Knowledge graph fusion method, device and equipment based on label propagation

Abstract

The invention discloses a knowledge graph fusion method, device and equipment based on label propagation. The method comprises the following steps: acquiring a label propagation command, wherein the command comprises a knowledge graph list and a result output path; respectively executing a label propagation algorithm on each knowledge graph in the knowledge graph list, and determining initial label value distribution of each knowledge graph; determining an equivalent entity set among different knowledge maps; based on an initial label value and a local topological structure of each equivalent entity in the equivalent entity set in the knowledge graph, fusing the initial label value and the topological structure to obtain a fused entity, a fused label value and a fused topological structure; and according to the fusion entity, the fusion label value and the fusion topological structure, fusing the at least two knowledge maps in the knowledge map list to obtain a fusion result, and storing the fusion result in a result output path. According to the embodiment of the invention, cross-graph complex semantic calculation and conjoint analysis are supported.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of knowledge graph, enterprise risk control and financial data analysis, and particularly relates to a knowledge graph fusion method and device based on label propagation and equipment. BACKGROUND

[0002] With the development of big data and artificial intelligence technology, as an effective knowledge representation and reasoning technology, knowledge graph has been widely used in search engines, intelligent question answering, risk control and other fields. In practical applications, knowledge data often comes from multiple heterogeneous and independently constructed knowledge bases, thus forming multiple isolated knowledge graphs. For example, in the field of financial risk control, the equity information, senior management information and investment relationship of an enterprise may be scattered in multiple independent knowledge graphs such as industry and commerce, justice and credit investigation.

[0003] In order to obtain a global knowledge view, the multiple knowledge graphs need to be fused. The existing technology usually focuses on entity alignment (i.e. judging whether the entities in different graphs point to the same real object) and attribute fusion. However, the existing fusion methods have obvious deficiencies when facing scenarios that require cross-graph deep semantic calculation. Specifically, when the complete semantics of an entity needs to be derived from multiple equivalent entities of the entity in different graphs and their respective local network relationships, the existing methods are not up to the task.

[0004] A typical scenario is to calculate the comprehensive net assets of an enterprise group. A holding company may have direct net asset records in one graph, while the net assets of its subsidiaries and associated companies are recorded in another graph. Although the existing methods can identify the shareholding relationship (equivalent entities) between these companies, they lack a mechanism to simulate how the net assets flow, aggregate and redistribute in the complex equity network composed of multiple graphs according to these shareholding ratios. Although traditional label propagation algorithms can spread information within a single graph, their inherent "information replication" model is not suitable for simulating the transfer of such conserved values, and they do not have the ability to perform cross-graph collaborative calculation.

[0005] Therefore, there is an urgent need in the art for a new fusion method that can coordinate multiple knowledge graphs and support complex calculation models based on value transfer, in order to solve the above deep semantic calculation problem. SUMMARY

[0006] The present application provides a knowledge graph fusion method, device and equipment based on label propagation, to solve the problem that traditional label propagation algorithms can only spread information within a single graph, cannot realize the transfer of conserved values between entities, and do not have the ability to perform cross-graph collaborative calculation; and the problem that existing knowledge graph fusion methods cannot perform deep semantic calculation across graphs.

[0007] According to an aspect of the present application, there is provided a label propagation based knowledge graph fusion method, comprising:

[0008] A label propagation command is obtained, the command comprising a knowledge graph list and a result output path;

[0009] A label propagation algorithm is executed on each knowledge graph in the knowledge graph list respectively to determine an initial label value distribution of each knowledge graph;

[0010] Based on a pre-established entity equivalence relationship, an equivalent entity set between different knowledge graphs is determined; the equivalent entities in the same equivalent entity set satisfy the entity equivalence relationship;

[0011] Based on the initial label value and local topology structure of each equivalent entity in the equivalent entity set in the knowledge graph, the initial label value and topology structure are fused to obtain a fused entity, a fused label value and a fused topology structure;

[0012] Based on the fused entity, the fused label value and the fused topology structure, at least two knowledge graphs in the knowledge graph list are fused to obtain a fusion result, and the fusion result is stored in the result output path.

[0013] According to another aspect of the present application, there is provided a label propagation based knowledge graph fusion device, comprising:

[0014] A command acquisition module is configured to obtain a label propagation command, the command comprising a knowledge graph list and a result output path;

[0015] A label propagation module is configured to execute a label propagation algorithm on each knowledge graph in the knowledge graph list respectively to determine an initial label value distribution of each knowledge graph;

[0016] An equivalent entity determination module is configured to determine an equivalent entity set between different knowledge graphs based on a pre-established entity equivalence relationship; the equivalent entities in the same equivalent entity set satisfy the entity equivalence relationship;

[0017] An entity fusion module is configured to fuse the initial label value and topology structure based on the initial label value and local topology structure of each equivalent entity in the equivalent entity set in the knowledge graph to obtain a fused entity, a fused label value and a fused topology structure;

[0018] A graph fusion module is configured to fuse at least two knowledge graphs in the knowledge graph list based on the fused entity, the fused label value and the fused topology structure to obtain a fusion result, and store the fusion result in the result output path.

[0019] According to another aspect of the present invention, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the tag-based knowledge graph fusion method according to any embodiment of the present invention.

[0020] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0021] At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to execute the tag-based knowledge graph fusion method according to any embodiment of the present invention.

[0022] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the tag-based knowledge graph fusion method according to any embodiment of the present invention.

[0023] According to another aspect of the present invention, a computer program product is provided, comprising a computer program / instructions that, when executed by a processor, implement the tag-based knowledge graph fusion method as described in any embodiment of the present invention.

[0024] This invention effectively solves the consistency challenges in entity alignment, attribute fusion, and topology integration of multi-source heterogeneous knowledge graphs by introducing a cross-knowledge graph tag propagation and collaborative fusion mechanism. First, an initial tag value distribution is generated in each independent graph using a tag propagation algorithm. Then, based on entity equivalence relations, equivalent entity sets across graphs are identified. Finally, by fusing their tag values ​​and local topological structures, a unified fused entity and global topology are constructed, ultimately achieving deep fusion of multiple knowledge graphs at the entity, attribute, and relation levels. This improves the automation and accuracy of knowledge fusion, supports complex semantic computation and joint analysis across graphs, and provides a reliable technical foundation for the unified management and application of large-scale knowledge graphs.

[0025] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0026] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0027] Figure 1 This is a first flowchart of a knowledge graph fusion method based on tag propagation provided in an embodiment of the present invention;

[0028] Figure 2 This is a second flowchart of a knowledge graph fusion method based on tag propagation provided in an embodiment of the present invention;

[0029] Figure 3 This is a schematic diagram of the structure of a knowledge graph fusion device based on tag propagation provided in an embodiment of the present invention;

[0030] Figure 4 This is a schematic diagram of the structure of an electronic device that implements an embodiment of the present invention. Detailed Implementation

[0031] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0032] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0033] Figure 1This is a first flowchart of a knowledge graph fusion method based on tag propagation provided in this embodiment of the invention. This embodiment is applicable to the conservation of value transfer between entities and the realization of deep semantic computation across the graph. This method can be executed by a knowledge graph fusion device based on tag propagation, which can be implemented in hardware and / or software and can be configured in an electronic device with corresponding data processing capabilities. Figure 1 As shown, the method includes:

[0034] S110. Obtain a tag propagation command, the command including a knowledge graph list and a result output path; on each knowledge graph in the knowledge graph list, execute the tag propagation algorithm to determine the initial tag value distribution of each knowledge graph.

[0035] Optionally, the knowledge graphs in the knowledge graph list reside in different graph databases; that is, each knowledge graph may be stored in a graph database from a different vendor or with a different version. The step of executing a tag propagation algorithm on each knowledge graph in the knowledge graph list to determine the initial tag value distribution of each knowledge graph includes: concurrently executing the tag propagation algorithm on each knowledge graph in the knowledge graph list and returning the task identifier of the tag propagation for each knowledge graph to the user, so that the user can asynchronously query the execution status of each task.

[0036] The process of executing the label propagation algorithm on any knowledge graph includes: converting the unified label propagation command into native computation instructions adapted to the target graph database where the knowledge graph is located through protocol conversion, so as to trigger the label propagation computation process of the target graph database server; obtaining and receiving the original computation results returned by the target graph database, and converting the original computation results into a standard format as the initial label value distribution of the knowledge graph.

[0037] The number of graph databases can be expanded, the service nodes of each graph database server can be expanded, and an additional gateway layer can be added between the graph database server and the upper-layer application to achieve load balancing or improve service availability.

[0038] By introducing a protocol conversion layer and a scalable service node architecture, unified computing capabilities across heterogeneous graph databases are effectively achieved. This design supports protocol adaptation and instruction conversion for graph databases from different vendors and versions, allowing upper-layer applications to concurrently trigger tag propagation computations across multiple knowledge graphs without needing to concern themselves with differences in the underlying data sources, simply through standardized interfaces. Simultaneously, asynchronous queries and status tracking are implemented through a task identification mechanism, and load balancing and high-availability deployment are supported at the data service layer, significantly improving the system's compatibility, scalability, and operational reliability in complex enterprise environments.

[0039] S120. Based on the pre-established entity equivalence relation, determine the equivalent entity set between different knowledge graphs; the equivalent entities in the same equivalent entity set satisfy the entity equivalence relation.

[0040] The equivalent entity set refers to a group of entities distributed across two or more different knowledge graphs, which are determined to point to the same real-world object based on pre-established entity equivalence relations; this group of entities forms the basis of a "fusion entity".

[0041] In one optional implementation, the command further includes an equivalence discrimination attribute. The step of determining equivalent entities between different knowledge graphs based on pre-established entity equivalence relations includes: traversing the entities in the current knowledge graph; comparing the current entity with entities in other knowledge graphs in the knowledge graph list one by one; the comparison uses the entity's equivalence discrimination attribute as the basis for similarity calculation; if the similarity between the current entity and any entity in another knowledge graph meets a preset condition, then the two are determined to be equivalent entities.

[0042] The tag propagation command also includes equivalence discrimination attributes, which are the core basis for calculating cross-knowledge graph entity similarity in each knowledge graph. By specifying comparable attribute fields, it provides a discrimination basis for determining equivalent entities between different knowledge graphs.

[0043] When determining equivalent entities between different knowledge graphs, each entity in the current knowledge graph is traversed, and it is compared one by one with all entities in other knowledge graphs in the knowledge graph list. During the comparison, the similarity between entities is calculated based on equivalence discrimination attributes (such as unified social credit code, registered name, etc.). If the similarity between the current entity and an entity in another knowledge graph reaches a preset threshold (e.g., semantic similarity > 0.9), the two are determined to be equivalent entities, and their attribute mapping relationship is recorded. Optionally, a multi-attribute weighted fusion similarity calculation strategy is supported to ensure the accuracy of the discrimination results.

[0044] By introducing configurable equivalence discrimination attributes and an automatic entity comparison mechanism, entity equivalence relationships across knowledge graphs can be established efficiently and accurately, reducing the reliance on manual annotation in multi-source data fusion. At the same time, the attribute similarity-based quantitative discrimination strategy enhances the fault tolerance for entity ambiguity and redundancy in heterogeneous data, providing a reliable data foundation for subsequent cross-graph label propagation and unified view generation, and improving the automation level and scalability of the knowledge fusion system.

[0045] S130. Based on the initial label value and local topology of each equivalent entity in the knowledge graph in the equivalent entity set, the initial label value and topology are fused to obtain a fused entity, a fused label value, and a fused topology.

[0046] A fused entity refers to an authoritative entity that represents a unique real-world object, generated by merging a set of equivalent entities from multiple knowledge graphs; it is a standard node in cross-knowledge graph operations.

[0047] Optionally, the local topology is a subgraph within one degree centered on the corresponding entity. For each identified set of equivalent entities, a multi-source information fusion operation is performed based on the initial label values ​​and local topology of each equivalent entity in its respective knowledge graph. Specifically, this includes: Entity generation: mapping the set of equivalent entities to a unified fusion entity, serving as the authoritative representation of the real-world object in the fusion graph; Tag value calculation: generating the fusion label value of the fusion entity based on the initial label values ​​of each equivalent entity through weighted averaging, voting, or confidence-based aggregation strategies; and Topology construction: integrating the local adjacency relationships of each equivalent entity in its respective knowledge graph, merging weights and removing redundancy from repeated edges to form the connection structure of the fusion entity in the fusion graph. This ensures that semantic information and structural features across the knowledge graph are effectively integrated, supporting subsequent global analysis and querying.

[0048] S140. Based on the fusion entity, fusion tag value, and fusion topology, at least two knowledge graphs in the knowledge graph list are fused to obtain a fusion result, and the fusion result is stored in the result output path.

[0049] Based on the generated fused entities, fused tag values, and fused topology, the knowledge graphs in the knowledge graph list are uniformly integrated. The fusion operation covers two types of entity objects: fused entities and unique entities. Unique entities refer to non-equivalent entities in each knowledge graph. Fused entities, their fused tag values, and the topological relationships between them are injected into the fused graph, serving as authoritative representations of real-world objects across the knowledge graph. The initial tag values ​​and local topology of unique entities are migrated to the fused graph, ensuring the integrity and consistency of connections. The fused graph is persistently stored in a standardized graph data format to the user-specified output path, supporting subsequent graph queries, visualization, and analysis tasks. Further tag propagation can also be performed on the fused knowledge graph according to preset rules.

[0050] By employing entity disambiguation, relation alignment, and weight normalization, a structurally consistent fused knowledge graph is constructed. The fusion results are serialized in a standardized format and output to a user-specified path, supporting cross-graph joint queries and macro-level analysis.

[0051] This invention effectively solves the consistency challenges in entity alignment, attribute fusion, and topology integration of multi-source heterogeneous knowledge graphs by introducing a cross-knowledge graph tag propagation and collaborative fusion mechanism. First, an initial tag value distribution is generated in each independent graph using a tag propagation algorithm. Then, based on entity equivalence relations, equivalent entity sets across graphs are identified. Finally, by fusing their tag values ​​and local topological structures, a unified fused entity and global topology are constructed, ultimately achieving deep fusion of multiple knowledge graphs at the entity, attribute, and relation levels. This improves the automation and accuracy of knowledge fusion, supports complex semantic computation and joint analysis across graphs, and provides a reliable technical foundation for the unified management and application of large-scale knowledge graphs.

[0052] In one optional implementation, a label propagation algorithm is executed to determine the initial label value distribution of the knowledge graph, including: synchronously updating the label values ​​in an iterative manner based on the original label values ​​of each entity in the knowledge graph and the propagation weights of the original edges between entities; in each iteration, each entity allocates its own label value according to the propagation weights of the entity's outgoing edges, and integrates the label values ​​of neighboring entities received from the entity's incoming edges to complete the update of its own label value; when the iteration process meets a preset termination condition, the entity's current label value is used as the entity's initial label value to obtain the initial label value distribution of the knowledge graph.

[0053] Specifically, the following operations are performed during each iteration:

[0054] a) For each entity, calculate the label value to be assigned to each outgoing neighbor entity based on the propagation weight of the entity's outgoing edge and the original label value; the product of the propagation weight of the outgoing edge and the original label value of the entity can be calculated as the label value to be assigned to that outgoing neighbor entity.

[0055] b) Subtract the label values ​​assigned to each outgoing neighbor entity from the current label value of the entity, and add the label values ​​assigned to the entity by all incoming neighbor entities to complete the label value update of the entity.

[0056] Repeat steps a) and b) until the change in the label values ​​of all entities is less than the preset threshold or the maximum number of iterations is reached.

[0057] By introducing a label propagation mechanism based on value conservation, the algorithm achieves precise allocation and synchronous updating of entity label values ​​according to propagation weights in each iteration. This not only effectively simulates the flow of resources along network paths in real-world scenarios (such as the transmission of equity funds along shareholding relationships), but also ensures the strict conservation of the total label value of the system through a closed-loop calculation of "allocation-reception-net value update". This iterative algorithm possesses both clear mathematical interpretability and good convergence, and can efficiently generate an initial label distribution that conforms to real-world patterns, providing a high-quality, interpretable data foundation for subsequent cross-knowledge graph fusion.

[0058] Figure 2 This is a second flowchart of a knowledge graph fusion method based on tag propagation provided in an embodiment of the present invention. This embodiment is an optimization and improvement based on the above embodiment. Figure 2 As shown, the method includes:

[0059] S210. Obtain a tag propagation command, the command including a knowledge graph list and a result output path; on each knowledge graph in the knowledge graph list, execute the tag propagation algorithm to determine the initial tag value distribution of each knowledge graph.

[0060] S220. Based on the pre-established entity equivalence relation, determine the equivalent entity set between different knowledge graphs; the equivalent entities in the same equivalent entity set satisfy the entity equivalence relation.

[0061] S230. Create a fused entity based on each equivalent entity in the equivalent entity set; determine the fused tag value of the fused entity based on the initial tag value of each equivalent entity.

[0062] Optionally, the arithmetic mean of the initial label values ​​of each equivalent entity is used as the fused label value of the fused entity.

[0063] S240. Based on the propagation weights of the original edges contained in the local topology of each equivalent entity, determine the fusion propagation weights between the fused entity and other fused entities through weight aggregation.

[0064] Original edges refer to edges that directly connect two entities within a single knowledge graph. The propagation weights attached to these edges are the foundational data for calculating the fusion propagation weights. By analyzing the local topology of each equivalent entity within its respective knowledge graph, the connection relationships between fusion entities are constructed. Specifically, direct association edges between different equivalent entities are identified, and the weights of multiple original edges connecting the same pair of fusion entities are aggregated (e.g., by taking the maximum value, summing, or weighted averaging). The aggregation result is then used as the fusion propagation weights between fusion entities, thereby establishing a unified fusion topology. This effectively preserves the structural features of multi-source graphs, ensuring the integrity of relationships and weight consistency between entities in the fusion graph.

[0065] Optionally, the step of determining the fusion propagation weight between the fused entity and other fused entities through weight aggregation based on the propagation weights of the original edges contained in the local topology of each equivalent entity includes:

[0066] For the first fused entity and the second fused entity, the sum of the propagation weights of the original edges between all equivalent entities in the first equivalent entity set and the second equivalent entity set is used as the initial propagation weight between the first fused entity and the second fused entity.

[0067] For each neighboring fusion entity connected to the first fusion entity, the total propagation weight of the first fusion entity is the sum of the propagation weights of the original edges between the first equivalent entity set and all equivalent entities in the neighboring equivalent entity set.

[0068] Divide the initial propagation weight by the total propagation weight to obtain the fusion propagation weight of the edge between the first fusion entity and the second fusion entity.

[0069] Wherein, the first equivalent entity set is the equivalent entity set that generates the first fused entity; the second equivalent entity set is the equivalent entity set that generates the second fused entity.

[0070] For the first and second fused entities that need to establish a connection, traverse their corresponding first and second equivalent entity sets, calculate the sum of the propagation weights of the original edges between all equivalent entity pairs in these two sets, and use this sum as the initial propagation weight between the first and second fused entities.

[0071] Identify all neighboring fusion entities that are connected to the first fusion entity, and for each neighboring fusion entity, calculate the sum of the propagation weights of all original edges between the first equivalent entity set and its corresponding neighboring equivalent entity set. Accumulate the sums of the propagation weights corresponding to each neighboring fusion entity to obtain the total propagation weight of the first fusion entity. In the fusion knowledge graph, fusion entities directly connected to the current fusion entity are called the neighboring fusion entities of the current fusion entity, and the equivalent entity set of the generated neighboring fusion entity is called the neighboring equivalent entity set.

[0072] The initial propagation weight between the first fusion entity and the second fusion entity is divided by the total propagation weight of the first fusion entity. This normalization operation yields the final fusion propagation weight of the edge between the first fusion entity and the second fusion entity.

[0073] By using weighted aggregation and normalization, the topological characteristics of the original graph are preserved, and the standardization and comparability of the propagation weights after fusion are ensured, providing reliable support for subsequent graph computation tasks.

[0074] S250. Based on the fusion entity, fusion tag value, and fusion topology, at least two knowledge graphs in the knowledge graph list are fused to obtain a fusion result, and the fusion result is stored in the result output path.

[0075] This invention achieves deep semantic fusion of multi-source heterogeneous knowledge graphs by constructing a collaborative computing framework across knowledge graphs. First, the initial label distribution of each knowledge graph is obtained through parallel label propagation. Then, fused entities and fused label values ​​are generated based on entity equivalence relations. Finally, topological connections between fused entities are constructed through weight aggregation technology, resulting in a unified knowledge graph fusion result. This effectively solves the consistency problems in entity alignment, attribute fusion, and structural integration of multi-source knowledge graphs, significantly improving the automation and accuracy of knowledge fusion, and providing reliable technical support for large-scale cross-domain knowledge reasoning and joint analysis.

[0076] In one alternative implementation, the entity is a company, the propagation weight is the company's shareholding ratio, and the tag value is the company's net assets; the method is used to realize the propagation and unification of the company's net assets across the knowledge graph composed of the group's affiliated companies.

[0077] Optionally, each knowledge graph is an enterprise-oriented knowledge graph, where the entities correspond to enterprise nodes, the propagation weight between entities is the enterprise's shareholding ratio, and the label value is the enterprise's net assets. First, net asset propagation calculations are performed in each enterprise knowledge graph based on the shareholding ratio, generating the initial net asset distribution of enterprises within each graph. Then, related enterprises within the group are identified through enterprise entity alignment, and the net asset data of these enterprises is integrated to construct a cross-graph equity topology network. Finally, a unified view reflecting the overall net asset distribution of the group is formed. By using the enterprise's shareholding ratio as the propagation weight and net assets as the label value, the propagation and unified calculation of net assets among related enterprises within the group across knowledge graphs are realized. This can accurately simulate the flow path of net assets along equity relationships, effectively solving problems such as incomplete net asset statistics and fragmented related shareholding relationships caused by data dispersion across different graphs for group enterprises, providing full-link data support for group asset valuation, risk transmission analysis, and capital structure optimization.

[0078] Figure 3 This is a schematic diagram of a knowledge graph fusion device based on tag propagation provided in an embodiment of the present invention. Figure 3 As shown, the device includes:

[0079] The command acquisition module 310 is used to acquire the tag propagation command, which includes a knowledge graph list and a result output path;

[0080] The tag propagation module 320 is used to execute the tag propagation algorithm on each knowledge graph in the knowledge graph list to determine the initial tag value distribution of each knowledge graph.

[0081] The equivalent entity determination module 330 is used to determine the equivalent entity set between different knowledge graphs based on the pre-established entity equivalence relation; the equivalent entities in the same equivalent entity set satisfy the entity equivalence relation;

[0082] The entity fusion module 340 is used to fuse the initial label value and the local topology of each equivalent entity in the knowledge graph based on the initial label value and the local topology of each equivalent entity in the equivalent entity set, to obtain a fused entity, a fused label value and a fused topology.

[0083] The graph fusion module 350 is used to fuse at least two knowledge graphs in the knowledge graph list according to the fusion entity, fusion tag value and fusion topology, to obtain the fusion result, and to store the fusion result in the result output path.

[0084] The knowledge graph fusion device based on tag propagation provided in the embodiments of the present invention can execute the knowledge graph fusion method based on tag propagation provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0085] Optionally, the entity fusion module includes:

[0086] The entity fusion determination unit is used to create a fusion entity based on each equivalent entity in the equivalent entity set.

[0087] A fusion tag value determination unit is used to determine the fusion tag value of the fusion entity based on the initial tag values ​​of each equivalent entity;

[0088] The fusion topology determination unit is used to determine the fusion propagation weight between the fusion entity and other fusion entities by weight aggregation based on the propagation weight of the original edges contained in the local topology of each equivalent entity.

[0089] Optionally, the fusion topology determination unit is specifically used for:

[0090] For the first fused entity and the second fused entity, the sum of the propagation weights of the original edges between all equivalent entities in the first and second equivalent entity sets is used as the initial propagation weight between the first fused entity and the second fused entity. For each neighboring fused entity connected to the first fused entity, the sum of the propagation weights of the original edges between all equivalent entities in the first and neighboring equivalent entity sets is used as the total propagation weight of the first fused entity. The initial propagation weight is divided by the total propagation weight to obtain the fusion propagation weight of the edge between the first fused entity and the second fused entity.

[0091] Optionally, the label propagation module includes a label propagation unit, used to perform synchronous update of label values ​​in an iterative manner based on the original label values ​​of each entity in the knowledge graph and the propagation weights of the original edges between entities; in each iteration, each entity allocates its own label value according to the propagation weights of the entity's outgoing edges, and integrates the label values ​​of neighboring entities received from the entity's incoming edges to complete the update of its own label value; when the iteration process meets a preset termination condition, the entity's current label value is used as the entity's initial label value to obtain the initial label value distribution of the knowledge graph.

[0092] Optionally, the knowledge graphs in the knowledge graph list reside in different graph databases; the tag propagation module includes: a concurrent execution unit, used to concurrently execute the tag propagation algorithm on each knowledge graph in the knowledge graph list; specifically, the concurrent execution unit is used to convert the unified tag propagation command into native computation instructions adapted to the target graph database where the knowledge graph resides through protocol conversion, so as to trigger the tag propagation computation process of the target graph database server; obtain and receive the original computation results returned by the target graph database, and convert the original computation results into a standard format as the initial tag value distribution of the knowledge graph.

[0093] Optionally, the entity is an enterprise, the propagation weight is the enterprise's shareholding ratio, and the tag value is the enterprise's net assets; the method is used to realize the propagation and unification of the enterprise's net assets across the knowledge graph composed of various affiliated enterprises of the group.

[0094] The tag-based knowledge graph fusion device further illustrates that it can also execute the tag-based knowledge graph fusion method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0095] According to embodiments of the present invention, the present invention also provides an electronic device, a readable storage medium, and a computer program product.

[0096] Figure 4 A schematic diagram of an electronic device 40 that can be used to implement embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0097] like Figure 4As shown, the electronic device 40 includes at least one processor 41 and a memory, such as a read-only memory 42 or a random access memory 43, communicatively connected to the at least one processor 41. The memory stores computer programs executable by the at least one processor. The processor 41 can perform various appropriate actions and processes based on the computer program stored in the read-only memory 42 or loaded from storage unit 48 into the random access memory 43. The random access memory 43 may also store various programs and data required for the operation of the electronic device 40. The processor 41, read-only memory 42, and random access memory 43 are interconnected via a bus 44. An input / output interface 45 is also connected to the bus 44.

[0098] Multiple components in electronic device 40 are connected to input / output interface 45, including: input unit 46, such as keyboard, mouse, etc.; output unit 47, such as various types of monitors, speakers, etc.; storage unit 48, such as disk, optical disk, etc.; and communication unit 49, such as network card, modem, wireless transceiver, etc. Communication unit 49 allows electronic device 40 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0099] Processor 41 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 41 include, but are not limited to, central processing units, graphics processing units, various special-purpose artificial intelligence computing chips, various processors running machine learning model algorithms, digital signal processors, and any suitable processor, controller, microcontroller, etc. Processor 41 performs the various methods and processes described above, such as label propagation-based knowledge graph fusion methods.

[0100] In some embodiments, the tag-based knowledge graph fusion method can be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 48. In some embodiments, part or all of the computer program can be loaded and / or installed on electronic device 40 via read-only memory 42 and / or communication unit 49. When the computer program is loaded into random access memory 43 and executed by processor 41, one or more steps of the tag-based knowledge graph fusion method described above can be performed. Alternatively, in other embodiments, processor 41 can be configured to perform the tag-based knowledge graph fusion method by any other suitable means (e.g., by means of firmware).

[0101] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays, application-specific integrated circuits (ASICs), application-specific standard products (ASICs), systems-on-a-chip (SoCs), payload programmable logic devices, computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0102] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0103] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory, read-only memory, erasable programmable read-only memory, optical fibers, portable compact disk read-only memory, optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0104] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a cathode ray tube, liquid crystal display, or monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0105] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0106] A computing system can include clients and servers. Clients and servers are generally geographically separated and typically interact via communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a host product within the cloud computing service system to address the shortcomings of traditional physical hosts and virtual private servers, such as high management difficulty and weak business scalability.

[0107] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0108] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A knowledge graph fusion method based on tag propagation, characterized in that, The method includes: Obtain the tag propagation command, which includes a knowledge graph list and a result output path; On each knowledge graph in the knowledge graph list, a label propagation algorithm is executed to determine the initial label value distribution of each knowledge graph; Based on pre-established entity equivalence relations, the sets of equivalent entities between different knowledge graphs are determined; the equivalent entities in the same set of equivalent entities satisfy the entity equivalence relations. Based on the initial label value and local topology of each equivalent entity in the knowledge graph in the equivalent entity set, the initial label value and topology are fused to obtain the fused entity, fused label value and fused topology. Based on the fused entity, fused tag value, and fused topology, at least two knowledge graphs in the knowledge graph list are fused to obtain a fusion result, and the fusion result is stored in the result output path.

2. The method according to claim 1, characterized in that, The step involves fusing the initial label value and local topology of each equivalent entity in the knowledge graph based on the set of equivalent entities to obtain a fused entity, a fused label value, and a fused topology, including: Create a merged entity based on each equivalent entity in the equivalent entity set; The fusion label value of the fused entity is determined based on the initial label values ​​of each equivalent entity; Based on the propagation weights of the original edges contained in the local topology of each equivalent entity, the fusion propagation weights between the fused entity and other fused entities are determined through weight aggregation.

3. The method according to claim 2, characterized in that, The step of determining the fusion propagation weight between the fused entity and other fused entities through weight aggregation, based on the propagation weights of the original edges contained in the local topology of each equivalent entity, includes: For the first fused entity and the second fused entity, the sum of the propagation weights of the original edges between all equivalent entities in the first equivalent entity set and the second equivalent entity set is used as the initial propagation weight between the first fused entity and the second fused entity. For each neighboring fusion entity connected to the first fusion entity, the total propagation weight of the first fusion entity is the sum of the propagation weights of the original edges between the first equivalent entity set and all equivalent entities in the neighboring equivalent entity set. Divide the initial propagation weight by the total propagation weight to obtain the fusion propagation weight of the edge between the first fusion entity and the second fusion entity.

4. The method according to claim 1, characterized in that, Execute the label propagation algorithm to determine the initial label value distribution of the knowledge graph, including: Based on the original label values ​​of each entity in the knowledge graph and the propagation weights of the original edges between entities, the label values ​​are updated synchronously in an iterative manner. In each iteration, each entity assigns its own label value according to the propagation weight of its outgoing edges, and integrates the label values ​​of neighboring entities received from its incoming edges to complete the update of its own label value. When the preset termination condition is met during the iteration process, the current label value of the entity is used as the initial label value of the entity to obtain the initial label value distribution of the knowledge graph.

5. The method according to claim 1, characterized in that, The knowledge graphs in the knowledge graph list are located in different graph databases; the label propagation algorithm is executed on each knowledge graph in the knowledge graph list to determine the initial label value distribution of each knowledge graph, including: The label propagation algorithm is executed concurrently on each knowledge graph in the knowledge graph list; The process of performing the label propagation algorithm on any knowledge graph includes: Through protocol conversion, the unified tag propagation command is converted into native computation instructions adapted to the target graph database where the knowledge graph is located, so as to trigger the tag propagation computation process of the target graph database server. The system acquires and receives the raw calculation results returned by the target graph database, converts the raw calculation results into a standard format, and uses them as the initial label value distribution of the knowledge graph.

6. The method according to any one of claims 2 to 5, characterized in that, The entity is an enterprise, the propagation weight is the enterprise's shareholding ratio, and the tag value is the enterprise's net assets; the method is used to realize the propagation and unification of the enterprise's net assets across the knowledge graph composed of various affiliated enterprises of the group.

7. A knowledge graph fusion device based on tag propagation, characterized in that, The device includes: The command acquisition module is used to acquire tag propagation commands, which include a knowledge graph list and a result output path. The tag propagation module is used to execute the tag propagation algorithm on each knowledge graph in the knowledge graph list to determine the initial tag value distribution of each knowledge graph. The equivalent entity determination module is used to determine the equivalent entity set between different knowledge graphs based on pre-established entity equivalence relations; the equivalent entities in the same equivalent entity set satisfy the entity equivalence relation. The entity fusion module is used to fuse the initial label value and local topology of each equivalent entity in the knowledge graph based on the initial label value and local topology of each equivalent entity in the equivalent entity set, to obtain fused entity, fused label value and fused topology. The graph fusion module is used to fuse at least two knowledge graphs in the knowledge graph list according to the fusion entity, fusion tag value and fusion topology, to obtain the fusion result, and to store the fusion result in the result output path.

8. An electronic device, characterized in that, The electronic device includes: At least one processor; and a memory communicatively connected to the at least one processor; The memory stores a computer program that can be executed by the at least one processor, which is then executed by the at least one processor to enable the at least one processor to perform the knowledge graph fusion method based on tag propagation as described in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the tag-based knowledge graph fusion method according to any one of claims 1-6.

10. A computer program product comprising a computer program that, when executed by a processor, implements the tag-based knowledge graph fusion method according to any one of claims 1-6.

Images