CN116010615A - Entity alignment method and device, electronic equipment and computer storage medium - Google Patents
Entity alignment method and device, electronic equipment and computer storage medium Download PDFInfo
- Publication number
- CN116010615A CN116010615A CN202211682651.4A CN202211682651A CN116010615A CN 116010615 A CN116010615 A CN 116010615A CN 202211682651 A CN202211682651 A CN 202211682651A CN 116010615 A CN116010615 A CN 116010615A
- Authority
- CN
- China
- Prior art keywords
- entity
- graph
- knowledge graph
- entities
- triplet
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The application provides an entity alignment method, an entity alignment device, electronic equipment and a computer storage medium, wherein the method comprises the following steps: preprocessing the obtained physical model knowledge graph and logic model knowledge graph to obtain a relationship triplet and an attribute triplet of the physical model knowledge graph and a relationship triplet and an attribute triplet of the logic model knowledge graph; inputting the knowledge graph into a multi-layer graph convolutional neural network to obtain an embedded representation of an entity in the knowledge graph; for each entity, calculating a cross-graph matching vector of neighbor nodes of the entity; determining a target embedded representation of the entity according to the cross-graph matching vector of the neighbor node of the entity and the embedded representation of the entity; aggregating the entities in the knowledge graph to obtain an embedded representation of the optimized entity; and determining the distance between the two knowledge-graph entities according to the embedded representation of the entity in the knowledge-graph after the optimization and the vectorization representation of the entity in the other knowledge-graph, thereby realizing the alignment of the entities in the two graphs.
Description
Technical Field
The present invention relates to the field of computer technologies, and in particular, to an entity alignment method, an entity alignment device, an electronic device, and a computer storage medium.
Background
At present, the entity alignment problem in the knowledge graph is one of the most deeply studied problems in the knowledge graph field, and the entity alignment problem related to the knowledge graph in the early stage is an entity alignment method based on embedding.
The most typical method is based on a translation model-TransE model, and the method considers that a correct knowledge-graph triplet meets the methods of head entity embellishment plus relation embellishment which is equal to tail entity embellishment, and then the method is improved on the basis, and MTransE, IPTransE, JAPE and the like are deformed. However, some problems still exist in the methods, such as the problem that two knowledge-graph neighbors are different in size, and the problem that popular neighbor nodes commonly appear in the knowledge-graph.
Disclosure of Invention
In view of this, the present application provides a method, apparatus, electronic device, and computer storage medium for entity alignment, which efficiently completes the task of entity alignment matching between two knowledge maps of a physical model knowledge map and a logical model knowledge map.
The first aspect of the present application provides an alignment method for an entity, including:
acquiring a physical model knowledge graph and a logic model knowledge graph; the knowledge graph is represented by a directed graph, and the directed graph comprises entities, relations and triples; the triples consist of entities, relationships/attributes, entities;
preprocessing the physical model knowledge graph and the logic model knowledge graph respectively to obtain a relationship triplet and an attribute triplet of the physical model knowledge graph and a relationship triplet and an attribute triplet of the logic model knowledge graph;
inputting the relation triplets and the attribute triplets of the physical model knowledge graph and the logic model knowledge graph into a multi-layer graph convolutional neural network to obtain the embedded representation of the entity in the knowledge graph;
for each entity, calculating a cross-graph matching vector of neighbor nodes of the entity;
determining a target embedded representation of the entity according to the cross-graph matching vector of the neighbor node of the entity and the embedded representation of the entity;
aggregating the entities in the knowledge graph to obtain an embedded representation of the optimized entity;
determining the distance between two knowledge-graph entities according to the embedded representation of the entity in the knowledge-graph after optimization and the vectorized representation of the entity in the other knowledge-graph;
and according to the distance between the two knowledge graph entities, realizing the alignment of the entities in the two graphs.
Optionally, the preprocessing is performed on the physical model knowledge graph and the logic model knowledge graph respectively to obtain a relationship triplet and an attribute triplet of the physical model knowledge graph, and the relationship triplet and the attribute triplet of the logic model knowledge graph, which include:
extracting the triples in the physical model knowledge graph, and performing normalization processing to obtain a relation triplet and an attribute triplet of the physical model knowledge graph;
and extracting the triples in the logic model knowledge graph, and performing normalization processing to obtain the relation triples and the attribute triples of the logic model knowledge graph.
Optionally, for each entity, calculating a cross-graph matching vector of a neighboring node of the entity includes:
calculating entity sampling probability of the entity neighbor node for each entity;
comparing the sub-graph of the entity with all sub-graphs in another knowledge graph, and selecting the most similar sub-graph as a target sub-graph;
and determining a cross-graph matching vector of the entity according to the target subgraph, the attention weight and the embedded representation of the entity in the knowledge graph.
Optionally, the aggregating the entities in the knowledge graph to obtain the optimized embedded representation of the entities includes:
updating the embedded representation of the central node by aggregating the neighbor node information of the central node, and aggregating all the associated node information to obtain the embedded representation of the optimized entity.
A second aspect of the present application provides an alignment device for an entity, including:
the acquisition unit is used for acquiring a physical model knowledge graph and a logic model knowledge graph; the knowledge graph is represented by a directed graph, and the directed graph comprises entities, relations and triples; the triples consist of entities, relationships/attributes, entities;
the preprocessing unit is used for respectively preprocessing the physical model knowledge graph and the logic model knowledge graph to obtain a relation triplet and an attribute triplet of the physical model knowledge graph and a relation triplet and an attribute triplet of the logic model knowledge graph;
the input unit is used for inputting the relation triplet and the attribute triplet of the physical model knowledge graph and the logic model knowledge graph into the multi-layer graph convolutional neural network to obtain the embedded representation of the entity in the knowledge graph;
a calculating unit, configured to calculate, for each entity, a cross-graph matching vector of a neighboring node of the entity;
a first determining unit, configured to determine a target embedded representation of the entity according to a cross-graph matching vector of a neighboring node of the entity and the embedded representation of the entity;
the aggregation unit is used for aggregating the entities in the knowledge graph to obtain an embedded representation of the optimized entity;
the second determining unit is used for determining the distance between the two knowledge-graph entities according to the embedded representation of the entity in the knowledge-graph after the optimization and the vectorized representation of the entity in the other knowledge-graph;
and the alignment unit is used for realizing the alignment of the entities in the two atlases according to the distance between the two knowledge atlas entities.
Optionally, the preprocessing unit includes:
the first preprocessing subunit is used for extracting the triples in the physical model knowledge graph and performing normalization processing to obtain the relation triples and the attribute triples of the physical model knowledge graph;
and the second preprocessing subunit is used for extracting the triples in the logic model knowledge graph and performing normalization processing to obtain the relation triples and the attribute triples of the logic model knowledge graph.
Optionally, the computing unit includes:
a first calculating subunit, configured to calculate, for each entity, an entity sampling probability of a neighbor node of the entity;
the selecting unit is used for comparing the sub-graph of the entity with all sub-graphs in another knowledge graph, and selecting the most similar sub-graph as a target sub-graph;
and the second calculating subunit is used for determining a cross-graph matching vector of the entity according to the target subgraph, the attention weight and the embedded representation of the entity in the knowledge graph.
Optionally, the polymerization unit includes:
and the aggregation subunit is used for updating the embedded representation of the central node by aggregating the neighbor node information of the central node, and aggregating all the related node information to obtain the embedded representation of the optimized entity.
A third aspect of the present application provides an electronic device, comprising:
one or more processors;
a storage device having one or more programs stored thereon;
the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the method of alignment of entities of any of the first aspects.
A fourth aspect of the present application provides a computer storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the method of alignment of entities according to any of the first aspects.
As can be seen from the above solutions, the present application provides an entity alignment method, an apparatus, an electronic device, and a computer storage medium, where the entity alignment method includes: firstly, acquiring a physical model knowledge graph and a logic model knowledge graph; the knowledge graph is represented by a directed graph, and the directed graph comprises entities, relations and triples; the triples consist of entities, relationships/attributes, entities; then, preprocessing the physical model knowledge graph and the logic model knowledge graph respectively to obtain a relation triplet and an attribute triplet of the physical model knowledge graph and a relation triplet and an attribute triplet of the logic model knowledge graph; then, inputting the relation triplet and the attribute triplet of the physical model knowledge graph and the logic model knowledge graph into a multi-layer graph convolutional neural network to obtain the embedded representation of the entity in the knowledge graph; for each entity, calculating a cross-graph matching vector of neighbor nodes of the entity; determining a target embedded representation of the entity according to the cross-graph matching vector of the neighbor node of the entity and the embedded representation of the entity; then, the entities in the knowledge graph are aggregated to obtain an embedded representation of the optimized entities; determining the distance between two knowledge-graph entities according to the embedded representation of the entity in the knowledge-graph after optimization and the vectorized representation of the entity in the other knowledge-graph; finally, according to the distance between the two knowledge graph entities, the alignment of the entities in the two graphs is realized. Therefore, the entity alignment matching task between the two knowledge maps of the physical model knowledge map and the logic model knowledge map is efficiently completed.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings that are required to be used in the embodiments or the description of the prior art will be briefly described below, and it is obvious that the drawings in the following description are only embodiments of the present application, and that other drawings may be obtained according to the provided drawings without inventive effort to a person skilled in the art.
Fig. 1 is a specific flowchart of an entity alignment method provided in an embodiment of the present application;
FIG. 2 is a flow chart of an entity alignment method according to another embodiment of the present application;
FIG. 3 is a flowchart of an entity alignment method according to another embodiment of the present application;
FIG. 4 is a schematic diagram of an alignment device for an entity according to another embodiment of the present disclosure;
fig. 5 is a schematic diagram of an electronic device for implementing an entity alignment method according to another embodiment of the present application.
Detailed Description
The following description of the embodiments of the present application will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are only some, but not all, of the embodiments of the present application. All other embodiments, which can be made by one of ordinary skill in the art without undue burden from the present disclosure, are within the scope of the present disclosure.
It should be noted that the terms "first," "second," and the like in this application are used merely to distinguish between different devices, modules, or units and are not intended to limit the order or interdependence of functions performed by such devices, modules, or units, but the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but also other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
The embodiment of the application provides an entity alignment method, as shown in fig. 1, specifically including the following steps:
s101, acquiring a physical model knowledge graph and a logic model knowledge graph.
The knowledge graph is represented by a directed graph, and the directed graph comprises entities, relations and triples; a triplet consists of entities, relationships/attributes, entities.
Specifically, according to a physical model knowledge graph KG1 and a logic model knowledge graph KG2 of a national power grid unified data model (SG-CIM), the knowledge graphs are represented by a directed graph g= (E, R, T), wherein E represents an entity, R represents a relationship, T represents a triplet, and each triplet is composed of (entity, relationship/attribute, entity).
S102, respectively preprocessing the physical model knowledge graph and the logic model knowledge graph to obtain a relationship triplet and an attribute triplet of the physical model knowledge graph and a relationship triplet and an attribute triplet of the logic model knowledge graph.
Optionally, in another embodiment of the present application, an implementation manner of step S102, as shown in fig. 2, includes:
s201, extracting the triples in the physical model knowledge graph, and performing normalization processing to obtain the relation triples and the attribute triples of the physical model knowledge graph.
S202, extracting the triples in the logic model knowledge graph, and performing normalization processing to obtain the relation triples and the attribute triples of the logic model knowledge graph.
S103, inputting the relation triplet and the attribute triplet of the physical model knowledge graph and the logic model knowledge graph into the multi-layer graph convolutional neural network to obtain the embedded representation of the entity in the knowledge graph.
Specifically, a relation triplet and an attribute triplet of the physical model knowledge graph and the logic model knowledge graph are used as input of a multi-layer graph convolutional neural network, and vector expression of entities in the knowledge graph is continuously updated to obtain embedded representation of the entities in the knowledge graph.
Entity e in a multi-layer graph convolutional neural network i Is expressed by the following formula:
wherein, the liquid crystal display device comprises a liquid crystal display device,
is entity e i Embedded representation in layer l GNN, e i Is a regularized constant, N i Is entity e i Is>
Representing entity e j Embedded representation in layer 1 GNN.
S104, calculating a cross-graph matching vector of the neighbor nodes of the entity for each entity.
Optionally, in another embodiment of the present application, an implementation of step S104, as shown in fig. 3, includes:
s301, calculating entity sampling probability of the entity neighbor nodes for each entity.
Specifically, for entity set E 1 Entity e in (3) i Is sub-sampled with the neighbors of the entity set E 2 The neighbor sampling subgraphs of each entity are compared, and the entity sampling is performed once. The calculation function of the entity sampling probability is expressed by the following formula:
wherein p (h i_j |h i ) Is entity e i Neighbor e of (2) i_j N, N i Is node e i Index set of adjacent nodes, h i And h i-j Is the learned entity vector expression, W S Is a matrix of parameters that are to be shared,
is h i-j Transpose of the entity vector expression, +.>
Entity vector h of yes i-k Transpose of expression.
S302, comparing the sub-graph of the entity with all sub-graphs in another knowledge graph, and selecting the most similar sub-graph as a target sub-graph.
S303, determining a cross-graph matching vector of the entity according to the target subgraph, the attention weight and the embedded representation of the entity in the knowledge graph.
Specifically, all subgraphs are taken as input, and the corresponding matching vector is calculated for each adjacent point, (e) i ,e ik ) Are entity pairs to be evaluated, respectively belong to an entity set E 1 And entity set E 2 P and q are each e i Sum e of ik And calculating a cross-graph matching vector of the neighbor node based on the candidate subgraph corresponding to the candidate entity. The computation function of the cross-graph matching vector of node q is expressed by:
wherein a is pq Is a weight coefficient of the attention mechanism,
is c ik Is the neighbor sample, q' and q, m p Is the cross-graph matching vector of node p, which is the measure h p Differences from their nearest neighbors in other subgraphs, h p And h q Is the GCN output embedded representation of the entity nodes p and q.
S105, determining target embedded representation of the entity according to the cross-graph matching vector of the neighbor node of the entity and the embedded representation of the entity.
Specifically, considering the result of GCN expression of the node p itself, the embedded representation of the node p is calculated by:
wherein, the liquid crystal display device comprises a liquid crystal display device,
is e i Is an embedded representation, h, of the neighbor node p after calculation p Is the GCN embedded representation of the previous neighbor node p itself, m p Is the cross-graph matching vector for node p.
S106, aggregating the entities in the knowledge graph to obtain the embedded representation of the optimized entities.
Optionally, in another embodiment of the present application, an implementation of step S106 includes:
updating the embedded representation of the central node by aggregating the neighbor node information of the central node, and aggregating all the associated node information to obtain the embedded representation of the optimized entity.
Specifically, each entity node of the knowledge graph is aggregated, the embedded representation of the center node is updated by aggregating the neighbor node information of the center node, and all the associated node information is aggregated. The optimized embedded representation of the compute farm entity is expressed by:
wherein, the liquid crystal display device comprises a liquid crystal display device,
as the central entity e i Optimized embedded representation, g i As the central entity e i The adjacent point representation, h, learned by the gate-controlled graph neural network i Representing a central entity e i The previous embedded representation.
S107, determining the distance between the two knowledge-graph entities according to the embedded representation of the entity in the knowledge-graph after optimization and the vectorized representation of the entity in the other knowledge-graph.
Specifically, the distance function of the computing entity is expressed by:
wherein d (r, t) is the distance between the logic model knowledge graph entity r and the physical model knowledge graph entity t,
is a vectorized representation of the logical model center entity r,/->
Is an optimized embedded representation of the physical model center entity t.
S108, according to the distance between the two knowledge graph entities, the alignment of the entities in the two graphs is realized.
Specifically, the entity alignment in the two maps is achieved by calculating an objective function, which is expressed by the following formula:
wherein L is w Is an objective function, L is a set of pre-aligned entity pairs, r and t are node pairs to be entity aligned,
is the integrated all adjacency information of the entity node r. />
Is the integrated all adjacency information of the entity node t.
As can be seen from the above solutions, the present application provides an entity alignment method: firstly, acquiring a physical model knowledge graph and a logic model knowledge graph; the knowledge graph is represented by a directed graph, and the directed graph comprises entities, relations and triples; the triples consist of entities, relationships/attributes, entities; then, respectively preprocessing the physical model knowledge graph and the logic model knowledge graph to obtain a relationship triplet and an attribute triplet of the physical model knowledge graph and a relationship triplet and an attribute triplet of the logic model knowledge graph; then, inputting a relation triplet and an attribute triplet of the physical model knowledge graph and the logic model knowledge graph into a multi-layer graph convolutional neural network to obtain an embedded representation of an entity in the knowledge graph; for each entity, calculating a cross-graph matching vector of neighbor nodes of the entity; determining a target embedded representation of the entity according to the cross-graph matching vector of the neighbor node of the entity and the embedded representation of the entity; then, the entities in the knowledge graph are aggregated to obtain an embedded representation of the optimized entities; determining the distance between two knowledge-graph entities according to the embedded representation of the entity in the knowledge-graph after optimization and the vectorized representation of the entity in the other knowledge-graph; finally, according to the distance between the two knowledge graph entities, the alignment of the entities in the two graphs is realized. Therefore, the entity alignment matching task between the two knowledge maps of the physical model knowledge map and the logic model knowledge map is efficiently completed.
Another embodiment of the present application provides an alignment device for an entity, as shown in fig. 4, specifically including:
an obtaining unit 401 is configured to obtain a physical model knowledge graph and a logical model knowledge graph.
The knowledge graph is represented by a directed graph, and the directed graph comprises entities, relations and triples; a triplet consists of entities, relationships/attributes, entities.
The preprocessing unit 402 is configured to perform preprocessing on the physical model knowledge graph and the logic model knowledge graph respectively, so as to obtain a relationship triplet and an attribute triplet of the physical model knowledge graph, and a relationship triplet and an attribute triplet of the logic model knowledge graph.
Optionally, in another embodiment of the present application, an implementation of the preprocessing unit 402 includes:
and the first preprocessing subunit is used for extracting the triples in the physical model knowledge graph and performing normalization processing to obtain the relation triples and the attribute triples of the physical model knowledge graph.
And the second preprocessing subunit is used for extracting the triples in the logic model knowledge graph and performing normalization processing to obtain the relation triples and the attribute triples of the logic model knowledge graph.
The specific working process of the unit disclosed in the foregoing embodiments of the present application may refer to the content of the corresponding method embodiment, as shown in fig. 2, which is not described herein again.
And the input unit 403 is configured to input the relationship triplet and the attribute triplet of the physical model knowledge graph and the logical model knowledge graph into the multi-layer graph convolutional neural network, so as to obtain an embedded representation of the entity in the knowledge graph.
A calculating unit 404, configured to calculate, for each entity, a cross-graph matching vector of neighboring nodes of the entity.
Optionally, in another embodiment of the present application, an implementation of the computing unit 404 includes:
and the first calculating subunit is used for calculating the entity sampling probability of the entity neighbor node for each entity.
And the selecting unit is used for comparing the sub-graph of the entity with all sub-graphs in another knowledge graph, and selecting the most similar sub-graph as the target sub-graph.
And the second calculation subunit is used for determining a cross-graph matching vector of the entity according to the target subgraph, the attention weight and the embedded representation of the entity in the knowledge graph.
The specific working process of the unit disclosed in the foregoing embodiments of the present application may refer to the content of the corresponding method embodiment, as shown in fig. 3, which is not described herein again.
A first determining unit 405, configured to determine a target embedded representation of the entity according to the cross-graph matching vector of the neighboring node of the entity and the embedded representation of the entity.
And the aggregation unit 406 is configured to aggregate the entities in the knowledge graph to obtain an embedded representation of the optimized entity.
Optionally, in another embodiment of the present application, an implementation of the aggregation unit 406 includes:
and the aggregation subunit is used for updating the embedded representation of the central node by aggregating the neighbor node information of the central node, and aggregating all the related node information to obtain the embedded representation of the optimized entity.
The specific working process of the unit disclosed in the foregoing embodiments of the present application may refer to the content of the corresponding method embodiment, as shown in fig. 3, which is not described herein again.
A second determining unit 407, configured to determine a distance between two knowledge-graph entities according to the embedded representation after the entity optimization in the knowledge-graph and the vectorized representation of the entity in the other knowledge-graph.
An alignment unit 408, configured to align entities in the two atlases according to the distance between the two knowledge-atlas entities.
The specific working process of the unit disclosed in the foregoing embodiments of the present application may refer to the content of the corresponding method embodiment, as shown in fig. 1, which is not repeated herein.
As can be seen from the above solutions, the present application provides an alignment device for an entity: first, the acquisition unit 401 acquires a physical model knowledge graph and a logical model knowledge graph; the knowledge graph is represented by a directed graph, and the directed graph comprises entities, relations and triples; the triples consist of entities, relationships/attributes, entities; then, the preprocessing unit 402 performs preprocessing on the physical model knowledge graph and the logic model knowledge graph respectively to obtain a relationship triplet and an attribute triplet of the physical model knowledge graph and a relationship triplet and an attribute triplet of the logic model knowledge graph; then, the input unit 403 inputs the relationship triplet and the attribute triplet of the physical model knowledge graph and the logic model knowledge graph into the multi-layer graph convolutional neural network, so as to obtain an embedded representation of the entity in the knowledge graph; the calculating unit 404 calculates, for each entity, a cross-graph matching vector of neighbor nodes of the entity; the first determining unit 405 determines a target embedded representation of the entity according to the cross-graph matching vector of the neighbor node of the entity and the embedded representation of the entity; the aggregation unit 406 aggregates the entities in the knowledge graph to obtain an embedded representation of the optimized entities; the second determining unit 407 determines the distance between the two knowledge-graph entities according to the embedded representation of the entity in the knowledge-graph after optimization and the vectorized representation of the entity in the other knowledge-graph; finally, the alignment unit 408 aligns the entities in the two atlases according to the distance between the two knowledge atlas entities. Therefore, the entity alignment matching task between the two knowledge maps of the physical model knowledge map and the logic model knowledge map is efficiently completed.
Another embodiment of the present application provides an electronic device, as shown in fig. 5, including:
one or more processors 501.
A storage device 502 on which one or more programs are stored.
The one or more programs, when executed by the one or more processors 501, cause the one or more processors 501 to implement the method of alignment of entities as set forth in any of the embodiments above.
Another embodiment of the present application provides a computer storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements a method of aligning entities as in any of the above embodiments.
In the above embodiments of the disclosure, it should be understood that the disclosed apparatus and method may be implemented in other manners. The apparatus and method embodiments described above are merely illustrative, for example, flow diagrams and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of apparatus, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
In addition, functional modules in various embodiments of the present disclosure may be integrated together to form a single portion, or each module may exist alone, or two or more modules may be integrated to form a single portion. The functions, if implemented in the form of software functional modules and sold or used as a stand-alone product, may be stored in a computer-readable storage medium. Based on such understanding, the technical solution of the present disclosure may be embodied in essence or a part contributing to the prior art or a part of the technical solution, or in the form of a software product stored in a storage medium, including several instructions for causing a computer device (which may be a personal computer, a live device, or a network device, etc.) to perform all or part of the steps of the method described in the embodiments of the present disclosure. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
Those skilled in the art can make or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (10)
1. A method of aligning entities, comprising:
acquiring a physical model knowledge graph and a logic model knowledge graph; the knowledge graph is represented by a directed graph, and the directed graph comprises entities, relations and triples; the triples consist of entities, relationships/attributes, entities;
preprocessing the physical model knowledge graph and the logic model knowledge graph respectively to obtain a relationship triplet and an attribute triplet of the physical model knowledge graph and a relationship triplet and an attribute triplet of the logic model knowledge graph;
inputting the relation triplets and the attribute triplets of the physical model knowledge graph and the logic model knowledge graph into a multi-layer graph convolutional neural network to obtain the embedded representation of the entity in the knowledge graph;
for each entity, calculating a cross-graph matching vector of neighbor nodes of the entity;
determining a target embedded representation of the entity according to the cross-graph matching vector of the neighbor node of the entity and the embedded representation of the entity;
aggregating the entities in the knowledge graph to obtain an embedded representation of the optimized entity;
determining the distance between two knowledge-graph entities according to the embedded representation of the entity in the knowledge-graph after optimization and the vectorized representation of the entity in the other knowledge-graph;
and according to the distance between the two knowledge graph entities, realizing the alignment of the entities in the two graphs.
2. The method of aligning an entity according to claim 1, wherein the preprocessing the physical model knowledge graph and the logical model knowledge graph to obtain a relationship triplet and an attribute triplet of the physical model knowledge graph and a relationship triplet and an attribute triplet of the logical model knowledge graph respectively includes:
extracting the triples in the physical model knowledge graph, and performing normalization processing to obtain a relation triplet and an attribute triplet of the physical model knowledge graph;
and extracting the triples in the logic model knowledge graph, and performing normalization processing to obtain the relation triples and the attribute triples of the logic model knowledge graph.
3. The method of aligning entities according to claim 1, wherein for each entity, calculating a cross-graph matching vector of neighboring nodes of the entity comprises:
calculating entity sampling probability of the entity neighbor node for each entity;
comparing the sub-graph of the entity with all sub-graphs in another knowledge graph, and selecting the most similar sub-graph as a target sub-graph;
and determining a cross-graph matching vector of the entity according to the target subgraph, the attention weight and the embedded representation of the entity in the knowledge graph.
4. The method for aligning entities according to claim 1, wherein the aggregating the entities in the knowledge graph to obtain the embedded representation of the optimized entity includes:
updating the embedded representation of the central node by aggregating the neighbor node information of the central node, and aggregating all the associated node information to obtain the embedded representation of the optimized entity.
5. An alignment device for an entity, comprising:
the acquisition unit is used for acquiring a physical model knowledge graph and a logic model knowledge graph; the knowledge graph is represented by a directed graph, and the directed graph comprises entities, relations and triples; the triples consist of entities, relationships/attributes, entities;
the preprocessing unit is used for respectively preprocessing the physical model knowledge graph and the logic model knowledge graph to obtain a relation triplet and an attribute triplet of the physical model knowledge graph and a relation triplet and an attribute triplet of the logic model knowledge graph;
the input unit is used for inputting the relation triplet and the attribute triplet of the physical model knowledge graph and the logic model knowledge graph into the multi-layer graph convolutional neural network to obtain the embedded representation of the entity in the knowledge graph;
a calculating unit, configured to calculate, for each entity, a cross-graph matching vector of a neighboring node of the entity;
a first determining unit, configured to determine a target embedded representation of the entity according to a cross-graph matching vector of a neighboring node of the entity and the embedded representation of the entity;
the aggregation unit is used for aggregating the entities in the knowledge graph to obtain an embedded representation of the optimized entity;
the second determining unit is used for determining the distance between the two knowledge-graph entities according to the embedded representation of the entity in the knowledge-graph after the optimization and the vectorized representation of the entity in the other knowledge-graph;
and the alignment unit is used for realizing the alignment of the entities in the two atlases according to the distance between the two knowledge atlas entities.
6. The alignment device of an entity of claim 5, wherein the preprocessing unit comprises:
the first preprocessing subunit is used for extracting the triples in the physical model knowledge graph and performing normalization processing to obtain the relation triples and the attribute triples of the physical model knowledge graph;
and the second preprocessing subunit is used for extracting the triples in the logic model knowledge graph and performing normalization processing to obtain the relation triples and the attribute triples of the logic model knowledge graph.
7. The alignment device of an entity of claim 5, wherein the computing unit comprises:
a first calculating subunit, configured to calculate, for each entity, an entity sampling probability of a neighbor node of the entity;
the selecting unit is used for comparing the sub-graph of the entity with all sub-graphs in another knowledge graph, and selecting the most similar sub-graph as a target sub-graph;
and the second calculating subunit is used for determining a cross-graph matching vector of the entity according to the target subgraph, the attention weight and the embedded representation of the entity in the knowledge graph.
8. The alignment device of an entity of claim 5, wherein the aggregation unit comprises:
and the aggregation subunit is used for updating the embedded representation of the central node by aggregating the neighbor node information of the central node, and aggregating all the related node information to obtain the embedded representation of the optimized entity.
9. An electronic device, comprising:
one or more processors;
a storage device having one or more programs stored thereon;
the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the method of alignment of entities of any of claims 1-4.
10. A computer storage medium, characterized in that it has stored thereon a computer program, wherein the computer program, when executed by a processor, implements the method of alignment of entities according to any of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211682651.4A CN116010615A (en) | 2022-12-27 | 2022-12-27 | Entity alignment method and device, electronic equipment and computer storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211682651.4A CN116010615A (en) | 2022-12-27 | 2022-12-27 | Entity alignment method and device, electronic equipment and computer storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116010615A true CN116010615A (en) | 2023-04-25 |
Family
ID=86018683
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211682651.4A Pending CN116010615A (en) | 2022-12-27 | 2022-12-27 | Entity alignment method and device, electronic equipment and computer storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116010615A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116610820A (en) * | 2023-07-21 | 2023-08-18 | 智慧眼科技股份有限公司 | Knowledge graph entity alignment method, device, equipment and storage medium |
-
2022
- 2022-12-27 CN CN202211682651.4A patent/CN116010615A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116610820A (en) * | 2023-07-21 | 2023-08-18 | 智慧眼科技股份有限公司 | Knowledge graph entity alignment method, device, equipment and storage medium |
| CN116610820B (en) * | 2023-07-21 | 2023-10-20 | 智慧眼科技股份有限公司 | Knowledge graph entity alignment method, device, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110532436B (en) | Cross-social network user identity recognition method based on community structure | |
| CN110837602B (en) | User recommendation method based on representation learning and multi-mode convolutional neural network | |
| CN112069398A (en) | Information pushing method and device based on graph network | |
| CN113536383B (en) | Method and device for training graph neural network based on privacy protection | |
| WO2021089013A1 (en) | Spatial graph convolutional network training method, electronic device and storage medium | |
| Vovan | An improved fuzzy time series forecasting model using variations of data | |
| CN111553215B (en) | Personnel association method and device, graph roll-up network training method and device | |
| CN111967972B (en) | Financial product recommendation method and device | |
| CN111339818B (en) | Face multi-attribute recognition system | |
| WO2019006976A1 (en) | Neural network weight discretizing method, system and device, and readable storage medium | |
| CN115018656B (en) | Risk identification method, and training method, device and equipment of risk identification model | |
| CN116010615A (en) | Entity alignment method and device, electronic equipment and computer storage medium | |
| Nguyen et al. | DPER: direct parameter estimation for randomly missing data | |
| CN111626311B (en) | Heterogeneous graph data processing method and device | |
| WO2022100607A1 (en) | Method for determining neural network structure and apparatus thereof | |
| CN113515519A (en) | Method, device and equipment for training graph structure estimation model and storage medium | |
| CN117312680A (en) | Resource recommendation method based on user-entity sub-graph comparison learning | |
| CN112270259A (en) | SAR image ship target rapid detection method based on lightweight convolutional neural network | |
| CN115238134A (en) | Method and apparatus for generating a graph vector representation of a graph data structure | |
| CN105760442A (en) | Image feature enhancing method based on database neighborhood relation | |
| CN110222196A (en) | Fishery knowledge mapping construction device, method and computer readable storage medium | |
| Khrulkov et al. | Approach to imputation multivariate missing data of urban buildings by chained equations based on geospatial information | |
| CN110598850A (en) | Method for determining neighbor function kernel parameters of self-organizing neural network and training method | |
| CN116523028B (en) | Image characterization model training method and device based on image space position | |
| CN116226674B (en) | Layout model training method, layout method and device for frame beams |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |