Method for checking multi-source property right data

Abstract

The application discloses a kind of verification methods of multi-source property right data, it is related to data processing technical field.The above-mentioned method obtains property right data from different data sources, according to the semantics of property right data and preset mapping rule, property right data is converted into uniform field;After building knowledge graph based on uniform field, business rule vector is encoded into the node attribute of knowledge graph;The node and edge in knowledge graph are traversed by preset graph traversal algorithm, and the property right data is verified according to business rule vector, and the verification result is obtained.The method integrates the association between entities by building knowledge graph, encodes business rules into knowledge graph for verification, so that the verification is no longer limited to single node data, but can be combined with cross-node verification of associated information, to improve the accuracy of verification.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a method for verifying multi-source property rights data. Background Technology

[0002] As property rights management scenarios become increasingly complex, the sources of property rights data are becoming more diversified, encompassing official property rights registration platforms, internal enterprise property rights management databases, third-party data service providers, and industry association information systems. Differences in construction standards, field definitions, and data formats among different data sources lead to fragmented and heterogeneous property rights data, posing challenges to unified data verification and application.

[0003] In the industry, the verification of multi-source property rights data mainly involves converting business rules into fixed code logic and performing batch verification on the pre-processed multi-source data. While this method achieves multi-source data coverage, it can only perform isolated verification on a single field and cannot leverage the relationships between fields for cross-entity verification, resulting in low verification accuracy.

[0004] The above content is only used to help understand the technical solution of this application and does not represent an admission that the above content is prior art. Summary of the Invention

[0005] The main purpose of this application is to provide a method for verifying multi-source property rights data, aiming to solve the technical problem of low verification accuracy of multi-source property rights data.

[0006] To achieve the above objectives, this application proposes a method for verifying multi-source property rights data, the method comprising:

[0007] Property rights data is obtained from different data sources, and the property rights data is converted into a unified field according to the semantics of the property rights data and the preset mapping rules.

[0008] After constructing a knowledge graph based on unified fields, business rule vectors are encoded into the node attributes of the knowledge graph;

[0009] The nodes and edges in the knowledge graph are traversed by a preset graph traversal algorithm, and the property rights data is verified according to the business rule vector to obtain the verification result.

[0010] In one embodiment, after the steps of obtaining property rights data from different data sources and converting the property rights data into a unified field according to the semantics of the property rights data and a preset mapping rule, the method further includes: determining the entities in the unified field corresponding to the different data sources and the entity type corresponding to each entity;

[0011] If entities from different data sources corresponding to the same entity type are consistent, then a relation triplet between entities corresponding to different entity types is determined. The relation triplet includes a head entity, a tail entity, and the relation type and relation attribute between the head entity and the tail entity.

[0012] The entities in the relation triples are used as nodes in the knowledge graph;

[0013] The relationship type and relationship attribute between the two entities in the relationship triple are used as edges of the knowledge graph.

[0014] In one embodiment, after the step of determining the entities in the unified field corresponding to different data sources and the entity type corresponding to each entity, the method further includes:

[0015] If the entities corresponding to the same entity type from different data sources are inconsistent, the confidence level of each data source is determined based on a preset evaluation dimension.

[0016] The entity corresponding to the data source with the highest confidence level shall be the entity corresponding to the entity type.

[0017] The steps include determining the relation triples between entities corresponding to different entity types, using the entities in the relation triples as nodes of the knowledge graph, and using the relation type and relation attribute between the two entities in the relation triples as edges of the knowledge graph.

[0018] In one embodiment, the step of traversing the nodes and edges in the knowledge graph using a preset graph traversal algorithm and verifying the property rights data according to the business rule vector includes:

[0019] Get the N-degree neighborhood subgraph of the target node currently being traversed;

[0020] The N-degree neighborhood subgraph and the business rule vector corresponding to the target node are input into the graph neural network model, and the graph neural network model generates an embedding vector for the target node based on the N-degree neighborhood subgraph.

[0021] The graph neural network model concatenates the embedded vector with each of the business rule vectors to obtain a fused feature vector. Then, a nonlinear transformation is performed on each of the fused feature vectors according to a preset activation function to obtain the transformed activation value of each of the fused feature vectors output by the graph neural network model, which serves as the risk value corresponding to each of the business rule vectors.

[0022] The risk value corresponding to each of the business rule vectors for each of the nodes is used as the verification result.

[0023] In one embodiment, the step of generating an embedding vector for the target node using the graph neural network model based on the N-degree neighborhood subgraph includes:

[0024] In the r-th iteration, the r-degree neighborhood nodes in the N-degree neighborhood subgraph are determined, wherein the r-degree neighborhood nodes represent nodes connected to the target node by r edges;

[0025] The entity type corresponding to the r-degree neighbor node is used as the initial feature vector of the r-degree neighbor node, and the relation type and relation attribute corresponding to each edge are used as the initial feature vector of the edge.

[0026] The initial feature vector of the r-degree neighboring node is concatenated with the initial feature vector of the edge between the center node obtained in the (r-1)th iteration to obtain the concatenated vector;

[0027] The embedding vector of the center node obtained in the r-th iteration is determined by the dot product of the parameter matrix corresponding to the r-th layer of the graph neural network model and the concatenated vector.

[0028] When the Nth iteration is completed, the embedding vector of the center node obtained in the Nth iteration is used as the embedding vector of the target node, where r is less than N.

[0029] In one embodiment, the step of determining the confidence level of each of the data sources based on a preset evaluation dimension includes:

[0030] Based on the update time of the different data sources obtained, the timeliness coefficient of the data source is determined;

[0031] The integrity coefficient of each data source is determined based on the data proportion of each data source in the historical property rights data form;

[0032] The consistency coefficient of each data source is determined based on the consistency ratio of the property rights data between each data source and other data sources.

[0033] The confidence level of a data source is obtained by weighted summing of the authority coefficient, timeliness coefficient, completeness coefficient, and consistency coefficient of each data source.

[0034] In one embodiment, after the steps of traversing the nodes and edges in the knowledge graph using a preset graph traversal algorithm, verifying the property rights data according to the business rule vector, and obtaining the verification result, the method further includes:

[0035] In response to the selection of a preset report template, the property rights data and the verification results are used to generate a verification report, which is then displayed on the interactive interface according to the output format of the preset report template.

[0036] In one embodiment, after the step of generating a verification report and displaying it on the interactive interface according to the output format of the preset report template based on the property rights data and the verification result, the method further includes:

[0037] The knowledge graph is displayed on the interactive interface;

[0038] If any of the business rule vectors at the node has a risk value greater than a preset risk threshold, an abnormal warning message will be displayed at the node.

[0039] In response to a click on the node, a prompt window is displayed, showing the rule item corresponding to the business rule vector.

[0040] This application provides a method for verifying multi-source property rights data. By obtaining property rights data from different data sources, the property rights data is converted into unified fields according to the semantics of the property rights data and preset mapping rules. After constructing a knowledge graph based on the unified fields, business rule vectors are encoded into the node attributes of the knowledge graph. The nodes and edges in the knowledge graph are traversed through a preset graph traversal algorithm, and the property rights data is verified according to the business rule vectors to obtain the verification result.

[0041] The above method uses a unified field transformation to eliminate format and semantic differences between multi-source data. When entities are inconsistent, a confidence calculation mechanism is used. By quantifying the reliability of data sources from multiple dimensions, the optimal entity is selected as the standard to solve the problem of multi-source data conflict and avoid the situation of multiple nodes or contradictory attributes of the same entity in the graph, thus ensuring the accuracy of the graph data. Then, by constructing a knowledge graph to integrate the relationship between entities, business rules are encoded into the knowledge graph for verification. This makes the verification no longer limited to single node data, but can combine the relationship information to perform cross-node verification, thereby improving the accuracy of the verification. Attached Figure Description

[0042] 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.

[0043] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 A flowchart illustrating the first embodiment of the verification method for multi-source property rights data in this application;

[0045] Figure 2A flowchart illustrating Embodiment 2 of the method for verifying multi-source property rights data in this application;

[0046] Figure 3 A flowchart illustrating Embodiment 3 of the method for verifying multi-source property rights data in this application;

[0047] Figure 4 This is a schematic diagram of the device structure of the hardware operating environment involved in the verification method for multi-source intellectual property data in this application embodiment.

[0048] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0049] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of this application and are not intended to limit this application.

[0050] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific embodiments. It should be noted that all actions involving the acquisition of signals, information, or data in this application are performed in accordance with the relevant data protection laws and regulations of the country where the application is located, and with authorization from the owner of the corresponding device.

[0051] As property rights management scenarios become increasingly complex, the sources of property rights data are becoming more diversified, encompassing official property rights registration platforms, internal enterprise property rights management databases, third-party data service providers, and industry association information systems. Differences in construction standards, field definitions, and data formats among different data sources lead to fragmented and heterogeneous property rights data, posing challenges to unified data verification and application.

[0052] In the industry, the verification of multi-source property rights data mainly involves converting business rules into fixed code logic and performing batch verification on the pre-processed multi-source data. While this method achieves multi-source data coverage, it can only perform isolated verification on a single field and cannot leverage the relationships between fields for cross-entity verification, resulting in low verification accuracy.

[0053] In view of the above problems, this application proposes a method for verifying multi-source property rights data. By obtaining property rights data from different data sources, the property rights data is converted into unified fields according to the semantics of the property rights data and preset mapping rules. After constructing a knowledge graph based on the unified fields, business rule vectors are encoded into the node attributes of the knowledge graph. The nodes and edges in the knowledge graph are traversed by a preset graph traversal algorithm, and the property rights data is verified according to the business rule vectors to obtain the verification result.

[0054] The above method uses a unified field transformation to eliminate format and semantic differences between multi-source data. When entities are inconsistent, a confidence calculation mechanism is used. By quantifying the reliability of data sources from multiple dimensions, the optimal entity is selected as the standard to solve the problem of multi-source data conflict and avoid the situation of multiple nodes or contradictory attributes of the same entity in the graph, thus ensuring the accuracy of the graph data. Then, by constructing a knowledge graph to integrate the relationship between entities, business rules are encoded into the knowledge graph for verification. This makes the verification no longer limited to single node data, but can combine the relationship information to perform cross-node verification, thereby improving the accuracy of the verification.

[0055] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer or personal computer, or an electronic device or multi-source intellectual property data verification system capable of performing the above functions. The following description uses a multi-source intellectual property data verification system as an example to illustrate this embodiment and the subsequent embodiments.

[0056] Based on this, the first embodiment of this application provides a method for verifying multi-source property rights data, referring to... Figure 1 In this embodiment, the verification method for multi-source property rights data includes steps S10 to S30:

[0057] Step S10: Obtain property rights data from different data sources, and convert the property rights data into a unified field according to the semantics of the property rights data and the preset mapping rules.

[0058] It should be noted that the preset mapping rules are a pre-defined set of rules used to associate fields from different data sources with a unified field. These data sources may include official property registration platforms, internal enterprise property management systems, and third-party data service providers. The preset mapping rules clearly define which field under the unified standard corresponds to a certain field from different data sources, and also define the conversion standards for field data types and formats. The unified field is a set of fields defined based on property business scenarios, with standardized names, meanings, and formats. For example, it unifies the "Enterprise Name" field from data source A and the "Company Name" field from data source B into the "Property Entity Name" field, which serves as the basic data unit for subsequent knowledge graph construction.

[0059] For example, the multi-source property rights data verification system accesses the interfaces of each data source, extracts the original property rights data from each data source, and forms an original dataset; then, it performs semantic parsing on each field in the original dataset to determine the actual meaning of the field, and then calls the preset mapping rule library to match the semantic parsing results with the unified field standard in the preset mapping rule library; based on the matching results, it converts the name, data type, format, etc. of the original fields so that the property rights data from all sources conforms to the unified field specification.

[0060] Step S20: After constructing the knowledge graph based on the unified fields, the business rule vectors are encoded into the node attributes of the knowledge graph.

[0061] Business rule vectors refer to the transformation of unstructured business verification rules, such as text descriptions, into numerical vectors through vector encoding algorithms such as word embedding and rule structured encoding, so that they can be recognized and calculated by computer algorithms, facilitating subsequent automated verification through graph traversal.

[0062] For example, the multi-source property rights data verification system parses entity information in a unified field, constructs nodes and edges of a knowledge graph based on the relationships between entities, and assigns node attributes to the unified field values ​​of the corresponding entities. Then, the system reads a pre-set business rule base, structurally decomposes each business rule, extracts rule elements such as entity type, attribute type, and verification conditions involved in each rule, and transforms these decomposed rule elements into numerical vectors using a vector encoding algorithm, serving as business rule vectors. Finally, based on the entity type corresponding to each business rule, the business rule vectors are associated with the attribute set of the corresponding node in the knowledge graph, forming a knowledge graph containing property rights data and rule vectors.

[0063] Optionally, after obtaining property rights data from different data sources and converting the property rights data into a unified field according to the semantics of the property rights data and preset mapping rules, the step of constructing a knowledge graph based on the unified field includes steps S21 to S24:

[0064] Step S21: Determine the entities in the unified field corresponding to different data sources and the entity type corresponding to each entity.

[0065] For example, the multi-source property rights data verification system parses the fields in a unified field format dataset and extracts objects with independent semantics as entities. Based on preset entity classification standards, such as subject class, target class, and institution class in property rights transactions, it matches each extracted entity with a corresponding entity type.

[0066] Step S22: If entities from different data sources corresponding to the same entity type are consistent, then determine the relation triplet between entities corresponding to different entity types. The relation triplet includes a head entity, a tail entity, and the relation type and relation attribute between the head entity and the tail entity.

[0067] In this context, a relation triple is a structured data unit used to represent the association between entities. It includes a head entity (the initiator of the association), a tail entity (the receiver of the association), a relation type (the type of association between entities, such as "ownership"), and relation attributes. Relationship attributes provide supplementary descriptions of the association, such as registration time and association validity period.

[0068] For example, the multi-source property rights data verification system first determines whether the field values ​​corresponding to entities of the same entity type from different data sources are consistent. If they are consistent, it continues to analyze the relationships between entities corresponding to different entity types.

[0069] Optionally, the steps for extracting relationships between entities include training an attribute association model during the initialization or periodic optimization of the multi-source property rights data verification system. Specifically, the multi-source property rights data verification system can internally store a historical property rights dataset, including cleaned and standardized property rights data from multiple data sources. The system groups all data records in the historical property rights dataset that describe the same relationship; for example, it groups all data records containing the relationship type "holding equity" together. Within each group, it analyzes all field values ​​appearing in each data record and counts the frequency of any two field values ​​appearing simultaneously in the same record, as the collinearity frequency of the associated fields. Then, the system uses a clustering algorithm to perform a second grouping of fields with co-occurrence frequencies greater than a preset threshold. Each group of fields after the second grouping can be understood as describing the same relationship attribute from different perspectives. Therefore, the system can generate a standardized attribute name for each cluster as the relationship attribute. For example, it can use a domain dictionary to find a hypernym for this group of fields as the relationship attribute, or it can select a representative field name from the cluster as the relationship attribute.

[0070] Step S23: The entities in the relation triples are used as nodes of the knowledge graph.

[0071] Step S24: Use the relationship type and relationship attribute between the two entities in the relationship triple as the edge of the knowledge graph.

[0072] The process iterates through the set of relation triples, extracting the head and tail entities contained in each triple. The extracted entities are deduplicated to ensure that each entity corresponds to only one node in the knowledge graph. The deduplicated entities are then used as nodes in the knowledge graph, and each node is associated with its attribute information from the unified field dataset, such as the entity's name (i.e., the unified field value) and code. The resulting set of nodes in the knowledge graph is then output. Next, the relation type and relation attributes from the triples are extracted and used as edges connecting the head and tail entity nodes, thus constructing the unified field dataset into a knowledge graph.

[0073] Optionally, after step S21, steps S25 to S27 are also included:

[0074] Step S25: If the entities corresponding to different data sources of the same entity type are inconsistent, then the confidence level of each data source is determined based on the preset evaluation dimension.

[0075] If there are records of the same entity type from different data sources with inconsistent entity information, such as "Company A" and "Limited Company A" for the same company in Data Source A and Data Source B respectively, then a preset evaluation dimension and its weights will be loaded, and the confidence level of each data source for the inconsistent entities will be determined according to the preset evaluation dimension.

[0076] Optionally, step S25 includes steps S251 to S254:

[0077] Step S251: Determine the timeliness coefficient of the data source based on the obtained update time of the different data sources.

[0078] For example, the multi-source property rights data verification system reads the metadata of each property rights data source, extracts the original value of the property rights data update time field in the metadata, and simultaneously reads the current time of the multi-source property rights data verification system. It then performs a difference calculation on the property rights data update time and the current time of the multi-source property rights data verification system to obtain the corresponding update time difference for each data source. Based on this update time difference, it determines the timeliness coefficient of the data source; the smaller the update time difference, the larger the timeliness coefficient of the data source.

[0079] Step S252: Determine the integrity coefficient of the data source based on the data proportion of each data source in the historical property rights data form.

[0080] For example, the multi-source property rights data verification system counts the number of data records corresponding to each data source in the verified and corrected historical property rights data forms, denoted as the sample size n_i of each data source. For instance, the sample size of data source A in the historical property rights data forms is n_A, and the sample size of data source B is n_B. If a data source has no records in the historical property rights data forms, its sample size is set to 0. Then, the multi-source property rights data verification system performs a percentage calculation on each data source: Data percentage = (Sample size of the data source n_i / Total sample size of the entire historical data set) × 100%. The integrity coefficient of the data source is determined based on this data percentage, where a larger data percentage results in a larger integrity coefficient.

[0081] Step S253: Determine the consistency coefficient of the data source based on the consistency ratio of the property rights data between each data source and other data sources.

[0082] For example, a multi-source property rights data verification system selects a target data source, such as data source A, from the data sources. Simultaneously, it filters out all other data sources belonging to the same entity type as the target data source, such as data source B, data source C, etc. It extracts the uniform field values ​​corresponding to all entities in the target data source and also extracts the uniform field values ​​corresponding to the same entities in each of the other data sources. The system compares the uniform field values ​​corresponding to all entities in the target data source with those in a single comparison data source to determine the number of entities with the same uniform field values. Based on the ratio of the number of entities in the target data source with the same uniform field values ​​as those in other data sources to the total number of entities in the target data source, it determines the consistency ratio of the property rights data between the target data source and other data sources. Based on this consistency ratio, it determines the consistency coefficient of the target data source, where a higher consistency ratio results in a higher consistency coefficient.

[0083] Step S254: The authority coefficient, timeliness coefficient, integrity coefficient and consistency coefficient of each data source are weighted and summed to obtain the confidence level of the data source.

[0084] The multi-source property rights data verification system uses a pre-defined data source authority coefficient mapping table. This table records the authority coefficients of different data sources, which can be customized by the user. For example, the coefficient for the official property registration platform is higher than that for third-party data service providers. After determining these coefficients, they are weighted and summed to calculate the data source with the highest confidence level. When the property rights data from different data sources are inconsistent, the property rights data from the data source with the highest confidence level is used to improve the accuracy of the initial property rights data.

[0085] Step S26: The entity corresponding to the data source with the highest confidence level is taken as the entity corresponding to the entity type.

[0086] Step S27: Execute the steps of determining the relation triples between entities corresponding to different entity types, using the entities in the relation triples as nodes of the knowledge graph, and using the relation type and relation attribute between the two entities in the relation triples as edges of the knowledge graph.

[0087] For example, by comparing the confidence values ​​of various data sources under the same entity type, the data source with the highest confidence is determined, and the entity data corresponding to that data source is determined as the standard entity corresponding to the current entity type.

[0088] Understandably, in a knowledge graph, the same node, or entity, should have only one unified value. For example, the business scope of a company might be described using different field values ​​in different data sources, such as investment management or investment services, or some data sources might omit certain types of business scopes. In this case, the business scope data from the data source with the highest confidence level is used as the value corresponding to the "business scope" node in the knowledge graph. This avoids the problem of multiple nodes for the same entity or contradictory node attributes in the knowledge graph, thus improving the accuracy of subsequent data validation.

[0089] Step S30: Traverse the nodes and edges in the knowledge graph using a preset graph traversal algorithm, and verify the property rights data according to the business rule vector to obtain the verification result.

[0090] For example, the multi-source property rights data verification system initiates a preset graph traversal algorithm, sequentially accessing all nodes and associated edges in the knowledge graph according to a set path. For each accessed node, the multi-source property rights data verification system extracts the unified field values ​​and associated business rule vectors from its node attributes.

[0091] Optionally, the multi-source property rights data verification system can verify single-node attribute rules: it compares the unified field values ​​of a node with the corresponding business rule vectors to determine whether the data conforms to the rules. The system can also verify cross-node association rules: it locates associated nodes through edge information, extracts the attribute data of associated nodes, and compares the current node data with the corresponding business rule vectors to determine whether the associated data conforms to the rules. The system records the verification result of each business rule vector. For items that do not conform to the rules, it records the corresponding nodes, attributes, and rule information, and outputs a verification report containing the verification results of all business rule vectors.

[0092] In this embodiment, unified field conversion eliminates format and semantic differences between multi-source data, laying a unified data foundation for subsequent processing and avoiding verification deviations caused by inconsistent fields. The confidence calculation mechanism for inconsistent entities quantifies the reliability of data sources from multiple dimensions, selecting the optimal entity as the standard, effectively resolving multi-source data conflicts, and preventing the same entity from having multiple nodes or contradictory attributes in the knowledge graph, thus ensuring the accuracy of the knowledge graph data. This embodiment integrates the relationships between entities by constructing a knowledge graph, achieving standardization, accuracy, and efficiency in multi-source property rights data verification.

[0093] Based on the first embodiment of this application, in the second embodiment of this application, the content that is the same as or similar to that in Embodiment 1 above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 2 Step S30 includes steps S31 to S34:

[0094] Step S31: Obtain the N-degree neighborhood subgraph of the target node currently being traversed.

[0095] It should be noted that the N-degree neighborhood subgraph is a subgraph centered on the target node, containing all nodes whose distance to the target node is between 1 and N, as well as the edges between these nodes. Here, distance refers to the number of edges in the shortest path between two nodes.

[0096] For example, the multi-source property rights data verification system determines the target node currently being traversed, loads a preset N value, and starts searching for neighboring nodes from the target node: first, it finds nodes directly connected to the target node and marks them as 1-degree neighboring nodes, while recording the edges connecting the target node and the 1-degree neighboring nodes; then, starting from each 1-degree neighboring node, it finds nodes directly connected to it that are not marked, marks them as 2-degree neighboring nodes, and records the edges connecting the 1-degree neighboring nodes and the 2-degree neighboring nodes; repeating the above steps until the target node and all 1 to N-degree neighboring nodes, as well as all edges between these nodes, are found, forming an N-degree neighborhood subgraph.

[0097] Step S32: Input the N-degree neighborhood subgraph and the business rule vector corresponding to the target node into the graph neural network model, and generate an embedding vector for the target node based on the N-degree neighborhood subgraph through the graph neural network model.

[0098] Optionally, step S32 includes steps S321 to S325:

[0099] Step S321: In the r-th iteration, determine the r-degree neighboring nodes in the N-degree neighborhood subgraph, wherein the r-degree neighboring nodes represent nodes connected to the target node by r edges.

[0100] For example, the multi-source property rights data verification system reads structured data from an N-degree neighborhood subgraph, denoting nodes within the N-degree neighborhood subgraph as node_id and edges as edge_list, in the format [source node_id, target node_id]. The target node is denoted as target_node_id.

[0101] The multi-source property rights data verification system first uses `target_node_id` as the starting point and traverses `edge_list` to construct the path relationships between nodes. When r=1, it directly filters nodes in `edge_list` that are directly connected to `target_node_id`, i.e., the other node corresponding to the edge where the source node or target node is `target_node_id`, and treats them as 1-degree neighbors. When r>1, it first obtains the list of r-1 degree neighbors determined in the (r-1)th round, and then filters nodes in `edge_list` that are directly connected to r-1 degree neighbors and are not marked as 1~r-1 degree neighbors, treating them as r-degree neighbors. In the above process, the multi-source property rights data verification system records the shortest path between each r-degree neighbor node and `target_node_id`, ensuring that the distance of the shortest path is r.

[0102] Step S322: The entity type corresponding to the r-degree neighbor node is used as the initial feature vector of the r-degree neighbor node, and the relation type and relation attribute corresponding to each edge are used as the initial feature vector of the edge.

[0103] For example, for each r-degree neighboring node, the entity type of that node is queried, and the entity type is converted into a d1-dimensional numerical vector, which serves as the initial feature vector for the r-degree neighboring node. Simultaneously, for each r-degree neighboring node, the edge between it and the intermediate node of degree r-1 is determined, and the relation type and relation attribute of that edge are extracted. The relation type is converted into a d2-dimensional vector, and the relation attribute is converted into a d3-dimensional vector. Then, the d2-dimensional and d3-dimensional vectors are sequentially concatenated to obtain the initial feature vector of the edge.

[0104] Step S323: Concatenate the initial feature vector of the edge between the r-degree neighbor node and the center node obtained in the (r-1)th iteration with the initial feature vector of the r-degree neighbor node to obtain the concatenated vector.

[0105] It should be noted that the central node obtained in the (r-1)th iteration is the intermediate feature carrier of the target node after absorbing the information of the (r-1)-degree neighborhood. It represents the target node that has fused the feature information of all neighboring nodes within the (r-1)-degree range.

[0106] Exemplarily, among the r-degree neighborhood nodes of the target node, each node is denoted as node_r, and the central node in the (r - 1)-th round is denoted as vec_center_{r - 1}. The edge between the r-degree neighborhood node and the central node obtained in the (r - 1)-th round of iteration, that is, the edge between the r-degree node node_r and the (r - 1)-degree node node_{r - 1} directly connected to it. Since the information of node_{r - 1} has been fully aggregated into vec_center_{r - 1}, the edge between node_r and node_{r - 1} is the associated edge between node_r and vec_center_{r - 1}.

[0107] Step S324, determine the embedding vector of the central node obtained in the r-th round of iteration according to the dot product of the parameter matrix corresponding to the r-th layer of the graph neural network model of the graph neural network and the concatenated vector.

[0108] The multi-source property data verification system reads the parameter matrix of the r-th layer of the graph neural network from the weight file of the graph neural network model. For each concatenated vector x_i obtained by concatenation in the r-th round, that is, the concatenated vector corresponding to the i-th r-degree neighborhood node, the parameter matrix and the concatenated vector are multiplied matrix-wise to determine the embedding vector of the central node obtained in the r-th round of iteration.

[0109] Optionally, a bias vector corresponding to the r-th layer of the graph neural network can also be set to adjust the offset of the dot product result and avoid the model output from being too concentrated. After obtaining the dot product of the parameter matrix and the concatenated vector, the sum of the dot product and the bias vector is used as the embedding vector of the central node obtained in the r-th round of iteration.

[0110] Step S325, when the N-th round of iteration is completed, use the embedding vector of the central node obtained in the N-th round of iteration as the embedding vector of the target node, where r is less than N.

[0111] The multi-source property data verification system reads the current iteration round r and the preset maximum iteration round N. If r < N, increment r by 1, return to step S321, and continue the next round of iteration; if r = N: trigger the iteration termination condition, stop subsequent iterations, and retrieve the embedding vector of the central node generated in the N-th round from the iteration cache as the embedding vector of the target node target_node_id.

[0112] Step S33, splice the embedding vector with each of the business rule vectors through the graph neural network model to obtain a fused feature vector, and perform a non-linear transformation on each of the fused feature vectors according to a preset activation function to obtain the activation value after transformation of each of the fused feature vectors output by the graph neural network model as the risk value corresponding to each of the business rule vectors.

[0113] After determining the embedding vector of the target node, the embedding vector of the target node is concatenated with the individual business rule vector along its dimensions to form a fused feature vector. This fused feature vector contains both the feature information of the node and the feature information of the rule. Then, each fused feature vector is input into the output layer of the graph neural network model, and a preset activation function is called to perform a non-linear transformation on the fused feature vector, mapping the value of the fused feature vector to a specific range. The activation function can be a Sigmoid function, ReLU function, or Tanh function, etc. These activation functions calculate the value of each dimension of the fused feature vector and finally output a value, which is the risk value of the corresponding business rule vector.

[0114] Understandably, when training a graph neural network model, known risk labels can be compared with the output values ​​of the activation function, and the model parameters can be adjusted through the loss function. Ultimately, high activation values ​​correspond to higher risk values, and low activation values ​​correspond to lower risk values.

[0115] Step S34: The risk value corresponding to each of the business rule vectors corresponding to each node is used as the verification result.

[0116] In this embodiment, the introduction of the N-degree neighborhood subgraph allows the verification to no longer rely solely on the target node's own data, but instead incorporates information from its associated nodes, avoiding misjudgments due to incomplete information. The multi-round iterative aggregation of the graph neural network enables the target node's embedded vector to absorb structural and attribute information from its 1 to N-degree neighborhood layer by layer, resulting in a more comprehensive feature representation. Compared to a single feature vector, this better reflects the true association state of property rights data, improving the reliability of the calibration risk value and enhancing the accuracy of multi-source property rights data verification.

[0117] Based on the above embodiments of this application, in the third embodiment of this application, the same or similar content as the above embodiments can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 3 After step S30, the verification method for multi-source property rights data further includes step S40:

[0118] Step S40: In response to the selection of the preset report template, the property rights data and the verification results are used to generate a verification report and display it on the interactive interface according to the output format of the preset report template.

[0119] For example, users can select a report template from a list of preset templates in an interactive interface such as a software client or web platform by clicking drop-down menus or radio buttons. The multi-source property rights data verification system then matches the property rights data with the corresponding verification results and fills them into the designated positions of the report template according to the preset rules of the selected template. Optionally, the multi-source property rights data verification system converts the completed report into an interactive format and displays it in the report viewing panel of the interactive interface, where users can scroll through and search for specific node reports.

[0120] Optionally, after step S40, steps S50 to S70 are also included:

[0121] Step S50: Display the knowledge graph on the interactive interface.

[0122] Step S60: If any of the business rule vectors in the node has a risk value greater than a preset risk threshold, an abnormal warning message is displayed at the node.

[0123] Step S70: In response to the click action of the node, a prompt window is displayed, and the rule item corresponding to the business rule vector is displayed in the prompt window.

[0124] For example, the multi-source property rights data verification system can assign different visual identifiers to nodes of different entity types. The node size can be adjusted according to the number of associated edges, and the knowledge graph is displayed on an interactive interface. For each node, the multi-source property rights data verification system checks the risk values ​​corresponding to all its business rules. If any rule's risk value exceeds a preset risk threshold, the node is marked as an abnormal node, and a visual warning identifier is overlaid on the abnormal node, which may include color warnings, dynamic warnings, or label warnings. When a user clicks on any node on the knowledge graph in the interactive interface, the multi-source property rights data verification system queries the business rule vector and risk value corresponding to the clicked node based on the node's unique identifier, and generates a floating prompt window on the interactive interface, displaying the rule details in a preset format.

[0125] In this embodiment, after the multi-source property rights data is verified, it first responds to the user's selection of a preset report template, matching and filling the property rights data and verification results according to the template's preset format to generate an interactive verification report that can be scrolled and searched, and displays it on the interface. Then, a knowledge graph is displayed on the interface, and for nodes where any business rule risk value exceeds a preset threshold, color, dynamic, or tag-based anomaly warnings are overlaid. When the user clicks on a node in the graph, the multi-source property rights data verification system queries the corresponding rule information based on the node's unique identifier, displaying rule details in a pop-up window. The interactive report and visual graph transform abstract data into an intuitive form, ensuring that users can clearly grasp entity relationships and specific risk points, providing accurate basis for subsequent property rights data correction and compliance management.

[0126] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the verification method of multi-source property data in this application. Any simple modifications based on this technical concept are within the protection scope of this application.

[0127] This application provides a verification device for multi-source intellectual property data. The verification device for multi-source intellectual property data includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the multi-source intellectual property data verification method in the above embodiment 1.

[0128] The following is for reference. Figure 4 The diagram illustrates a structural schematic of a verification device suitable for implementing embodiments of this application for multi-source intellectual property data. The verification device for multi-source intellectual property data in the embodiments of this application may include, but is not limited to, mobile terminals such as laptops and tablets (PADs), and fixed terminals such as desktop computers. Figure 4 The verification device for multi-source property data shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0129] like Figure 4As shown, the verification device for multi-source property data may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM) 1004. The RAM 1004 also stores various programs and data required for the operation of the verification device for multi-source property data. The processing unit 1001, the read-only memory 1002, and the RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to I / O interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. Communication device 1009 allows the verification device for multi-source proprietary data to communicate wirelessly or wiredly with other devices to exchange data. Although a verification device for multi-source proprietary data with various systems is shown in the figure, it should be understood that it is not required to implement or possess all the systems shown. More or fewer systems may be implemented alternatively.

[0130] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from read-only memory 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0131] The verification device for multi-source intellectual property data provided in this application, employing the verification method for multi-source intellectual property data described in the above embodiments, can solve the technical problem of low verification accuracy of multi-source intellectual property data. Compared with the prior art, the beneficial effects of the verification device for multi-source intellectual property data provided in this application are the same as those of the verification method for multi-source intellectual property data provided in the above embodiments, and other technical features in this verification device are the same as those disclosed in the method of the previous embodiment, and will not be repeated here.

[0132] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0133] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0134] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, which are used to execute the verification method for multi-source property data in the above embodiments.

[0135] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), or flash memory, optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, radio frequency (RF), etc., or any suitable combination thereof.

[0136] The aforementioned computer-readable storage medium may be included in a verification device for multi-source property data; or it may exist independently and not be assembled into a verification device for multi-source property data.

[0137] The aforementioned computer-readable storage medium carries one or more programs that, when executed by a verification device for multi-source intellectual property data, enable the verification device to write computer program code for performing the operations of this application in one or more programming languages ​​or a combination thereof. These programming languages ​​include object-oriented programming languages—such as Java, Smalltalk, and C++—and conventional procedural programming languages—such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, or as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0138] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0139] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0140] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described verification method for multi-source intellectual property data, thereby solving the technical problem of low verification accuracy of multi-source intellectual property data. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the multi-source intellectual property data verification method provided in the above embodiments, and will not be repeated here.

[0141] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the multi-source property data verification method described above.

[0142] The computer program product provided in this application can solve the technical problem of low accuracy in the verification of multi-source intellectual property data. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the multi-source intellectual property data verification method provided in the above embodiments, and will not be repeated here.

[0143] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A method for verifying multi-source property rights data, characterized in that, The verification method for the multi-source property rights data includes: Property rights data is obtained from different data sources, and the property rights data is converted into a unified field according to the semantics of the property rights data and the preset mapping rules. After constructing a knowledge graph based on unified fields, business rule vectors are encoded into the node attributes of the knowledge graph; The nodes and edges in the knowledge graph are traversed by a preset graph traversal algorithm to obtain the N-degree neighborhood subgraph of the target node currently traversed. The N-degree neighborhood subgraph is a subgraph centered on the target node and includes all nodes with a distance of 1 to N from the target node and the edges between the nodes. The N-degree neighborhood subgraph and the business rule vector corresponding to the target node are input into the graph neural network model, and the graph neural network model generates an embedding vector for the target node based on the N-degree neighborhood subgraph. The graph neural network model concatenates the embedded vector with each of the business rule vectors to obtain a fused feature vector. Then, a nonlinear transformation is performed on each of the fused feature vectors according to a preset activation function to obtain the transformed activation value of each of the fused feature vectors output by the graph neural network model, which serves as the risk value corresponding to each of the business rule vectors. The risk value corresponding to each of the business rule vectors for each node is used as the verification result.

2. The method for verifying multi-source property rights data as described in claim 1, characterized in that, After the steps of obtaining property rights data from different data sources and converting the property rights data into a unified field according to the semantics of the property rights data and preset mapping rules, the method further includes: Determine the entities in the unified field corresponding to different data sources and the entity type corresponding to each entity; If entities from different data sources corresponding to the same entity type are consistent, then a relation quadruple is determined between entities corresponding to different entity types. The relation quadruple includes a head entity, a tail entity, and the relationship type and relation attributes between the head entity and the tail entity. The entities in the relation quadruple are used as nodes in the knowledge graph; The relationship type and relationship attribute between two entities in the relation quadruple are used as edges of the knowledge graph.

3. The verification method for multi-source property rights data as described in claim 2, characterized in that, After the step of determining the entities in the unified field corresponding to different data sources and the entity type corresponding to each entity, the method further includes: If the entities corresponding to the same entity type from different data sources are inconsistent, the confidence level of each data source is determined based on a preset evaluation dimension. The entity corresponding to the data source with the highest confidence level shall be the entity corresponding to the entity type. The steps include determining the relation quadruples between entities corresponding to different entity types, using the entities in the relation quadruples as nodes of the knowledge graph, and using the relation type and relation attribute between two entities in the relation quadruples as edges of the knowledge graph.

4. The method for verifying multi-source property rights data as described in claim 1, characterized in that, The step of generating an embedding vector for the target node using the graph neural network model based on the N-degree neighborhood subgraph includes: In the r-th iteration, the r-degree neighborhood nodes in the N-degree neighborhood subgraph are determined, wherein the r-degree neighborhood nodes represent nodes connected to the target node by r edges; The entity type corresponding to the r-degree neighbor node is used as the initial feature vector of the r-degree neighbor node, and the relation type and relation attribute corresponding to each edge are used as the initial feature vector of the edge. The initial feature vector of the r-degree neighboring node is concatenated with the initial feature vector of the edge between the center node obtained in the (r-1)th iteration to obtain the concatenated vector; The embedding vector of the center node obtained in the r-th iteration is determined by the dot product of the parameter matrix corresponding to the r-th layer of the graph neural network model and the concatenated vector. When the Nth iteration is completed, the embedding vector of the center node obtained in the Nth iteration is used as the embedding vector of the target node, where r is less than N.

5. The method for verifying multi-source property rights data as described in claim 3, characterized in that, The step of determining the confidence level of each data source based on a preset evaluation dimension includes: Based on the update time of the different data sources obtained, the timeliness coefficient of the data source is determined; The integrity coefficient of each data source is determined based on the data proportion of each data source in the historical property rights data form; The consistency coefficient of each data source is determined based on the consistency ratio of the property rights data between each data source and other data sources. The confidence level of a data source is obtained by weighted summing of the authority coefficient, timeliness coefficient, completeness coefficient, and consistency coefficient of each data source.

6. The method for verifying multi-source property rights data as described in claim 1, characterized in that, After the step of using the risk value corresponding to each of the business rule vectors corresponding to each of the nodes as the verification result, the method further includes: In response to the selection of a preset report template, the property rights data and the verification results are used to generate a verification report, which is then displayed on the interactive interface according to the output format of the preset report template.

7. The method for verifying multi-source property rights data as described in claim 6, characterized in that, After the step of generating a verification report and displaying it on the interactive interface according to the output format of the preset report template based on the property rights data and the verification results, the method further includes: The knowledge graph is displayed on the interactive interface; If any of the business rule vectors at the node has a risk value greater than a preset risk threshold, an abnormal warning message will be displayed at the node. In response to a click on the node, a prompt window is displayed, showing the rule item corresponding to the business rule vector.

Images