Cross-dataset graph association analysis method and apparatus, and device and medium

Abstract

The present application is applicable to the technical field of big data mining, and particularly relates to a cross-dataset graph association analysis method and apparatus, and a medium. The method comprises: determining matching pairs in external graph data and target graph data; acquiring a first path of any matching pair between a first vertex and a second vertex in the external graph data, and calculating a first ranking score representing the semantic importance of the first path in the external graph data; acquiring a second path in the target graph data that uses the first vertex as a starting point and has a secondary intersection with the first path, and calculating a second ranking score representing the semantic importance of the second path in the target graph data; and filtering the external graph data on the basis of the first ranking score and the second ranking score to obtain filtered external graph data, and performing association rule mining on the basis of the filtered external graph data and the target graph data to obtain a rule mining result. While reducing noise and errors, the present application expands the scope of rule discovery for target graph data, thereby improving the accuracy of graph association analysis.

Description

Cross-dataset graph association analysis method, device, equipment and medium

[0001] This application is based on the Chinese invention application with application number 202410216256.X filed on February 27, 2024, and entitled “Method, device, equipment and medium for graph association analysis across data sets”, and claims priority. Technical Field

[0002] The present application is applicable to the field of big data mining technology, and in particular relates to a method, apparatus, device and medium for cross-dataset graph association analysis. Background Art

[0003] Graph data analysis is an important branch of data science, primarily involving the analysis of nodes and edges in graph data. Within graph data analysis, graph association rule discovery is a technique used to discover associations between graph data items. This technique has been widely applied in multiple fields, such as shopping basket analysis, recommendation systems, and social network research. Therefore, graph association analysis is the analysis of association rules on graph data. Currently, association rule analysis on graph data primarily focuses on graph data from a single data source. However, in actual applications, graph data (such as social networks and knowledge graphs) is typically complex and highly correlated. Due to data incompleteness and limitations, graph data from a single data source often cannot provide sufficient information for a comprehensive and in-depth analysis of association rules. This limits the depth and breadth of graph association analysis, reducing the effectiveness and accuracy of graph association rule discovery.

[0004] Therefore, how to improve the accuracy of graph association analysis has become an urgent problem to be solved.

[0005] Summary of the Invention

[0006] In view of this, embodiments of the present application provide a method, apparatus, device, and medium for cross-dataset graph association analysis to solve the problem of how to improve the accuracy of graph association analysis.

[0007] In a first aspect, an embodiment of the present application provides a graph association analysis method across data sets, the graph association analysis method comprising:

[0008] Perform heterogeneous entity resolution on the external graph data and the target graph data, identify the vertices representing the same entity in the two graph data as matching pairs, merge the attributes and links of the matching pairs, obtain graph connections, and enrich the graph data;

[0009] Obtaining a first path between a first vertex and a second vertex corresponding to any matching pair in the external graph data, and calculating a first ranking score representing the semantic importance of the first path in the external graph data, wherein the second vertex is any vertex in the external graph data except the first vertex;

[0010] Obtaining a second path in the target graph data that has the first vertex as a starting point and has a secondary intersection with the first path, and calculating a second ranking score representing the semantic importance of the second path in the target graph data;

[0011] The external graph data is filtered according to the first ranking score and the second ranking score to obtain filtered external graph data, and association rule mining is performed based on the filtered external graph data and the target graph data to obtain rule mining results.

[0012] In a second aspect, an embodiment of the present application provides a graph association analysis device across data sets, the graph association analysis device comprising:

[0013] A graph connection module is used to perform heterogeneous entity resolution on the external graph data and the target graph data, determine the vertices representing the same entity in the two graph data as matching pairs, merge the attributes and links of the matching pairs, and enrich the graph data after obtaining the graph connection;

[0014] a first calculation module, configured to obtain a first path between a first vertex and a second vertex corresponding to any matching pair in the external graph data, and calculate a first ranking score representing the semantic importance of the first path in the external graph data, wherein the second vertex is any vertex in the external graph data except the first vertex;

[0015] a second calculation module, configured to obtain a second path in the target graph data that has a second intersection with the first path starting from the first vertex, and calculate a second ranking score representing the semantic importance of the second path in the target graph data;

[0016] A graph filtering module is used to filter the external graph data according to the first ranking score and the second ranking score to obtain filtered external graph data, and perform association rule mining based on the filtered external graph data and the target graph data to obtain rule mining results.

[0017] In a third aspect, an embodiment of the present application provides a computer device, comprising a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor implements the graph association analysis method as described in the first aspect when executing the computer program.

[0018] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the graph association analysis method as described in the first aspect is implemented.

[0019] The advantageous effects of the embodiments of the present application compared to the prior art are as follows: the present application determines that vertices representing the same entity in the external graph data and the target graph data are matched pairs, obtains a first path between the first vertex and the second vertex corresponding to any matching pair in the external graph data, calculates a first ranking score representing the semantic importance of the first path in the external graph data, obtains a second path in the target graph data that has a second intersection with the first path starting from the first vertex, calculates a second ranking score representing the semantic importance of the second path in the target graph data, filters the external graph data based on the first ranking score and the second ranking score to obtain filtered external graph data, and performs association rule mining based on the filtered external graph data and the target graph data to obtain rule mining results. The filtering of the external graph data based on the first ranking score and the second ranking score removes irrelevant or misleading data in the external graph, reduces noise and error, and improves the accuracy of graph association analysis. Furthermore, the association rule mining performed on the filtered external graph data and the target graph data reduces noise and error while also expanding the depth and breadth of rule discovery in the target graph data, thereby improving the accuracy of graph association analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0021] FIG1 is a schematic diagram of an application environment of a graph association analysis method across data sets provided in Example 1 of the present application;

[0022] FIG2 is a flow chart of a cross-dataset graph association analysis method provided in Example 2 of the present application;

[0023] FIG3 is a flow chart of a graph association analysis method across data sets provided in Example 3 of the present application;

[0024] FIG4 is a flow chart of a graph association analysis method across data sets provided in Example 4 of the present application;

[0025] FIG5 is a flow chart of a graph association analysis method across data sets provided in Example 5 of the present application;

[0026] FIG6 is a flow chart of a cross-dataset graph association analysis method provided in Example 6 of the present application;

[0027] FIG7 is a flow chart of a cross-dataset graph association analysis method provided in Example 7 of the present application;

[0028] FIG8 is a flow chart of a cross-dataset graph association analysis method provided in Example 8 of the present application;

[0029] FIG9 is a schematic structural diagram of a graph association analysis device across data sets provided in Example 9 of the present application;

[0030] FIG10 is a schematic structural diagram of a computer device provided in Example 10 of the present application. DETAILED DESCRIPTION

[0031] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0032] In order to illustrate the technical solution of the present application, specific embodiments are provided below.

[0033] A cross-dataset graph association analysis method provided in Example 1 of the present application can be applied in an application environment such as Figure 1, wherein a server communicates with a client, the server provides a graph association analysis service, and the client triggers an analysis task to the server. The client includes but is not limited to a PDA, a desktop computer, a laptop computer, an ultra-mobile personal computer (UMPC), a netbook, a cloud computer device, a personal digital assistant (PDA), and other devices. The computer device corresponding to the server can be implemented using an independent server or a server cluster consisting of multiple servers.

[0034] See Figure 2, which is a flow chart of a cross-dataset graph association analysis method provided in Example 2 of this application. The cross-dataset graph association analysis method is applied to the server in Figure 1. The server connects to the client to obtain external graph data and target graph data sent by the client. As shown in Figure 2, the graph association analysis method may include the following steps:

[0035] Step S201 , performing heterogeneous entity resolution on the external graph data and the target graph data, determining that the vertices representing the same entity in the two graph data are matching pairs, merging the attributes and links of the matching pairs, and enriching the graph data after obtaining graph connections.

[0036] In an embodiment of the present application, graph data may refer to a mesh data structure consisting of vertices and edges, wherein vertices may represent entities or objects in the graph data, edges may represent relationships between vertices, vertices and edges may be collectively referred to as elements of the graph, target graph data may refer to graph data for graph association analysis, external graph data may refer to graph data having overlapping elements with the target graph data, and heterogeneous entity resolution may refer to a technology for identifying and matching entities with the same or similar meanings in different data sources or data patterns.

[0037] An entity can refer to a vertex in graph data, an attribute can refer to data associated with a vertex (entity) that characterizes the entity's characteristics or status, a link can refer to an edge in graph data, a matching pair can refer to vertices in the target graph data and external graph data that represent the same entity, and graph-connected enriched graph data can refer to the target graph data enriched with information from the external graph data. For example, if the external graph data comes from a social network and the target graph data comes from an enterprise database, a vertex in the external graph data represents a user, and a vertex in the target graph data represents an employee. If a user is also an employee of the enterprise, then the vertex representing the user and the vertex representing the employee can represent the same entity, that is, the same person. In this case, the vertex representing the user and the vertex representing the employee are a matching pair. The data associated with the name, age, and gender corresponding to the two vertices are the attributes of the matching pair, and the edge corresponding to the two vertices is the link of the matching pair.

[0038] In the process of enriching graph data after obtaining graph connection, first, entity recognition is performed on the target graph data and external graph data, and the features and attributes of the entities in the two graph data are extracted respectively. Then, based on the features and attributes of the entities in the two graph data, feature vectors of each entity are created respectively, and the similarity between the entities in the two graph data is calculated by a similarity algorithm. According to the similarity calculation results, matching pairs representing the same entity in the two graph data are determined. Finally, for the matching vertices in the external graph data, these vertices and the attributes and links corresponding to the vertices are merged into the target graph data to obtain the enriched graph data after graph connection.

[0039] Step S202: Obtain a first path between the first vertex and the second vertex corresponding to any matching pair in the external graph data, and calculate a first ranking score representing the semantic importance of the first path in the external graph data.

[0040] In an embodiment of the present application, the first vertex may refer to a matching vertex in the external graph data, the second vertex may refer to any vertex in the external graph data except the first vertex, the first path may refer to a path composed of vertices and edges between the first vertex and the second vertex, the semantic importance may refer to the degree of importance of the first path in the external graph data, and the first ranking score may refer to a score that characterizes the semantic importance of the first path in the external graph data.

[0041] In the process of calculating the first ranking score of the first path, first, based on the matching pairs in the two graph data determined in step S201, the first vertex matched by the matching pair in the external graph data is determined, and then, based on the vertices and edges between the first vertex and the second vertex, the first path from the first vertex as the starting point to the second vertex is determined. Finally, the first path can be processed by a model that has been trained for semantic importance to determine the first ranking score of the first path in the external graph data.

[0042] Step S203 : obtaining a second path in the target graph data that has a second intersection with the first path starting from the first vertex, and calculating a second ranking score representing the semantic importance of the second path in the target graph data.

[0043] In an embodiment of the present application, the second path may refer to a path in the target graph data that has a second intersection with the first path starting from the first vertex, and the second ranking score may refer to a score that represents the semantic importance of the second path in the target graph data.

[0044] In the process of calculating the second ranking score of the second path, first, according to the matching pairs in the two graph data determined in step S201, in the target graph data, the vertex that matches the first vertex in the first path is determined, and the vertex is the first vertex in the target graph data. Then, for the first vertex that matches in the target graph data, the path between the first vertex and any vertex other than the vertex is determined. If the path contains vertices that are matching pairs with the vertices matched in the first path, the path between the first vertex and the vertices that are matching pairs with the vertices matched in the first path is determined as the second path. Finally, the second path can be processed by a model that has been trained for semantic importance to determine the second ranking score of the second path in the target graph data.

[0045] Step S204: Filter the external graph data according to the first ranking score and the second ranking score to obtain filtered external graph data, and perform association rule mining based on the filtered external graph data and the target graph data to obtain rule mining results.

[0046] In an embodiment of the present application, filtered external graph data may refer to external graph data after removing data that is irrelevant or misleading to the target graph data, association rule mining may refer to a technology used to discover the association between elements in graph data, and rule mining results may refer to the results of association rule mining.

[0047] In the process of obtaining rule mining results, first, for any first path, according to the second ranking scores of all second paths that have a secondary intersection with the first path determined in step S203, a third ranking score representing the semantic relevance of the first path and the target graph data is obtained, and according to the first ranking score and the third ranking score of the first path, a dual-path ranking score of the first path is calculated, and then, according to the dual-path ranking scores of all first paths in the external graph data, the external graph data is filtered to remove data that is irrelevant or misleading to the target graph data to obtain filtered external graph data, and finally, the filtered external graph data is physically merged into the target graph data to obtain enriched graph data after graph filtering, and association rule mining is performed on the enriched graph data after graph filtering through graph data analysis technology to obtain rule mining results.

[0048] The embodiment of the present application determines that the vertices representing the same entity in the external graph data and the target graph data are matched pairs, obtains the first path between the first vertex and the second vertex corresponding to any matching pair in the external graph data, calculates a first ranking score representing the semantic importance of the first path in the external graph data, obtains a second path in the target graph data that has a second intersection with the first path starting from the first vertex, calculates a second ranking score representing the semantic importance of the second path in the target graph data, filters the external graph data according to the first ranking score and the second ranking score to obtain filtered external graph data, performs association rule mining based on the filtered external graph data and the target graph data, and obtains rule mining results. Among them, filtering the external graph data by the first ranking score and the second ranking score removes irrelevant or misleading data in the external graph, reduces noise and errors, and improves the accuracy of graph association analysis. Moreover, performing association rule mining by the filtered external graph data and the target graph data, due to the reference and integration of the filtered external graph data, reduces noise and errors while also expanding the depth and breadth of rule discovery in the target graph data, thereby improving the accuracy of graph association analysis.

[0049] See Figure 3, which is a flow chart of a cross-dataset graph association analysis method provided in Example 3 of the present application. As shown in Figure 3, the calculation of the first ranking score representing the semantic importance of the first path in the external graph data in the above step S202 may include the following steps:

[0050] Step S301: Obtain the out-degree value of each vertex in the first path.

[0051] Step S302: For any vertex in the first path, perform a reciprocal operation on the out-degree value of the vertex to obtain the inverted out-degree value of the vertex.

[0052] Step S303 : performing a multiplication operation on the inverted out-degree values ​​of all vertices in the first path to obtain a first ranking score representing the semantic importance of the first path in the external graph data.

[0053] In the embodiment of the present application, the out-degree value may refer to the number of edges starting from a vertex.

[0054] For example, if the first path can be expressed as ρ1=(v0,v1,···,v m ), where ρ1 is the first path, v0 is the first vertex, and v m is the second vertex, (v0,v1,···,v m ) is the set of all vertices in the first path, then the calculation formula for the first ranking score of the first path can be: Among them, D(v i ) is the node v i The out-degree value, R importance (ρ1) is the first ranking score of the first path. For example, if there are 4 vertices in the first path, namely v0, v1, v2 and v3, and the out-degree values ​​of these 4 vertices are 1, 2, 2 and 1 respectively, then according to the calculation formula of the first ranking score, the out-degree values ​​of these 4 vertices are inversely calculated, and the inverse out-degree values ​​are 1, 2, 2 and 1 respectively. and 1, then perform a multiplication operation on the out-degree value after inversion, that is, It can be obtained that the first ranking score of the first path is 0.25.

[0055] In an embodiment of the present application, the out-degree value of each vertex in the first path is inverted to obtain the inverted out-degree value of the vertex, and the inverted out-degree values ​​of all vertices in the first path are multiplied to obtain a first ranking score representing the semantic importance of the first path in the external graph data. Specifically, by calculating the first ranking score of the first path based on the out-degree value of each vertex in the first path, the contribution of each vertex in the first path to the semantic importance is taken into account, thereby improving the comprehensiveness and integrity of calculating the semantic importance of the first path, thereby further improving the accuracy of calculating the first ranking score.

[0056] See Figure 4, which is a flow chart of a cross-dataset graph association analysis method provided in Example 4 of the present application. As shown in Figure 4, obtaining a second path in the target graph data starting from the first vertex and having a secondary intersection with the first path in the above step S203 may include the following steps:

[0057] Step S401: Obtain secondary intersection points where the third vertices corresponding to all matching pairs in the target graph data intersect with the first path.

[0058] Step S402: obtaining a second path starting from the first vertex and having a secondary intersection with the first path based on the vertices and links between the first vertex and the secondary intersection in the target graph data.

[0059] In an embodiment of the present application, the third vertex may refer to a matching vertex in the target graph data, and the secondary intersection point may refer to a vertex that is a matching pair with the matching vertex in the first path in addition to the first vertex in the target graph data.

[0060] In the process of obtaining the second path, first, according to the matching pairs in the two graph data determined in step S201, the third vertex matched by the matching pair in the target graph data except the first vertex is determined, and the vertex matched by the matching pair in the first path is determined. Then, according to the two parts of the matched vertices determined in the target graph data and the first path, the vertices among all the third vertices of the target graph data that are matching pairs with the vertices matched in the first path are determined. The determined vertex is the secondary intersection point. Finally, in the target graph data, according to all the vertices and edges between the first vertex and the secondary intersection point, it is determined that the path between the first vertex and the secondary intersection point is the second path.

[0061] In an embodiment of the present application, by obtaining the secondary intersection points where the third vertices corresponding to all matching pairs in the target graph data intersect with the first path, a second path starting from the first vertex and having a secondary intersection with the first path is obtained based on the vertices and links between the first vertex and the secondary intersection points in the target graph data. Among them, the secondary intersection points in the target graph data that secondary intersect with the first path except the first vertex are determined through the vertices matched in the target graph data and the vertices matched in the first path, and based on the secondary intersection points, a second path that has a deeper association with the first path is determined, thereby improving the reliability of analyzing the correlation between the first path and the target graph data.

[0062] 5 , which is a flow chart of a cross-dataset graph association analysis method provided in Example 5 of the present application. As shown in FIG5 , calculating the second ranking score representing the semantic importance of the second path in the target graph data in step S203 may include the following steps:

[0063] Step S501: Obtain the out-degree value of each vertex in the second path.

[0064] Step S502: For any vertex in the second path, perform a reciprocal operation on the out-degree value of the vertex to obtain the inverted out-degree value of the vertex.

[0065] Step S503 : performing a multiplication operation on the inverted out-degree values ​​of all vertices in the second path to obtain a second ranking score representing the semantic importance of the second path in the target graph data.

[0066] For example, if the first path can be expressed as ρ1=(v0,v1,···,v m ), where ρ1 is the first path, v0 is the first vertex, and v m is the second vertex, (v0,v1,···,v m ) is the set of all vertices in the first path, then the second path can be expressed as ρ2=(u0,u1,···,u n ), where ρ2 is the second path, u0 and v0 are a matching pair, i.e. u0 is the first vertex in the target graph data, u n is the secondary intersection point, (u0,u1,···,u n ) is the set of all vertices in the second path, then the calculation formula for the second ranking score of the second path can be: Among them, D(u i ) is node u i The out-degree value, R importance (ρ2) is the second ranking score of the second path. For the specific process of calculating the second ranking score of the second path, reference may be made to the contents in steps S301 to S303.

[0067] In an embodiment of the present application, the out-degree value of each vertex in the second path is inverted to obtain the inverted out-degree value of the vertex, and the inverted out-degree values ​​of all vertices in the second path are multiplied to obtain a second ranking score representing the semantic importance of the second path in the target graph data. Specifically, by calculating the second ranking score of the second path based on the out-degree value of each vertex in the second path, the contribution of each vertex in the second path to the semantic importance is taken into account, thereby improving the comprehensiveness and integrity of calculating the semantic importance of the second path, thereby further improving the accuracy of calculating the second ranking score.

[0068] See Figure 6, which is a flow chart of a cross-dataset graph association analysis method provided in Example 6 of the present application. As shown in Figure 6, in step S204, filtering the external graph data according to the first ranking score and the second ranking score to obtain the filtered external graph data may include the following steps:

[0069] Step S601 : In the target graph data, a third ranking score representing the semantic relevance between the first path and the target graph data is calculated based on the second ranking scores of all second paths starting from the first vertex and having a secondary intersection with the first path.

[0070] Step S602: Add the first ranking score and the third ranking score to obtain a dual-path ranking score of the first path.

[0071] Step S603 : Aggregate the dual-path ranking scores of all first paths between the first vertex and the second vertex to obtain a total score of the path set between the first vertex and the second vertex.

[0072] In an embodiment of the present application, semantic relevance may refer to the degree of relevance between the first path and the target graph data, the third ranking score may refer to the score characterizing the semantic relevance between the first path and the target graph data, the dual-path ranking score may refer to a comprehensive score characterizing the semantic importance of the first path in the external graph data and the semantic relevance between the first path and the target graph data, the path set may refer to the set of all paths between the first vertex and the second vertex in the external graph data, and the total score of the path set may refer to a comprehensive score characterizing the semantic importance of the path set between the first vertex and the second vertex and the semantic relevance to the target graph data.

[0073] In the process of obtaining the total score of the path set between the first vertex and the second vertex, first, based on the second ranking score of the second path calculated in step S203, the second ranking scores of all second paths that have a secondary intersection with the first path are determined, and based on the second ranking scores of all the determined second paths, a third ranking score of the semantic relevance of the first path to the target graph data is determined. Secondly, the first ranking score representing the semantic importance of the first path in the external graph data calculated in steps S301 to S303 is added to the third ranking score of the first path to obtain a dual-path ranking score of the first path. Then, the first ranking scores of all first paths between the first vertex and the second vertex are respectively calculated according to steps S301 to S303, the third ranking scores of all first paths between the first vertex and the second vertex are respectively calculated according to step S601, and the dual-path ranking scores of all first paths between the first vertex and the second vertex are calculated based on the calculated first ranking scores and third ranking scores. Finally, the calculated dual-path ranking scores of all first paths between the first vertex and the second vertex are summed to obtain the total score of the path set between the first vertex and the second vertex.

[0074] Step S604 : filtering the external graph data according to the total scores of all path sets in the external graph data and a preset total score threshold to obtain filtered external graph data.

[0075] In the embodiment of the present application, the preset total score threshold may refer to a preset total score value of a path set.

[0076] In the process of obtaining the filtered external graph data, first, according to steps S601 to S603, the total score of the path set between the first vertex and the second vertex corresponding to all matching pairs in the external graph data can be calculated, and then, for the total score of the path set between the first vertex and the second vertex corresponding to any matching pair, the calculated total score of the path set is compared with the preset total score threshold. If the total score of the path set is greater than the preset total score threshold, all vertices and edges involved in the path set are retained. Finally, the elements other than the vertices and edges of all retained path sets in the external graph data are filtered to obtain the filtered external graph data.

[0077] In an embodiment of the present application, a third ranking score representing the semantic relevance between the first path and the target graph data is calculated based on the second ranking scores of all second paths that have a second intersection with the first path starting from the first vertex in the target graph data. The first ranking score and the third ranking score are summed to obtain a dual-path ranking score for the first path. The dual-path ranking scores of all first paths between the first vertex and the second vertex are aggregated to obtain a total score for the path set between the first vertex and the second vertex. The external graph data is filtered based on the total score of all path sets in the external graph data and a preset total score threshold to obtain filtered external graph data. The dual-path ranking score of the first path is determined based on the calculated first ranking score and the second ranking score. This comprehensively considers the semantic importance of the first path in the external graph data and the degree of semantic relevance between the first path and the target graph data, so that the calculated total score of the path set better reflects the actual degree of relevance between the path set and the target graph data. The external graph data is filtered based on the total score of all path sets in the external graph, effectively removing data in the external graph data that is irrelevant or misleading to the target graph data, reducing noise and error, and improving the quality and accuracy of the integration of the two graph data.

[0078] See Figure 7, which is a flowchart illustrating a cross-dataset graph association analysis method provided in Example 7 of the present application. As shown in Figure 7, the calculation of a third ranking score representing the semantic relevance between the first path and the target graph data based on the second ranking scores of all second paths starting from the first vertex that have a secondary intersection with the first path in step S601 may include the following steps:

[0079] Step S701: In the target graph data, for any secondary intersection, if the number of second paths from the first vertex as the starting point to the secondary intersection is greater than a preset path value, the second ranking scores of all second paths from the first vertex as the starting point to the secondary intersection are maximized to obtain a second comprehensive ranking score of the second path from the first vertex as the starting point to the secondary intersection.

[0080] In an embodiment of the present application, the preset path value may refer to a preset value of the number of second paths involved in any secondary intersection, wherein the preset path value may be 1, and the second comprehensive ranking score may refer to a comprehensive score of the semantic importance of all second paths involved in any secondary intersection whose number of second paths involved is greater than the preset path value.

[0081] In the process of obtaining the second comprehensive ranking score, first, in the target graph data, for any secondary intersection, based on the vertices and corresponding edges between the first vertex and the secondary intersection, the second path between the first vertex and the secondary intersection is determined, and based on the determined second path, the number of second paths between the first vertex and the secondary intersection is determined. Then, the number of second paths involved in each secondary intersection is compared with a preset path value. Finally, for a secondary intersection where the number of second paths involved is greater than the preset path value, the second ranking scores of all second paths involved in the secondary intersection are obtained, and the maximum value of the second ranking scores of all the obtained second paths is taken as the second comprehensive ranking score of the second path between the first vertex and the secondary intersection.

[0082] Step S702 , calculating a third ranking score representing the semantic relevance between the first path and the target graph data based on the second ranking scores and the second comprehensive ranking scores of all second paths that intersect with the first path starting from the first vertex.

[0083] In the process of calculating the third ranking score, first, for the secondary intersection points whose number of second paths involved is greater than the preset path value, the second comprehensive ranking scores involved in these secondary intersection points are determined, and for the secondary intersection points whose number of second paths involved is less than or equal to the preset path value, the second ranking scores involved in these secondary intersection points are determined. Then, the second comprehensive ranking scores corresponding to the secondary intersection points whose number of second paths is greater than the preset path value and the second ranking scores corresponding to the secondary intersection points whose number of second paths is less than or equal to the preset path value are added to obtain the third ranking score that characterizes the semantic relevance between the first path and the target graph data.

[0084] For example, if the preset path value is 1, if there are two secondary intersection points with the first path, namely u1 and u2, where the item value of the second path involved in u1 is 2, namely ρ1 and ρ2, the second ranking score of ρ1 is 0.1, the second ranking score of ρ2 is 0.2, the item value of the second path involved in u2 is 1, the path is ρ3, and the second ranking score of ρ3 is 0.3, then the process of calculating the third ranking score representing the semantic relevance between the first path and the target graph data can be: 1) If the item value 2 of the second path involved in u1 is greater than the preset path value 1, then for u1, take the maximum value of 0.1 and 0.2, and obtain the second comprehensive ranking score involved in u1 as 0.2; 2) add the second comprehensive ranking score of 0.2 involved in u1 and the second ranking score of 0.3 involved in u2 to obtain a third ranking score of 0.5.

[0085] In an embodiment of the present application, for any secondary intersection in the target graph data, if the number of second paths from the first vertex to the secondary intersection is greater than a preset path value, the second ranking scores of all second paths from the first vertex to the secondary intersection are maximized to obtain a second comprehensive ranking score of the second path from the first vertex to the secondary intersection. A third ranking score representing the semantic relevance between the first path and the target graph data is calculated based on the second ranking scores and the second comprehensive ranking scores of all second paths that intersect with the first path from the first vertex. The third ranking score representing the semantic relevance between the first path and the target graph data is calculated based on the second ranking scores of the second paths involved in each secondary intersection in the target graph data. The third ranking score representing the semantic relevance between the first path and the target graph data is calculated, taking into account the contribution of all second paths from the first vertex to the secondary intersection to the semantic relevance between the first path and the target graph data, thereby improving the comprehensiveness and integrity of the calculation of the semantic relevance between the first path and the target graph data, thereby further improving the accuracy of the calculation of the third ranking score.

[0086] See Figure 8, which is a flow chart of a cross-dataset graph association analysis method provided in Example 8 of the present application. As shown in Figure 8, in the above step S204, association rule mining is performed based on the filtered external graph data and the target graph data to obtain rule mining results, which may include the following steps:

[0087] Step S801 : Physically merge the filtered external graph data with the target graph data to obtain enriched graph data after graph filtering.

[0088] Step S802 , performing association rule mining on the enriched graph data after graph filtering to obtain rule mining results.

[0089] In the embodiment of the present application, the enriched graph data after graph filtering may refer to the graph data obtained by merging the filtered external graph data and the target graph data.

[0090] In the process of obtaining rule mining results, first, according to the obtained matching pairs, the matching vertices and vertex-related attributes and links in the filtered external graph data are physically merged into the matching vertices and vertex-related attributes and links in the target graph data to obtain the enriched graph data after graph filtering. Then, association rule mining can be performed on the enriched graph data after graph filtering through graph data mining technology to obtain rule mining results.

[0091] For example, through the graph data mining technology of cascade traversal and pattern matching, association rule mining is performed on the enriched graph data after graph filtering. First, the graph data is enriched after graph filtering through depth-first traversal or breadth-first traversal, and the attributes of the nodes in the graph data and the corresponding edge relationships are recorded. Candidate association rules are generated based on the recorded node attributes and the corresponding edge relationships. Then, according to the predefined association rule pattern, the pattern matching algorithm is used to search for candidate rules that match the predefined association rule pattern in the candidate association rules. Finally, according to predefined metrics, such as confidence and support, the matching candidate rules are filtered and sorted to obtain rule mining results.

[0092] For example, if the enriched graph data after graph filtering is a user purchase behavior graph, where vertices represent items purchased by users, edges represent simultaneous purchase relationships, and the predefined association rule pattern is that users who purchased item A usually also purchased item B, then association rule mining can be performed on this enriched graph data after graph filtering. First, each vertex and corresponding edge in the graph data can be cascaded to record each user's purchase path, that is, the sequence of purchased items. Candidate association rules are then generated based on the purchase path. For example, the candidate association rule could be that users who purchased item A also purchased items B and C; users who purchased item B also purchased item D; users who purchased item A also purchased item C, etc. Then, a pattern matching algorithm is used to search among the candidate association rules for candidate rules that match the predefined association rule pattern. Finally, the matching candidate rules are filtered and sorted according to predefined metrics such as confidence and support to obtain the rule mining results.

[0093] In an embodiment of the present application, the filtered external graph data is physically merged with the target graph data to obtain enriched graph data after graph filtering, and association rule mining is performed on the enriched graph data after graph filtering to obtain rule mining results. Specifically, by physically merging the filtered external graph data with the target graph data, the enriched graph data after graph filtering enriches the data of the external graph data, improves the data integrity of the enriched graph data after graph filtering, and also removes irrelevant or misleading information in the external graph data, thereby improving the accuracy of association rule mining on the enriched graph data after graph filtering.

[0094] Corresponding to the cross-dataset graph association analysis method described in the preceding embodiment, FIG9 shows a block diagram of a cross-dataset graph association analysis apparatus provided in Embodiment 9 of the present application. This cross-dataset graph association analysis apparatus is applied to the server shown in FIG1 . The server connects to the client to obtain external graph data and target graph data sent by the client. For ease of illustration, only the portion relevant to the present embodiment is shown.

[0095] Referring to FIG9 , the cross-dataset graph association analysis apparatus includes:

[0096] A graph connection module 91 is used to perform heterogeneous entity resolution on the external graph data and the target graph data, determine vertices representing the same entity in the two graph data as matching pairs, merge the attributes and links of the matching pairs, and obtain enriched graph data after graph connection;

[0097] a first calculation module 92 configured to obtain a first path between a first vertex and a second vertex corresponding to any matching pair in the external graph data, and calculate a first ranking score representing the semantic importance of the first path in the external graph data, wherein the second vertex is any vertex in the external graph data other than the first vertex;

[0098] A second calculation module 93 is configured to obtain a second path in the target graph data that has a second intersection with the first path starting from the first vertex, and calculate a second ranking score representing the semantic importance of the second path in the target graph data;

[0099] The graph filtering module 94 is used to filter the external graph data according to the first ranking score and the second ranking score to obtain filtered external graph data, and perform association rule mining based on the filtered external graph data and the target graph data to obtain rule mining results.

[0100] Optionally, the first calculation module 92 includes:

[0101] A first acquiring unit, configured to acquire the out-degree value of each vertex in the first path;

[0102] a first inverse unit, configured to perform an inverse operation on the out-degree value of any vertex in the first path to obtain the inverse out-degree value of the vertex;

[0103] The first multiplication unit is used to perform a multiplication operation on the inverted out-degree values ​​of all vertices in the first path to obtain a first ranking score representing the semantic importance of the first path in the external graph data.

[0104] Optionally, the second calculation module 93 includes:

[0105] A second acquisition unit is used to acquire secondary intersection points where the third vertices corresponding to all matching pairs in the target graph data intersect with the first path;

[0106] The third acquisition unit is used to obtain a second path starting from the first vertex and having a secondary intersection with the first path based on the vertices and links between the first vertex and the secondary intersection in the target graph data.

[0107] Optionally, the second calculation module 93 includes:

[0108] a fourth acquiring unit, configured to acquire an out-degree value of each vertex in the second path;

[0109] A second inverting unit is configured to perform an inverting operation on the out-degree value of any vertex in the second path to obtain the inverted out-degree value of the vertex;

[0110] The second multiplication unit is used to perform a multiplication operation on the inverted out-degree values ​​of all vertices in the second path to obtain a second ranking score representing the semantic importance of the second path in the target graph data.

[0111] Optionally, the graph filtering module 94 includes:

[0112] a third calculation unit configured to calculate, in the target graph data, a third ranking score representing the semantic relevance between the first path and the target graph data based on the second ranking scores of all second paths starting from the first vertex and having secondary intersections with the first path;

[0113] a fourth calculation unit, configured to add the first ranking score and the third ranking score to obtain a dual-path ranking score of the first path;

[0114] a fifth calculation unit, configured to aggregate the dual-path ranking scores of all first paths between the first vertex and the second vertex to obtain a total score of the path set between the first vertex and the second vertex;

[0115] The filtering unit is configured to filter the external graph data according to the total scores of all path sets in the external graph data and a preset total score threshold to obtain filtered external graph data.

[0116] Optionally, the third computing unit includes:

[0117] a maximum subunit configured to, for any secondary intersection in the target graph data, if the number of second paths from the first vertex as a starting point to the secondary intersection is greater than a preset path value, perform a maximum operation on the second ranking scores of all second paths from the first vertex as a starting point to the secondary intersection to obtain a second comprehensive ranking score of the second paths from the first vertex as a starting point to the secondary intersection;

[0118] The sixth computing subunit is configured to calculate a third ranking score representing the semantic relevance between the first path and the target graph data based on the second ranking scores and the second comprehensive ranking scores of all second paths that intersect with the first path starting from the first vertex.

[0119] Optionally, the graph filtering module 94 includes:

[0120] a merging unit, configured to physically merge the filtered external graph data with the target graph data to obtain enriched graph data after graph filtering;

[0121] The mining unit is used to perform association rule mining on the enriched graph data after the graph is filtered to obtain a rule mining result.

[0122] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as shown in FIG10 . The computer device includes a processor, a memory, a network interface, and a database connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a readable storage medium, and a database. The internal memory provides an environment for the operation of the operating system and the readable storage medium in the non-volatile storage medium. The database of the computer device is used to store user original data. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the readable storage medium is executed by the processor, a graph association analysis method across data sets is implemented.

[0123] In one embodiment, a computer device is provided, comprising a memory, a processor, and a readable storage medium stored in the memory and executable on the processor. When the processor executes the readable storage medium, the steps of the cross-dataset graph association analysis method in the above-described embodiment are implemented, such as steps S201-S204 shown in FIG2 , or the steps shown in FIG3 to FIG8 . To avoid repetition, these steps are not described here. Alternatively, when the processor executes the readable storage medium, the functions of the modules / units in the embodiment of the user data processing device are implemented, such as the functions of the graph connection module 91, the first calculation module 92, the second calculation module 93, and the graph filtering module 94 shown in FIG9 . To avoid repetition, these steps are not described here.

[0124] In one embodiment, one or more readable storage media storing computer-readable instructions are provided, and the computer-readable storage media stores computer-readable instructions. When the computer-readable instructions are executed by one or more processors, the one or more processors implement the steps of the graph association analysis method across data sets in the above-mentioned embodiment, such as steps S201-S204 shown in Figure 2, or the steps shown in Figures 3 to 8. To avoid repetition, they are not described here. Alternatively, when the processor executes the readable storage medium, the functions of each module / unit in this embodiment of the user data processing device are implemented, such as the functions of the graph connection module 91, the first calculation module 92, the second calculation module 93 and the graph filtering module 94 shown in Figure 9. To avoid repetition, they are not described here. The readable storage medium in this embodiment includes a non-volatile readable storage medium and a volatile readable storage medium.

[0125] Those skilled in the art will understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing related hardware through a readable storage medium, and the readable storage medium can be stored in a non-volatile computer-readable storage medium. When the readable storage medium is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct RAM bus dynamic RAM (DRDRAM), and RAM bus dynamic RAM (RDRAM).

[0126] Those skilled in the art will clearly understand that for the sake of convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0127] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present application, and should all be included in the scope of protection of the present application.

Claims

1. A graph association analysis method across data sets, characterized in that: The graph association analysis method comprises: Perform heterogeneous entity resolution on the external graph data and the target graph data, determine the vertices representing the same entity in the two graph data as matching pairs, merge the attributes and links of the matching pairs, obtain graph connections, and enrich the graph data; Obtaining a first path between a first vertex and a second vertex corresponding to any matching pair in the external graph data, and calculating a first ranking score representing the semantic importance of the first path in the external graph data, wherein the second vertex is any vertex in the external graph data except the first vertex; Obtaining a second path in the target graph data that has the first vertex as a starting point and has a secondary intersection with the first path, and calculating a second ranking score representing the semantic importance of the second path in the target graph data; The external graph data is filtered according to the first ranking score and the second ranking score to obtain filtered external graph data, and association rule mining is performed based on the filtered external graph data and the target graph data to obtain rule mining results.

2. The cross-dataset graph association analysis method according to claim 1, characterized in that: The calculating a first ranking score representing the semantic importance of the first path in the external graph data includes: Obtaining the out-degree value of each vertex in the first path; For any vertex in the first path, performing a reciprocal operation on the out-degree value of the vertex to obtain the inverted out-degree value of the vertex; The inverted out-degree values ​​of all vertices in the first path are multiplied together to obtain a first ranking score representing the semantic importance of the first path in the external graph data.

3. The cross-dataset graph association analysis method according to claim 1, characterized in that: The acquiring, in the target graph data, a second path starting from the first vertex and having a secondary intersection with the first path, includes: Obtaining secondary intersection points where the third vertices corresponding to all matching pairs in the target graph data intersect with the first path; According to the vertices and links between the first vertex as the starting point and the second intersection point in the target graph data, a second path with the first vertex as the starting point and having a second intersection with the first path is obtained.

4. The cross-dataset graph association analysis method according to claim 1, characterized in that: The calculating a second ranking score representing the semantic importance of the second path in the target graph data includes: Obtaining the out-degree value of each vertex in the second path; For any vertex in the second path, perform a reciprocal operation on the out-degree value of the vertex to obtain the inverted out-degree value of the vertex; The inverted out-degree values ​​of all vertices in the second path are multiplied together to obtain a second ranking score representing the semantic importance of the second path in the target graph data.

5. The cross-dataset graph association analysis method according to claim 3, characterized in that: The filtering of the external graph data according to the first ranking score and the second ranking score to obtain filtered external graph data includes: In the target graph data, calculating a third ranking score representing the semantic relevance between the first path and the target graph data based on the second ranking scores of all second paths starting from the first vertex and having a secondary intersection with the first path; Adding the first ranking score and the third ranking score to obtain a dual-path ranking score of the first path; Aggregating the dual-path ranking scores of all first paths between the first vertex and the second vertex to obtain a total score of the path set between the first vertex and the second vertex; The external graph data is filtered according to the total scores of all path sets in the external graph data and a preset total score threshold to obtain filtered external graph data.

6. The cross-dataset graph association analysis method according to claim 5, characterized in that: The calculating, based on the second ranking scores of all second paths that have a second intersection with the first path and start from the first vertex, a third ranking score representing the semantic relevance between the first path and the target graph data includes: In the target graph data, for any secondary intersection, if the number of second paths from the first vertex as a starting point to the secondary intersection is greater than a preset path value, a maximum value is taken for the second ranking scores of all second paths from the first vertex as a starting point to the secondary intersection to obtain a second comprehensive ranking score of the second paths from the first vertex as a starting point to the secondary intersection; A third ranking score representing the semantic relevance between the first path and the target graph data is calculated based on the second ranking scores and the second comprehensive ranking scores of all second paths that have the first vertex as a starting point and intersect with the first path.

7. The cross-dataset graph association analysis method according to claim 1, characterized in that: The performing association rule mining based on the filtered external graph data and the target graph data to obtain rule mining results includes: Physically merging the filtered external graph data with the target graph data to obtain enriched graph data after graph filtering; Association rule mining is performed on the enriched graph data after filtering the graph to obtain a rule mining result.

8. A graph association analysis device across data sets, characterized in that: The graph association analysis device comprises: A graph connection module is used to perform heterogeneous entity resolution on the external graph data and the target graph data, identify vertices representing the same entity in the two graph data as matching pairs, merge the attributes and links of the matching pairs, and enrich the graph data after obtaining graph connections; a first calculation module, configured to obtain a first path between a first vertex and a second vertex corresponding to any matching pair in the external graph data, and calculate a first ranking score representing the semantic importance of the first path in the external graph data, wherein the second vertex is any vertex in the external graph data except the first vertex; a second calculation module, configured to obtain a second path in the target graph data that has a second intersection with the first path starting from the first vertex, and calculate a second ranking score representing the semantic importance of the second path in the target graph data; A graph filtering module is used to filter the external graph data according to the first ranking score and the second ranking score to obtain filtered external graph data, and perform association rule mining based on the filtered external graph data and the target graph data to obtain rule mining results.

9. A computer device comprising a memory, a processor, and a readable storage medium stored in the memory and operable on the processor, wherein: When the processor executes the readable storage medium, the following steps are implemented: Perform heterogeneous entity resolution on the external graph data and the target graph data, determine the vertices representing the same entity in the two graph data as matching pairs, merge the attributes and links of the matching pairs, obtain graph connections, and enrich the graph data; Obtaining a first path between a first vertex and a second vertex corresponding to any matching pair in the external graph data, and calculating a first ranking score representing the semantic importance of the first path in the external graph data, wherein the second vertex is any vertex in the external graph data except the first vertex; Obtaining a second path in the target graph data that has the first vertex as a starting point and has a secondary intersection with the first path, and calculating a second ranking score representing the semantic importance of the second path in the target graph data; The external graph data is filtered according to the first ranking score and the second ranking score to obtain filtered external graph data, and association rule mining is performed based on the filtered external graph data and the target graph data to obtain rule mining results.

10. The computer device according to claim 9, wherein: The calculating a first ranking score representing the semantic importance of the first path in the external graph data includes: Obtaining the out-degree value of each vertex in the first path; For any vertex in the first path, performing a reciprocal operation on the out-degree value of the vertex to obtain the inverted out-degree value of the vertex; The inverted out-degree values ​​of all vertices in the first path are multiplied together to obtain a first ranking score representing the semantic importance of the first path in the external graph data.

11. The computer device according to claim 9, wherein: The acquiring, in the target graph data, a second path starting from the first vertex and having a secondary intersection with the first path, includes: Obtaining secondary intersection points where the third vertices corresponding to all matching pairs in the target graph data intersect with the first path; According to the vertices and links between the first vertex as the starting point and the second intersection point in the target graph data, a second path with the first vertex as the starting point and having a second intersection with the first path is obtained.

12. The computer device according to claim 9, wherein: The calculating a second ranking score representing the semantic importance of the second path in the target graph data includes: Obtaining the out-degree value of each vertex in the second path; For any vertex in the second path, perform a reciprocal operation on the out-degree value of the vertex to obtain the inverted out-degree value of the vertex; The inverted out-degree values ​​of all vertices in the second path are multiplied together to obtain a second ranking score representing the semantic importance of the second path in the target graph data.

13. The computer device according to claim 11, wherein: The filtering of the external graph data according to the first ranking score and the second ranking score to obtain filtered external graph data includes: In the target graph data, calculating a third ranking score representing the semantic relevance between the first path and the target graph data based on the second ranking scores of all second paths starting from the first vertex and having a secondary intersection with the first path; Adding the first ranking score and the third ranking score to obtain a dual-path ranking score of the first path; Aggregating the dual-path ranking scores of all first paths between the first vertex and the second vertex to obtain a total score of the path set between the first vertex and the second vertex; The external graph data is filtered according to the total scores of all path sets in the external graph data and a preset total score threshold to obtain filtered external graph data.

14. The computer device according to claim 13, wherein: The calculating, based on the second ranking scores of all second paths that have a second intersection with the first path and start from the first vertex, a third ranking score representing the semantic relevance between the first path and the target graph data includes: In the target graph data, for any secondary intersection, if the number of second paths from the first vertex as a starting point to the secondary intersection is greater than a preset path value, a maximum value is taken for the second ranking scores of all second paths from the first vertex as a starting point to the secondary intersection to obtain a second comprehensive ranking score of the second paths from the first vertex as a starting point to the secondary intersection; A third ranking score representing the semantic relevance between the first path and the target graph data is calculated based on the second ranking scores and the second comprehensive ranking scores of all second paths that have the first vertex as a starting point and intersect with the first path.

15. The computer device according to claim 9, wherein: The performing association rule mining based on the filtered external graph data and the target graph data to obtain rule mining results includes: Physically merging the filtered external graph data with the target graph data to obtain enriched graph data after graph filtering; Association rule mining is performed on the enriched graph data after filtering the graph to obtain a rule mining result.

16. One or more computer-readable storage media storing computer-readable instructions, wherein: When the computer readable instructions are executed by one or more processors, the one or more processors are caused to perform the following steps: Perform heterogeneous entity resolution on external graph data and target graph data to determine the top entities representing the same entity in the two graph data. The points are matched pairs, and the attributes and links of the matched pairs are merged to obtain graph connections and enrich the graph data; Obtaining a first path between a first vertex and a second vertex corresponding to any matching pair in the external graph data, and calculating a first ranking score representing the semantic importance of the first path in the external graph data, wherein the second vertex is any vertex in the external graph data except the first vertex; Obtaining a second path in the target graph data that has the first vertex as a starting point and has a secondary intersection with the first path, and calculating a second ranking score representing the semantic importance of the second path in the target graph data; The external graph data is filtered according to the first ranking score and the second ranking score to obtain filtered external graph data, and association rule mining is performed based on the filtered external graph data and the target graph data to obtain rule mining results.

17. The readable storage medium according to claim 16, wherein: The calculating a first ranking score representing the semantic importance of the first path in the external graph data includes: Obtaining the out-degree value of each vertex in the first path; For any vertex in the first path, performing a reciprocal operation on the out-degree value of the vertex to obtain the inverted out-degree value of the vertex; The inverted out-degree values ​​of all vertices in the first path are multiplied together to obtain a first ranking score representing the semantic importance of the first path in the external graph data.

18. The readable storage medium according to claim 16, wherein: The acquiring, in the target graph data, a second path starting from the first vertex and having a secondary intersection with the first path, includes: Obtaining secondary intersection points where the third vertices corresponding to all matching pairs in the target graph data intersect with the first path; According to the vertices and links between the first vertex as the starting point and the second intersection point in the target graph data, a second path with the first vertex as the starting point and having a second intersection with the first path is obtained.

19. The readable storage medium according to claim 16, wherein: The calculating a second ranking score representing the semantic importance of the second path in the target graph data includes: Obtaining the out-degree value of each vertex in the second path; For any vertex in the second path, perform a reciprocal operation on the out-degree value of the vertex to obtain the inverted out-degree value of the vertex; The inverted out-degree values ​​of all vertices in the second path are multiplied together to obtain a second ranking score representing the semantic importance of the second path in the target graph data.

20. The readable storage medium according to claim 18, wherein The filtering of the external graph data according to the first ranking score and the second ranking score to obtain filtered external graph data includes: In the target graph data, calculating a third ranking score representing the semantic relevance between the first path and the target graph data based on the second ranking scores of all second paths starting from the first vertex and having a secondary intersection with the first path; Adding the first ranking score and the third ranking score to obtain a dual-path ranking score of the first path; Aggregating the dual-path ranking scores of all first paths between the first vertex and the second vertex to obtain a total score of the path set between the first vertex and the second vertex; The external graph data is filtered according to the total scores of all path sets in the external graph data and a preset total score threshold to obtain filtered external graph data.

Images