to implement the method of any one of claims 1 to 8.

Images