Attribute value extraction method and apparatus across data structures, device, and medium

Abstract

The present application is applicable to the technical field of big data, and particularly relates to an attribute value extraction method and apparatus across data structures, a device, and a medium. The method comprises: calculating, on the basis of attribute values of target data in a target dataset and feature values of graph data, first similarity values representing the grammatical similarities between nodes in the graph data and the target data, and determining first candidate nodes; matching existing attributes in the target data with first feature paths of the first candidate nodes to determine first candidate matching nodes; calculating, on the basis of feature vectors of the first candidate matching nodes and attribute vectors of the target data, second similarity values representing the semantic similarities between the first candidate matching nodes and the target data, and determining first matching nodes representing a same entity as the target data; and matching missing attributes in the target data with second feature paths of the first matching nodes to determine first candidate attribute values, sorting the first candidate attribute values, and determining batch attribute values corresponding to the missing attributes. The speed of attribute value extraction is improved.

Description

Attribute value extraction method, device and equipment across data structures and medium

[0001] The present application is based on the Chinese invention application No. 202410839018.4, filed on June 26, 2024, entitled "Attribute value extraction method, device and equipment across data structures and medium", and claims priority thereto. TECHNICAL FIELD

[0002] The present application is applicable to the field of big data technology, and particularly relates to an attribute value extraction method, device, equipment and medium across data structures. BACKGROUND

[0003] In the process of data integration and knowledge graph construction, attribute value extraction is a key task, which aims to locate and extract the attribute value of a specific entity from a knowledge graph to enrich and perfect the existing data set. The attribute value extraction method of the prior art mainly relies on pre-defined logical rules, which can be formulated by domain experts according to domain knowledge and experience to describe how to extract the corresponding attribute value from a specific path in the knowledge graph. However, the method based on pre-defined logical rules usually needs to rely on manual operation, and a large amount of time and effort is needed to discover and formulate rules. At the same time, due to the diversity and complexity of application scenarios, it is difficult to formulate logical rules that can cover all cases, and when the application demand or data changes slightly, it may be necessary to modify all original logical rules or previous attribute value extraction results, thereby increasing a large amount of workload and maintenance cost. Therefore, how to improve the speed of attribute value extraction has become a problem to be solved.

[0004] SUMMARY

[0005] Therefore, the embodiments of the present application provide an attribute value extraction method, device, equipment and medium across data structures to solve the problem of improving the speed of attribute value extraction.

[0006] In a first aspect, the embodiments of the present application provide an attribute value extraction method across data structures, which comprises:

[0007] obtaining graph data corresponding to a graph structure and a data set corresponding to a table structure, determining the attributes and corresponding attribute values of target data in the data set, and determining the feature values of each node in the graph data, wherein the attributes include existing attributes and missing attributes;

[0008] According to the attribute values of the target data and the feature values of each node in the graph data, a first similarity value representing the syntactic similarity between each node in the graph data and the target data is calculated, and a first candidate node in the graph data is determined according to the first similarity value;

[0009] For any first candidate node, a first feature path starting from the first candidate node is determined, the existing attribute in the target data is matched with the first feature path, and a first candidate node corresponding to a first feature path matched with the existing attribute in the target data is determined as a first candidate matching node;

[0010] For any first candidate matching node, a second similarity value representing semantic similarity between the first candidate matching node and the target data is calculated according to the feature vector of the first candidate matching node and the attribute vector of the target data, and a first matching node representing the same entity as the target data is determined according to the second similarity value.

[0011] For any matching node, a second feature path starting from the first matching node is determined, the missing attribute in the target data is matched with the second feature path, a feature value corresponding to a second path matched with the missing attribute in the target data is determined as a first candidate attribute value, all first candidate attribute values are sorted to obtain a first sorting result, and a batch attribute value corresponding to the missing attribute in the target data is determined according to the first sorting result.

[0012] In a second aspect, an attribute value extraction device across data structures is provided, and the attribute value extraction device comprises:

[0013] A first obtaining module is configured to obtain graph data corresponding to a graph structure and a data set corresponding to a table structure, determine an attribute and a corresponding attribute value of target data in the data set, and determine a feature value of each node in the graph data, wherein the attribute comprises an existing attribute and a missing attribute.

[0014] A first calculating module is configured to calculate a first similarity value representing syntactic similarity between each node in the graph data and the target data according to the attribute value of the target data and the feature value of each node in the graph data, and determine a first candidate node in the graph data according to the first similarity value.

[0015] A first matching module is configured to, for any first candidate node, determine a first feature path starting from the first candidate node, match the existing attribute in the target data with the first feature path, and determine a first candidate node corresponding to a first feature path matched with the existing attribute in the target data as a first candidate matching node.

[0016] The second computing module is configured to, for any first candidate matching node, compute a second similarity value representing semantic similarity between the first candidate matching node and the target data according to the feature vector of the first candidate matching node and the attribute vector of the target data, and determine the first matching node representing the same entity as the target data according to the second similarity value.

[0017] The second matching module is configured to, for any matching node, determine a second feature path starting from the first matching node, match the missing attribute in the target data with the second feature path, determine a feature value corresponding to the second path matched with the missing attribute in the target data as a first candidate attribute value, sort all the first candidate attribute values to obtain a first sorting result, and determine a batch attribute value corresponding to the missing attribute in the target data according to the first sorting result.

[0018] In a third aspect, an embodiment of the present application provides a computer device, which comprises a processor, a memory, and a computer program stored in the memory and executable on the processor, and the processor implements the attribute value extraction method across data structures when executing the computer program.

[0019] In a fourth aspect, an embodiment of the present application provides a computer readable storage medium, which stores a computer program, and the computer program is executable on a processor to implement the attribute value extraction method across data structures.

[0020] Compared with the prior art, the application has the beneficial effects that: the application calculates a first similarity value representing the syntactic similarity between each node in the graph data and the target data according to the attribute value of the target data in the target data set and the characteristic value of the graph data, determines a first candidate node according to the first similarity value, matches the existing attribute in the target data with a first feature path corresponding to the first candidate node to obtain a first candidate matching node, calculates a second similarity value representing the semantic similarity between the first candidate matching node and the target data according to the feature vector of the first candidate matching node and the attribute vector of the target data, determines a first matching node representing the same entity as the target data according to the second similarity value, matches the missing attribute in the target data with a second feature path corresponding to the first matching node to obtain a first candidate attribute value, and sorts all the first candidate attribute values to determine a batch of attribute values matched with the missing attribute. The first candidate node with syntactic similarity to the target data is screened out through the first similarity value, the first candidate node is screened through the existing attribute, the first matching node with both syntactic similarity and semantic similarity to the target data is screened out through the second similarity value, the automatic matching of the attribute and the path is realized, the matching cost is reduced, the speed and flexibility of screening the first matching node are improved, the accuracy of the screened first matching node is also improved, the missing attribute is matched with the second feature path corresponding to the first matching node to obtain the first candidate attribute value, and the batch of attribute values corresponding to the missing attribute is obtained through sorting, the speed and flexibility of extracting the attribute value of the missing attribute are improved, and the accuracy of the extracted attribute value is also improved. BRIEF DESCRIPTION OF DRAWINGS

[0021] In order to more clearly illustrate the technical solutions of the embodiments of the application, the following will briefly introduce the drawings needed to be used in the description of the embodiments of the application. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0022] FIG. 1 is a schematic diagram of an application environment of a cross-data-structure attribute value extraction method according to an embodiment of the application;

[0023] FIG. 2 is a flowchart of a cross-data-structure attribute value extraction method according to an embodiment of the application;

[0024] FIG. 3 is a schematic diagram of a table structure corresponding data set according to an embodiment of the application;

[0025] FIG. 4 is a schematic diagram of a graph structure corresponding graph data according to an embodiment of the application;

[0026] FIG. 5 is a flow diagram of a method for extracting attribute values across data structures according to an embodiment of the present application;

[0027] FIG. 6 is a flow diagram of a method for extracting attribute values across data structures according to an embodiment of the present application;

[0028] FIG. 7 is a flow diagram of a method for extracting attribute values across data structures according to an embodiment of the present application;

[0029] FIG. 8 is a flow diagram of a method for extracting attribute values across data structures according to an embodiment of the present application;

[0030] FIG. 9 is a flow diagram of a method for extracting attribute values across data structures according to an embodiment of the present application;

[0031] FIG. 10 is a flow diagram of a method for extracting attribute values across data structures according to an embodiment of the present application;

[0032] FIG. 11 is a structural diagram of an apparatus for extracting attribute values across data structures according to an embodiment of the present application;

[0033] FIG. 12 is a structural diagram of a computer device according to an embodiment of the present application. DETAILED DESCRIPTION

[0034] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of the present application.

[0035] To illustrate the technical solutions of the present application, the following will be described by specific embodiments.

[0036] The method for extracting attribute values across data structures provided by the embodiment one of the present application can be applied in the application environment as shown in FIG. 1, in which a server and a client communicate, the server provides the service of extracting attribute values across data structures, and the client triggers the attribute value extraction task to the server. The client includes, but is not limited to, a palm computer, a desktop computer, a notebook computer, an ultra-mobile personal computer (UMPC), a netbook, a cloud computer device, a personal digital assistant (PDA), and the like. The computer device corresponding to the server can be implemented by an independent server or a server cluster composed of multiple servers.

[0037] Referring to FIG. 2, it is a flow diagram of a cross-data-structure attribute value extraction method provided in Embodiment Two of the present application. The cross-data-structure attribute value extraction method is applied to the server in FIG. 1. The server connects to the client to obtain the graph data corresponding to the graph structure and the data set corresponding to the table structure sent by the client. As shown in FIG. 2, the attribute value extraction method can include the following steps:

[0038] In step S201, the graph data corresponding to the graph structure and the data set corresponding to the table structure are obtained, the attributes and corresponding attribute values of the target data in the data set are determined, and the feature values of each node in the graph data are determined.

[0039] In the present application, the data set can refer to a collection of text type data of the table structure. The data set includes data records of multiple record data information. Each data record is composed of a single or multiple attributes and corresponding attribute values. The attribute can refer to a variable or field describing the characteristics of the data record. The attribute value can refer to the specific numerical value of the corresponding attribute. The attribute can include existing attributes and missing attributes. The existing attribute can refer to an attribute with a corresponding attribute value. The missing attribute can refer to an attribute without a corresponding attribute value. The target data can refer to a data record containing a missing attribute. The graph data can refer to a knowledge graph of a graph structure composed of nodes, edges, node features, and edge relationship types. The node can represent an entity or object in the knowledge graph. The edge can represent the relationship between the nodes. The node feature can refer to the data associated with the node (entity) and used to represent the characteristics or state of the node. The node feature value is the specific numerical value of the corresponding node feature.

[0040] As shown in FIG. 3, it is a schematic diagram of a data set corresponding to a table structure provided in Embodiment Two of the present application. The target data set includes three target data. Tid refers to the serial number of each target data, which is t1, t2, and t3 in sequence. Eid refers to the serial number of the entity represented by each target data. As shown in the diagram, three data records correspond to two entities, i.e., t1 and t3 correspond to the same entity e1, and t2 corresponds to entity e2. Each target data includes six attributes, i.e., name, gender, email, address, age, and spouse name. The existing attributes of each target data include name, gender, email, and address. The missing attributes include age and spouse name.

[0041] As shown in FIG. 4, it is a schematic diagram of graph data corresponding to a graph structure provided in Embodiment Two of the present application. The graph data is a knowledge graph composed of nodes, edges, node features, and edge relationship types. The knowledge graph includes three nodes, i.e., v1 and v2, v3. Each node represents an entity in the knowledge graph. For example, for node v3, the node includes three features, i.e., name, gender, and age. The feature values of the respective features are A3, female, and 24 in sequence.

[0042] In step S202, a first similarity value representing the syntactic similarity between each node in the graph data and the target data is calculated according to the attribute value of the target data and the feature value of each node in the graph data, and a first candidate node in the graph data is determined according to the first similarity value.

[0043] In the embodiments of the present application, the syntactic similarity can refer to the similarity between the node in the graph data and the surface form or structure of the target data, the first similarity value is a score representing the syntactic similarity between the node in the graph data and the target data, and the first candidate node can refer to the node selected according to the first similarity value.

[0044] Specifically, the first similarity value representing the syntactic similarity between each node in the graph data and the target data can be calculated according to the Jaccard similarity coefficient, and all nodes in the graph data are selected according to the preset threshold and the first similarity value, and the node with the first similarity value higher than the preset threshold is determined as the first candidate node.

[0045] In step S203, for any first candidate node, a first feature path with the first candidate node as the starting point is determined, and the existing attributes in the target data are matched with the first feature path, and the first candidate node corresponding to the first feature path matching the existing attributes in the target data is determined as the first candidate matching node.

[0046] In the embodiments of the present application, the feature path can refer to a path composed of a node, a feature of the node and a corresponding feature value in the graph data, for example, as shown in FIG. 4, for the node v3, a feature path of the node v3 can refer to a feature path composed of the node v3, the feature “name” of the node and the corresponding feature value “A3”, which can be denoted as ρ=(v3, v14), the first feature path can refer to a path composed of the first candidate node, the feature of the first candidate node and the corresponding feature value, and the first candidate matching node can refer to the node selected according to the existing attributes in the target data.

[0047] Specifically, for any first candidate node, all first feature paths with the first candidate node as the starting point are determined, the existing attributes in the target data are matched with the first feature path, and if all the existing attributes in the target data can find a first feature path with the first candidate node as the starting point, the first candidate node is determined as the first candidate matching node.

[0048] In step S204, for any first candidate matching node, a second similarity value representing the semantic similarity between the first candidate matching node and the target data is calculated according to the feature vector of the first candidate matching node and the attribute vector of the target data, and a first matching node representing the same entity as the target data is determined according to the second similarity value.

[0049] In the embodiments of the present application, the feature vector can refer to a vector form of the feature value of the first candidate matching node, the attribute vector can refer to a vector form of the attribute value corresponding to the existing attribute in the target data, the semantic similarity can refer to the similarity between the first candidate matching node and the meaning or context of the target data, the second similarity value is a score representing the semantic similarity between the first candidate matching node and the target data, and the first matching node can refer to a node representing the same entity as the target data.

[0050] Specifically, the second similarity value of the semantic similarity between the first candidate matching node and the target data can be calculated according to the cosine similarity, the second similarity values of all first candidate matching nodes are sorted, and the first candidate matching node with the largest second similarity value is determined as the first matching node.

[0051] In step S205, for any matching node, a second feature path starting from the first matching node is determined, the missing attribute in the target data is matched with the second feature path, the feature value corresponding to the second path matched with the missing attribute in the target data is determined as the first candidate attribute value, all first candidate attribute values are sorted to obtain a first sorting result, and the batch attribute value corresponding to the missing attribute in the target data is determined according to the first sorting result.

[0052] In the embodiments of the present application, the second feature path can refer to a path composed of the first matching node, the feature of the first matching node and the corresponding feature value, the first candidate attribute value can refer to the feature value of the node in the graph data filtered according to the missing attribute in the target data, the first sorting result can refer to the result of sorting all first candidate attribute values, and the batch attribute value can refer to the attribute value corresponding to the missing attribute in the target data.

[0053] Specifically, for any first matching node, all second feature paths starting from the first matching node are determined, the missing attribute in the target data is matched with the second feature path, if the missing attribute in the target data can find a second feature path starting from the first matching node, the feature value represented by the last node in the second feature path is determined as the first candidate attribute value. Since each target data can have multiple first matching nodes, and each first matching node can have multiple second feature paths, for any missing attribute in the target data, after matching with the second feature path, there are multiple first candidate attribute values. The all first candidate attribute values corresponding to the missing attribute can be sorted by a pre-set sorting model, and the highest ranked attribute value is selected as the batch attribute value corresponding to the missing attribute.

[0054] For example, for the target data t2 in FIG. 3, if the vertex v2 in FIG. 4 represents the same entity as the target data, and for the missing attribute "age", a unique second feature path matching "age" can be found in FIG. 4, i.e. p = (v2, v13), then the attribute value corresponding to the missing attribute "age" is "24".

[0055] Optionally, after obtaining the first candidate attribute value corresponding to the missing attribute in the target data, all the first candidate attribute values of the missing attribute are formed into a first candidate attribute value set corresponding to the missing attribute.

[0056] In the embodiments of the present application, the first candidate node with syntactic similarity to the target data is screened out by the first similarity value, the first candidate node is screened by the existing attribute, the first matching node with both syntactic similarity and semantic similarity to the target data is screened out by the second similarity value, the automatic matching of attributes and paths is realized, the matching cost is reduced, the speed and flexibility of screening the first matching node are improved, the accuracy of the screened first matching node is also improved, the missing attribute is matched with the second feature path corresponding to the first matching node to obtain the first candidate attribute value, the first candidate attribute value is sorted to obtain the batch attribute value corresponding to the missing attribute, the speed and flexibility of extracting the attribute value of the missing attribute are improved, and the accuracy of the extracted attribute value is also improved.

[0057] Referring to FIG. 5, a flowchart of a cross-data-structure attribute value extraction method provided by an embodiment of the present application is shown. As shown in FIG. 3, the step S202 of calculating the first similarity value representing the syntactic similarity between each node in the graph data and the target data according to the attribute value of the target data and the feature value of each node in the graph data, and determining the first candidate node in the graph data according to the first similarity value, can include the following steps:

[0058] Step S501: Forming an attribute value set by using the attribute value corresponding to each existing attribute in the target data.

[0059] Step S502: For any node in the graph data, determining a first subgraph data corresponding to the node, and forming a feature value set by using the feature value of each node in the first subgraph data.

[0060] Step S503: Calculating the intersection of the attribute value set and the feature value set, and calculating the union of the attribute value set and the feature value set.

[0061] Step S504: Taking the ratio of the intersection to the union as the first similarity value of the node, and determining the first candidate node according to the first similarity value of all nodes in the graph data and a preset threshold.

[0062] In the embodiments of the present application, the attribute value set can refer to a set of attribute values of all existing attributes in the target data, the first subgraph data can refer to graph data of any node, including the node and all neighbors of the node, and the feature value set can refer to a set of feature values of all nodes in the first subgraph data.

[0063] For example, attribute values corresponding to all existing attributes in the target data form an attribute value set, denoted as Set(t), and feature values of all nodes in the first subgraph data form a feature value set, denoted as Set(v), and the first similarity value is denoted as Jacc(t, v). The first similarity value of each node in the graph data with respect to the syntax similarity of the target data according to the Jaccard similarity coefficient can be expressed as According to the preset threshold and the first similarity value, all nodes in the graph data are filtered, and nodes with a first similarity value higher than the preset threshold are determined as first candidate nodes.

[0064] In the embodiments of the present application, the first similarity value of the attribute value set corresponding to the existing attributes and the feature value set corresponding to any node is calculated according to the Jaccard similarity coefficient, and the first candidate nodes are obtained according to the first similarity value. The first candidate nodes with high syntax similarity with the graph data are obtained through screening, and the accuracy of the first matching nodes obtained through screening is improved.

[0065] Referring to FIG. 6, a flowchart of a cross-data-structure attribute value extraction method according to an embodiment of the present application is shown. As shown in FIG. 6, the step S204 of calculating, for any first candidate matching node, a second similarity value representing the semantic similarity between the first candidate matching node and the target data according to the feature vector of the first candidate matching node and the attribute vector of the target data, and determining the first matching node representing the same entity as the target data according to the second similarity value can include the following steps:

[0066] In step S601, for any first candidate matching node, second subgraph data corresponding to the first candidate matching node is determined, and feature values of all nodes in the second subgraph data are vectorized to obtain a feature vector.

[0067] In step S602, attribute values corresponding to all existing attributes in the target data are vectorized to obtain an attribute vector.

[0068] In step S603, the cosine similarity value of the feature vector and the attribute vector is calculated, the cosine similarity value is taken as the second similarity value of the first candidate matching node, the second similarity values of all first candidate matching nodes are sorted, and the first matching node representing the same entity as the target data is determined.

[0069] In the embodiments of the present application, the second sub-graph data can refer to graph data of any first candidate matching node and all neighbors of the first candidate matching node, the feature vector can refer to a vector form of feature values of all nodes in the second sub-graph data, and the attribute vector can refer to a vector form of attribute values of all existing attributes in the target data.

[0070] For example, the feature values of all nodes in the second sub-graph data are vectorized to obtain a feature vector, denoted as e w The attribute values corresponding to all existing attributes in the target data are vectorized to obtain an attribute vector, denoted as e t The second similarity value is denoted as sem(t, v), and the second similarity value of the feature vector and the attribute vector is calculated according to the cosine similarity, which can be expressed as sem(t, v) = max cos(e w , e t The second similarity values of all first candidate matching nodes are sorted, and the first candidate matching node with the largest second similarity value is determined as the first matching node.

[0071] In the embodiments of the present application, the second similarity value of the attribute vector corresponding to the existing attribute and the feature vector corresponding to any first candidate matching node is calculated according to the cosine similarity, and the optimal first matching node is obtained according to the second similarity value. On the basis of syntactic similarity, semantic similarity is screened, and on the basis of improving the speed and flexibility of screening the first matching node, the accuracy of the screened first matching node is also improved.

[0072] Referring to FIG. 7, it is a flowchart of a cross-data-structure attribute value extraction method according to an embodiment of the present application, as shown in FIG. 7, the attribute value extraction method can further include the following steps:

[0073] In step S701, an inverted index is obtained according to all nodes in the graph data and feature edges in all node feature paths.

[0074] In step S702, a matching data structure and a candidate data structure are obtained.

[0075] In step S703, after adding or deleting a target feature edge in the graph data to obtain updated graph data, an associated node is determined according to the target feature edge and the inverted index, and an associated feature path starting from the associated node is determined according to the associated node.

[0076] In the embodiments of the present application, the inverted index can refer to an index structure pre-constructed for storing the mapping relationship of all nodes and the feature edges corresponding to the nodes in the graph data, which can include key values and arrays corresponding to the key values. The key values can be used to store the inverted index items of the nodes, and the arrays corresponding to the key values can be used to store the feature edges corresponding to the nodes in the inverted index items. The candidate data structure can refer to a mapping table pre-constructed for recording the mapping relationship of all target data and the corresponding first matching nodes. The matching data structure can refer to a mapping table pre-constructed for recording the mapping relationship of all target data and the corresponding first candidate nodes. The feature edge can refer to a feature edge in a feature path. Each feature edge can represent a feature of a node. For example, as shown in FIG. 4, the node v3 includes three feature edges, which are the edges corresponding to the name, age and gender. The target feature edge can refer to a feature edge to be added or deleted in the graph data. The associated feature path can refer to a feature path containing the target feature edge. The associated node can refer to the first node in the associated feature path.

[0077] Specifically, after adding or deleting the target feature edge in the graph data to obtain updated graph data, the inverted index is queried with the target feature edge as an index item, the node corresponding to the target feature edge is determined as the associated node, and the feature path with the associated node as a starting point is determined as the associated feature path.

[0078] In step S704, when performing the missing attribute matching on the associated feature path corresponding to the associated node, it is determined that the target data corresponding to the associated node is found in the matching data structure as the first to-be-modified data, and the incremental attribute value corresponding to the missing attribute in the first to-be-modified data is determined according to the associated feature path.

[0079] In the embodiments of the present application, the first to-be-modified data can refer to the target data with the first matching node as the associated node, and the incremental attribute value can refer to the attribute value corresponding to the missing attribute in the first to-be-modified data.

[0080] Because the first matching node representing the same entity as the target data is calculated in steps S201 to S204, the result is determined according to the existing attributes of the target data and is irrelevant to the missing attribute. Therefore, when the associated feature path corresponding to the associated node is added or deleted to obtain updated graph data, the missing attribute matching on the associated feature path will not affect the result of the first matching node corresponding to the first to-be-modified data obtained previously.

[0081] Specifically, after the target feature edge is added or deleted in the graph data to obtain updated graph data, when performing the missing attribute matching on the association feature path corresponding to the association node, the association node is taken as a query condition to query a matching data structure, the target data corresponding to the association node is determined as the first to-be-modified data, and the incremental attribute value corresponding to the missing attribute in the first to-be-modified data is determined according to the association feature path.

[0082] In step S705, when performing the existing attribute matching on the association feature path corresponding to the association node, it is determined that the target data corresponding to the association node is found in the candidate data structure as the second to-be-modified data, the attribute value corresponding to the missing attribute in the second to-be-modified data is recalculated according to the updated graph data, and the incremental attribute value corresponding to the missing attribute in the second to-be-modified data is obtained.

[0083] In the embodiment of the application, the second to-be-modified data can refer to the target data of the first candidate node as the association node, and the incremental attribute value can refer to the attribute value corresponding to the missing attribute in the second to-be-modified data.

[0084] Because the first matching node representing the same entity as the target data is calculated in steps S201 to S204, and the first matching node is determined according to the existing attribute of the target data, when the existing attribute matching is performed on the association feature path after the target feature edge is added or deleted in the graph data to obtain updated graph data, the result of the first matching node corresponding to the second to-be-modified data obtained by the previous matching is affected.

[0085] Specifically, after the target feature edge is added or deleted in the graph data to obtain updated graph data, the association node is taken as a query condition to query a candidate data structure, the target data corresponding to the association node is determined as the second to-be-modified data, and the incremental attribute value corresponding to the missing attribute in the second to-be-modified data is obtained by recalculating the attribute value corresponding to the missing attribute in the second to-be-modified data according to the updated graph data.

[0086] In the embodiment of the application, by maintaining the three auxiliary structures of the inverted index, the candidate data structure, and the matching data structure, after the target feature edge is added or deleted in the graph data to obtain updated graph data, it is not necessary to recalculate the batch attribute values corresponding to the missing attributes extracted from all target data, but according to the auxiliary data structure, the target data affected by the update in the data set is located, and according to the matched attribute type, only these targets are updated accordingly, the data amount of the target data to be processed in the update process is reduced, thereby the workload is reduced and the update speed is improved.

[0087] Referring to FIG. 8, a flowchart of a method for extracting attribute values across data structures according to an embodiment of the present application is shown. As shown in FIG. 8, after the graph data is updated by adding or deleting the target feature edge in step S703, the method for extracting attribute values can further include the following steps:

[0088] In step S801, if the target feature edge is added in the graph data, the target feature edge is written into a blank key value of the inverted index, and the associated node is written into an array corresponding to the key value.

[0089] In step S802, if the target feature edge is deleted in the graph data, the key value and data corresponding to the target feature edge are removed from the inverted index.

[0090] In the embodiment of the present application, after the graph data is updated by adding or deleting the target feature edge, if the target feature edge is added in the graph data, the target feature edge is written into a blank key value of the inverted index, and the associated node is written into an array corresponding to the key value. If the target feature edge is deleted in the graph data, the key value and data corresponding to the target feature edge are removed from the inverted index, so as to update the inverted index.

[0091] In the embodiment of the present application, after the graph data is updated by adding or deleting the target feature edge, the inverted index is updated. By maintaining the inverted index, when the node corresponding to the feature edge needs to be queried, the inverted index can be directly obtained without traversing the entire graph data, thereby improving the query efficiency. In addition, when the feature edge changes, all associated nodes can be updated in time, the consistency of the graph data is maintained, and accurate results can be obtained when queried or analyzed.

[0092] Referring to FIG. 9, a flowchart of a method for extracting attribute values across data structures according to an embodiment of the present application is shown. As shown in FIG. 9, in step S704, the incremental attribute value corresponding to the missing attribute in the first to-be-modified data can be determined according to the associated feature path, and the method can include the following steps:

[0093] In step S901, the feature value of the last node in the associated feature path is obtained, and the first candidate attribute value corresponding to the first to-be-modified data is updated according to the feature value of the last node in the associated feature path, to obtain a second candidate attribute value.

[0094] In step S902, the second candidate attribute value is sorted to obtain a second sorting result, and the incremental attribute value corresponding to the missing attribute in the first to-be-modified data is determined according to the second sorting result.

[0095] In the embodiments of the present application, the first candidate attribute value can refer to a feature value filtered according to the missing attribute in the first to-be-modified data, the second candidate attribute value can refer to a feature value filtered according to the missing attribute in the updated first to-be-modified data, and the second sorting result can refer to a result of sorting all the second candidate attribute values.

[0096] Specifically, if the target feature edge is added in the graph data, the feature value of the last node in the associated feature path is added to the first candidate attribute value set corresponding to the first to-be-modified data to obtain an updated first candidate attribute value set. If the target feature edge is deleted in the graph data, the feature value of the last node in the associated feature path is deleted from the first candidate attribute value set corresponding to the first to-be-modified data to obtain an updated first candidate attribute value set. All the second candidate attribute values corresponding to the missing attribute in the first to-be-modified data are re-sorted according to a preset sorting model, and the attribute value with the highest ranking is selected as the incremental attribute value corresponding to the missing attribute in the first to-be-modified data.

[0097] In the embodiments of the present application, after the target feature edge is added or deleted in the graph data to obtain updated graph data, if the associated feature path is matched for the missing attribute, the first candidate attribute value corresponding to the missing attribute in the first to-be-modified data is updated, and the updated candidate attribute value is re-sorted to obtain the incremental attribute value corresponding to the missing attribute in the first to-be-modified data. Therefore, after the target feature edge is added or deleted in the graph data to obtain updated graph data, when the associated feature path corresponding to the associated node is matched for the missing attribute, it is not necessary to re-calculate the batch attribute values corresponding to the missing attribute extracted from all the target data, but to locate the affected first to-be-modified data according to the associated node, the associated feature path and the matching data structure, and only to quickly recalculate the first to-be-modified data, thereby reducing the amount of data to be processed in the updating process, and thus reducing the workload and improving the updating speed.

[0098] Referring to FIG. 10, FIG. 10 is a flowchart of a method for extracting attribute values across data structures according to an eighth embodiment of the present application. As shown in FIG. 10, the step S705 of recalculating the attribute values corresponding to the missing attribute in the second to-be-modified data according to the updated graph data to obtain the incremental attribute values corresponding to the missing attribute in the second to-be-modified data can include the following steps:

[0099] In step S1001, the updated graph data is obtained, the feature values of each node in the updated graph data are determined, and the attributes and corresponding attribute values of the second to-be-modified data are determined.

[0100] In step S1002, a third similarity value representing the syntactic similarity between each node in the updated graph data and the second to-be-modified data is calculated according to the attribute value of the second to-be-modified data and the feature value of each node in the updated graph data, and a second candidate node in the updated graph data is determined according to the third similarity value.

[0101] In step S1003, the first candidate node corresponding to the second to-be-modified data in the candidate data structure is updated according to the second candidate node, and an updated candidate data structure is obtained.

[0102] In the embodiments of the present application, the third similarity value can be a score representing the syntactic similarity between the node in the updated graph data and the second to-be-modified data, and the second candidate node can be a node selected according to the second similarity value.

[0103] Specifically, the process of obtaining the second candidate node according to the feature value of each node in the updated graph data and the attribute value of the second to-be-modified data can refer to the contents in steps S201 to S202 or steps S501 to S504. After obtaining the second candidate node corresponding to the second to-be-modified data, the first candidate node corresponding to the second to-be-modified data in the candidate data structure is updated, and an updated candidate data structure is obtained.

[0104] In step S1004, for any second candidate node, a third feature path starting from the second candidate node is determined, and the existing attribute in the second to-be-modified data is matched with the third feature path, and the second candidate node corresponding to the third feature path matched with the existing attribute in the second to-be-modified data is determined as a second candidate matching node.

[0105] In step S1005, for any second candidate matching node, a fourth similarity value representing the semantic similarity between the second candidate matching node and the second to-be-modified data is calculated according to the feature vector of the second candidate matching node and the attribute vector of the second to-be-modified data, and a second matching node representing the same entity as the second to-be-modified data is determined according to the fourth similarity value.

[0106] In step S1006, the first matching node corresponding to the second to-be-modified data in the matching data structure is updated according to the second matching node, and an updated matching data structure is obtained.

[0107] In the embodiments of the present application, the third feature path can be a path composed of the second candidate node, the feature of the second candidate node and the corresponding feature value, the second candidate matching node can be a node selected according to the existing attribute in the second to-be-modified data, the fourth similarity value can be a score representing the semantic similarity between the second candidate matching node and the second to-be-modified data, and the second matching node can be a node representing the same entity as the second to-be-modified data.

[0108] Specifically, the process of determining the second matching node representing the same entity as the second to-be-modified data can refer to the contents in steps S203 to S204 or steps S601 to S603. After obtaining the second matching node corresponding to the second to-be-modified data, the first matching node corresponding to the second to-be-modified data in the matching data structure is updated to obtain an updated matching data structure.

[0109] In step S1007, for any second matching node, a fourth feature path starting from the second matching node is determined, the missing attribute in the second to-be-modified data is matched with the fourth feature path, a feature value corresponding to the fourth feature path matched with the missing attribute in the second to-be-modified data is determined as a third candidate attribute value, all third candidate attribute values are sorted to obtain a third sorting result, and an incremental attribute value corresponding to the missing attribute in the second to-be-modified data is determined according to the third sorting result.

[0110] In the embodiments of the present application, the fourth feature path can refer to a path composed of the second matching node, the features of the second matching node, and the corresponding feature values, the third candidate attribute value can refer to a feature value filtered according to the missing attribute in the second to-be-modified data, and the third sorting result can refer to a result of sorting all third candidate attribute values.

[0111] Specifically, the process of determining the incremental attribute value corresponding to the missing attribute in the second to-be-modified data can refer to the contents in step S205.

[0112] After the target feature edge is added or deleted in the graph data to obtain the updated graph data, if the existing attribute matching is performed on the associated feature path, the second candidate node corresponding to the second to-be-modified data, the second matching node, and the incremental attribute value corresponding to the missing attribute in the second to-be-modified data are recalculated according to the updated graph data, the candidate data structure is updated according to the second candidate node, and the matching data structure is updated according to the second matching node. Therefore, after the target feature edge is added or deleted in the graph data to obtain the updated graph data, when the existing attribute matching is performed on the associated feature path corresponding to the associated node, it is not necessary to recalculate the batch attribute values corresponding to the missing attributes extracted from all target data, but to locate the affected second to-be-modified data according to the associated node, the associated feature path, and the candidate data structure, and only to quickly recalculate the second to-be-modified data, thereby reducing the amount of data to be processed in the updating process, reducing the workload, and improving the updating speed.

[0113] Corresponding to the attribute value extraction method across data structures of the above embodiment, Fig. 11 shows a structural block diagram of an attribute value extraction device across data structures provided by the embodiment nine of the present application, which is applied to the server in Fig. 1, and the server is connected with the client to obtain the graph data corresponding to the graph structure and the data set corresponding to the table structure sent by the client. For the convenience of description, only the part related to the present application is shown.

[0114] Referring to Fig. 11, the attribute value extraction device comprises:

[0115] The first acquisition module 1101 is configured to acquire the graph data corresponding to the graph structure and the data set corresponding to the table structure, determine the attributes and corresponding attribute values of the target data in the data set, and determine the feature values of each node in the graph data, wherein the attributes include existing attributes and missing attributes;

[0116] The first calculation module 1102 is configured to calculate the first similarity value representing the syntactic similarity between each node in the graph data and the target data according to the attribute values of the target data and the feature values of each node in the graph data, and determine the first candidate node in the graph data according to the first similarity value.

[0117] The first matching module 1103 is configured to determine the first feature path starting from the first candidate node for any first candidate node, match the existing attributes in the target data with the first feature path, and determine the first candidate node corresponding to the first feature path matched with the existing attributes in the target data as the first candidate matching node.

[0118] The second calculation module 1104 is configured to calculate the second similarity value representing the semantic similarity between the first candidate matching node and the target data according to the feature vector of the first candidate matching node and the attribute vector of the target data for any first candidate matching node, and determine the first matching node representing the same entity as the target data according to the second similarity value.

[0119] The second matching module 1105 is configured to determine the second feature path starting from the first matching node for any matching node, match the missing attributes in the target data with the second feature path, determine the feature value corresponding to the second path matched with the missing attributes in the target data as the first candidate attribute value, sort all the first candidate attribute values to obtain the first sorting result, and determine the batch attribute value corresponding to the missing attributes in the target data according to the first sorting result.

[0120] Optionally, the first calculation module 1102 comprises:

[0121] The first forming unit is configured to form an attribute value set by using attribute values corresponding to all existing attributes in the target data.

[0122] The second forming unit is configured to determine first subgraph data corresponding to any node in the graph data, and form a feature value set by using feature values of all nodes in the first subgraph data.

[0123] The third calculating unit is configured to calculate an intersection of the attribute value set and the feature value set, and calculate a union set of the attribute value set and the feature value set.

[0124] The first determining unit is configured to determine the first candidate node according to the first similarity value of all nodes in the graph data and a preset threshold.

[0125] Optionally, the second calculating module 1104 comprises:

[0126] The first vectorizing unit is configured to determine second subgraph data corresponding to any first candidate matching node, and vectorize feature values of all nodes in the second subgraph data to obtain a feature vector.

[0127] The second vectorizing unit is configured to vectorize attribute values corresponding to all existing attributes in the target data to obtain an attribute vector.

[0128] The second determining unit is configured to calculate a cosine similarity value of the feature vector and the attribute vector, take the cosine similarity value as a second similarity value of the first candidate matching node, sort second similarity values of all first candidate matching nodes, and determine the first matching node representing the same entity as the target data.

[0129] Optionally, the attribute value extraction apparatus further comprises:

[0130] The second obtaining module is configured to obtain an inverted index constructed according to all nodes in the graph data and feature edges in all node feature paths.

[0131] The third obtaining module is configured to obtain a matching data structure and a candidate data structure, wherein the matching data structure represents a mapping relationship between all target data and corresponding first matching nodes, and the candidate data structure represents a mapping relationship between all target data and corresponding first candidate nodes.

[0132] a third determining module, configured to determine, after a target feature edge is added or deleted in the graph data to obtain updated graph data, an associated node according to the target feature edge and the inverted index, and determine an associated feature path with the associated node as a starting point according to the associated node;

[0133] a first increment extraction module, configured to, when performing missing attribute matching on the associated feature path corresponding to the associated node, determine that target data corresponding to the associated node is first to-be-modified data in the matching data structure, and determine an increment attribute value corresponding to a missing attribute in the first to-be-modified data according to the associated feature path;

[0134] a second increment extraction module, configured to, when performing existing attribute matching on the associated feature path corresponding to the associated node, determine that target data corresponding to the associated node is second to-be-modified data in the candidate data structure, and recalculate an attribute value corresponding to a missing attribute in the second to-be-modified data according to the updated graph data to obtain an increment attribute value corresponding to the missing attribute in the second to-be-modified data.

[0135] Optionally, the attribute value extraction apparatus further includes:

[0136] a first updating module, configured to, if the target feature edge is added in the graph data, write the target feature edge into a blank key value in the inverted index, and write the associated node into an array corresponding to the key value;

[0137] a second updating module, configured to, if the target feature edge is deleted in the graph data, remove a key value and data corresponding to the target feature edge from the inverted index.

[0138] Optionally, the first increment extraction module includes:

[0139] a third updating unit, configured to obtain a feature value of a last node in the associated feature path, and update a first candidate attribute value corresponding to the first to-be-modified data according to the feature value of the last node in the associated feature path to obtain a second candidate attribute value;

[0140] a fourth determining unit, configured to sort the second candidate attribute value to obtain a second sorting result, and determine an increment attribute value corresponding to a missing attribute in the first to-be-modified data according to the second sorting result.

[0141] Optionally, the second increment extraction module includes:

[0142] a fourth obtaining unit, configured to obtain updated graph data, determine feature values of each node in the updated graph data, and determine attributes and corresponding attribute values of the second to-be-modified data.

[0143] a fourth computing unit, configured to calculate, according to the attribute value of the second to-be-modified data and the feature value of each node in the update graph data, a third similarity value representing the syntactic similarity between each node in the update graph data and the second to-be-modified data, and determine a second candidate node in the update graph data according to the third similarity value;

[0144] a fourth updating unit, configured to update, according to the second candidate node, the first candidate node corresponding to the second to-be-modified data in the candidate data structure, to obtain an updated candidate data structure;

[0145] a third matching unit, configured to, for any second candidate node, determine a third feature path starting from the second candidate node, match the existing attribute in the second to-be-modified data with the third feature path, and determine, as a second candidate matching node, the second candidate node corresponding to the third feature path that matches the existing attribute in the second to-be-modified data;

[0146] a fifth computing unit, configured to, for any second candidate matching node, calculate, according to the feature vector of the second candidate matching node and the attribute vector of the second to-be-modified data, a fourth similarity value representing the semantic similarity between the second candidate matching node and the second to-be-modified data, and determine, as a second matching node, the second candidate matching node that represents the same entity as the second to-be-modified data according to the fourth similarity value;

[0147] a fifth updating unit, configured to update, according to the second matching node, the first matching node corresponding to the second to-be-modified data in the matching data structure, to obtain an updated matching data structure;

[0148] the fourth matching unit is configured to, for any second matching node, determine a fourth feature path starting from the second matching node, match the missing attribute in the second to-be-modified data with the fourth feature path, determine, as a third candidate attribute value, the feature value corresponding to the fourth feature path that matches the missing attribute in the second to-be-modified data, sort all third candidate attribute values to obtain a third sorting result, and determine, according to the third sorting result, an incremental attribute value corresponding to the missing attribute in the second to-be-modified data.

[0149] In one embodiment, a computer device is provided, which can be a server, and an internal structure diagram of the computer device can be as shown in FIG. 12. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a readable storage medium and a database. The internal memory provides an environment for running the operating system and the readable storage medium in the non-volatile storage medium. The database of the computer device is configured to store user raw data. The network interface of the computer device is configured to communicate with an external terminal through a network connection. The readable storage medium is configured to be executed by the processor to implement a method for extracting attribute values across data structures.

[0150] In one embodiment, a computer device is provided, which includes a memory, a processor and a readable storage medium stored in the memory and executable on the processor. The processor is configured to execute the readable storage medium to implement the steps of the method for extracting attribute values across data structures in the above embodiments, such as the steps S201-S205 shown in FIG. 2, or the steps shown in FIGS. 5-10, which will not be repeated here. Alternatively, the processor is configured to execute the readable storage medium to implement the functions of the modules / units in the user data processing apparatus in this embodiment, such as the functions of the acquisition module 1101, the first calculation module 1102, the first matching module 1103, the second calculation module 1104 and the second matching module 1105 shown in FIG. 11, which will not be repeated here.

[0151] In one embodiment, one or more readable storage media storing computer readable instructions are provided. The computer readable instructions are executed by one or more processors to implement the steps of the method for extracting attribute values across data structures in the above embodiments, such as the steps S201-S205 shown in FIG. 2, or the steps shown in FIGS. 5-10, which will not be repeated here. Alternatively, the processor is configured to execute the readable storage medium to implement the functions of the modules / units in the user data processing apparatus in this embodiment, such as the functions of the acquisition module 1101, the first calculation module 1102, the first matching module 1103, the second calculation module 1104 and the second matching module 1105 shown in FIG. 11, which will not be repeated here. The readable storage medium in this embodiment includes a non-volatile readable storage medium and a volatile readable storage medium.

[0152] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing relevant hardware through a readable storage medium, and the readable storage medium can be stored in a non-volatile computer readable storage medium. When executed, the readable storage medium can include the processes of the above-mentioned embodiments of each method. Any reference to memory, storage, database or other medium used in each embodiment provided by the present application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0153] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the above-mentioned division of functional units and modules is exemplified, and in actual application, the above-mentioned functions can be completed by different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above.

[0154] The above-mentioned embodiments are only used to illustrate the technical solutions of the present application, but not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. A method for extracting attribute values ​​across data structures, wherein, The attribute value extraction method includes: Obtain the graph data corresponding to the graph structure and the dataset corresponding to the table structure, determine the attributes and corresponding attribute values ​​of the target data in the dataset, and determine the feature values ​​of each node in the graph data, wherein the attributes include existing attributes and missing attributes; Based on the attribute values ​​of the target data and the feature values ​​of each node in the graph data, a first similarity value is calculated to characterize the syntactic similarity between each node in the graph data and the target data. Based on the first similarity value, a first candidate node in the graph data is determined. For any first candidate node, a first feature path starting from the first candidate node is determined, and the existing attributes in the target data are matched with the first feature path. The first candidate node corresponding to the first feature path that matches the existing attributes in the target data is determined as the first candidate matching node. For any first candidate matching node, a second similarity value representing the semantic similarity between the first candidate matching node and the target data is calculated based on the feature vector of the first candidate matching node and the attribute vector of the target data. Based on the second similarity value, a first matching node representing the same entity as the target data is determined. For any matching node, a second feature path starting from the first matching node is determined. The missing attributes in the target data are matched with the second feature path. The feature value corresponding to the second path that matches the missing attributes in the target data is determined as the first candidate attribute value. All the first candidate attribute values ​​are sorted to obtain a first sorting result. Based on the first sorting result, the batch attribute values ​​corresponding to the missing attributes in the target data are determined.

2. The method for extracting attribute values ​​across data structures according to claim 1, wherein, The step of calculating a first similarity value, representing the syntactic similarity between each node in the graph data and the target data, based on the attribute values ​​of the target data and the feature values ​​of each node in the graph data, and determining a first candidate node in the graph data based on the first similarity value, includes: The attribute values ​​corresponding to all existing attributes in the target data are formed into an attribute value set; For any node in the graph data, determine the first subgraph data corresponding to the node, and form a feature value set by taking the feature values ​​of all nodes in the first subgraph data; Calculate the intersection of the attribute value set and the feature value set, and calculate the union of the attribute value set and the feature value set; The ratio of the intersection to the union is used as the first similarity value of the node. Based on the first similarity values ​​of all nodes in the graph data and a preset threshold, the first candidate node is determined.

3. The method for extracting attribute values ​​across data structures according to claim 1, wherein, For any first candidate matching node, calculating a second similarity value representing the semantic similarity between the first candidate matching node and the target data based on the feature vector of the first candidate matching node and the attribute vector of the target data, and determining a first matching node representing the same entity as the target data based on the second similarity value, includes: For any first candidate matching node, determine the second subgraph data corresponding to the first candidate matching node, and vectorize the feature values ​​of all nodes in the second subgraph data to obtain the feature vector. Vectorize the attribute values ​​corresponding to all existing attributes in the target data to obtain attribute vectors; Calculate the cosine similarity value between the feature vector and the attribute vector, use the cosine similarity value as the second similarity value of the first candidate matching node, sort the second similarity values ​​of all first candidate matching nodes, and determine the first matching node that represents the same entity as the target data.

4. The method for extracting attribute values ​​across data structures according to claim 1, wherein, The attribute value extraction method further includes: Obtain the inverted index constructed based on all nodes and feature edges in the feature paths of all nodes in the graph data; Obtain the matching data structure and candidate data structures, wherein the matching data structure represents all target data and their... The mapping relationship of the corresponding first matching node, the candidate data structure represents the mapping relationship between all target data and their corresponding first candidate nodes; After adding or deleting target feature edges in the graph data to obtain updated graph data, the associated nodes are determined based on the target feature edges and the inverted index, and the associated feature path starting from the associated node is determined based on the associated nodes. When performing missing attribute matching on the associated feature path corresponding to the associated node, the target data corresponding to the associated node found in the matching data structure is determined as the first data to be modified. Based on the associated feature path, the incremental attribute value corresponding to the missing attribute in the first data to be modified is determined. When performing existing attribute matching on the associated feature path corresponding to the associated node, the target data corresponding to the associated node found in the candidate data structure is determined to be the second data to be modified. The attribute values ​​corresponding to the missing attributes in the second data to be modified are recalculated according to the updated graph data to obtain the incremental attribute values ​​corresponding to the missing attributes in the second data to be modified.

5. The method for extracting attribute values ​​across data structures according to claim 4, wherein, After adding or deleting target feature edges in the graph data to obtain updated graph data, the method further includes: If the target feature edge is added to the graph data, the target feature edge is written into a blank key value of the inverted index, and the associated node is written into the array corresponding to the key value; If the target feature edge is deleted from the graph data, then the key value and data corresponding to the target feature edge are removed from the inverted index.

6. The attribute value extraction method across data structures according to claim 4, wherein, The step of determining the incremental attribute value corresponding to the missing attribute in the first data to be modified based on the associated feature path includes: Obtain the feature value of the last node in the associated feature path, and update the first candidate attribute value corresponding to the first data to be modified based on the feature value of the last node in the associated feature path to obtain the second candidate attribute value. The second candidate attribute values ​​are sorted to obtain a second sorting result. Based on the second sorting result, the incremental attribute values ​​corresponding to the missing attributes in the first data to be modified are determined.

7. The method for extracting attribute values ​​across data structures according to claim 4, wherein, The step of recalculating the attribute values ​​corresponding to the missing attributes in the second data to be modified based on the updated graph data to obtain the incremental attribute values ​​corresponding to the missing attributes in the second data to be modified includes: Obtain updated graph data, determine the feature value of each node in the updated graph data, and determine the attributes and corresponding attribute values ​​of the second data to be modified; Based on the attribute values ​​of the second data to be modified and the feature values ​​of each node in the updated graph data, a third similarity value is calculated to characterize the syntactic similarity between each node in the updated graph data and the second data to be modified. Based on the third similarity value, a second candidate node in the updated graph data is determined. Based on the second candidate node, the first candidate node corresponding to the second data to be modified in the candidate data structure is updated to obtain the updated candidate data structure; For any second candidate node, a third feature path starting from the second candidate node is determined. The existing attributes in the second data to be modified are matched with the third feature path, and the second candidate node corresponding to the third feature path that matches the existing attributes in the second data to be modified is determined as the second candidate matching node. For any second candidate matching node, a fourth similarity value representing the semantic similarity between the second candidate matching node and the second data to be modified is calculated based on the feature vector of the second candidate matching node and the attribute vector of the second data to be modified. Based on the fourth similarity value, a second matching node representing the same entity as the second data to be modified is determined. Based on the second matching node, the first matching node corresponding to the second data to be modified in the matching data structure is updated to obtain the updated matching data structure; For any second matching node, determine a fourth feature path starting from the second matching node, and then... The missing attributes in the second data to be modified are matched with the fourth feature path. The feature value corresponding to the fourth feature path that matches the missing attributes in the second data to be modified is determined as the third candidate attribute value. All the third candidate attribute values ​​are sorted to obtain the third sorting result. Based on the third sorting result, the incremental attribute value corresponding to the missing attributes in the second data to be modified is determined.

8. An attribute value extraction device across data structures, wherein, The attribute value extraction device includes: The first acquisition module is used to acquire graph data corresponding to the graph structure and dataset corresponding to the table structure, determine the attributes and corresponding attribute values ​​of the target data in the dataset, and determine the feature value of each node in the graph data, wherein the attributes include existing attributes and missing attributes; The first calculation module is used to calculate a first similarity value representing the syntactic similarity between each node in the graph data and the target data based on the attribute values ​​of the target data and the feature values ​​of each node in the graph data, and to determine a first candidate node in the graph data based on the first similarity value. The first matching module is used to determine a first feature path starting from any first candidate node, match the existing attributes in the target data with the first feature path, and determine the first candidate node corresponding to the first feature path that matches the existing attributes in the target data as the first candidate matching node. The second calculation module is used to calculate a second similarity value representing the semantic similarity between the first candidate matching node and the target data based on the feature vector of the first candidate matching node and the attribute vector of the target data for any first candidate matching node, and to determine a first matching node representing the same entity as the target data based on the second similarity value. The second matching module is used to determine a second feature path starting from the first matching node for any matching node, match the missing attributes in the target data with the second feature path, determine the feature value corresponding to the second path that matches the missing attributes in the target data as the first candidate attribute value, sort all the first candidate attribute values ​​to obtain a first sorting result, and determine the batch attribute value corresponding to the missing attributes in the target data based on the first sorting result.

9. A computer device comprising a memory, a processor, and a readable storage medium stored in the memory and operable on the processor, wherein, When the processor executes the readable storage medium, it performs the following steps: Obtain the graph data corresponding to the graph structure and the dataset corresponding to the table structure, determine the attributes and corresponding attribute values ​​of the target data in the dataset, and determine the feature values ​​of each node in the graph data, wherein the attributes include existing attributes and missing attributes; Based on the attribute values ​​of the target data and the feature values ​​of each node in the graph data, a first similarity value is calculated to characterize the syntactic similarity between each node in the graph data and the target data. Based on the first similarity value, a first candidate node in the graph data is determined. For any first candidate node, a first feature path starting from the first candidate node is determined, and the existing attributes in the target data are matched with the first feature path. The first candidate node corresponding to the first feature path that matches the existing attributes in the target data is determined as the first candidate matching node. For any first candidate matching node, a second similarity value representing the semantic similarity between the first candidate matching node and the target data is calculated based on the feature vector of the first candidate matching node and the attribute vector of the target data. Based on the second similarity value, a first matching node representing the same entity as the target data is determined. For any matching node, a second feature path starting from the first matching node is determined. The missing attributes in the target data are matched with the second feature path. The feature value corresponding to the second path that matches the missing attributes in the target data is determined as the first candidate attribute value. All the first candidate attribute values ​​are sorted to obtain a first sorting result. Based on the first sorting result, the batch attribute values ​​corresponding to the missing attributes in the target data are determined.

10. The computer device according to claim 9, wherein, The step of calculating a first similarity value, representing the syntactic similarity between each node in the graph data and the target data, based on the attribute values ​​of the target data and the feature values ​​of each node in the graph data, and determining a first candidate node in the graph data based on the first similarity value, includes: The attribute values ​​corresponding to all existing attributes in the target data are formed into an attribute value set; For any node in the graph data, determine the first subgraph data corresponding to the node, and form a feature value set by taking the feature values ​​of all nodes in the first subgraph data; Calculate the intersection of the attribute value set and the feature value set, and calculate the union of the attribute value set and the feature value set; The ratio of the intersection to the union is used as the first similarity value of the node. Based on the first similarity values ​​of all nodes in the graph data and a preset threshold, the first candidate node is determined.

11. The computer device according to claim 9, wherein, For any first candidate matching node, calculating a second similarity value representing the semantic similarity between the first candidate matching node and the target data based on the feature vector of the first candidate matching node and the attribute vector of the target data, and determining a first matching node representing the same entity as the target data based on the second similarity value, includes: For any first candidate matching node, determine the second subgraph data corresponding to the first candidate matching node, and vectorize the feature values ​​of all nodes in the second subgraph data to obtain the feature vector. Vectorize the attribute values ​​corresponding to all existing attributes in the target data to obtain attribute vectors; Calculate the cosine similarity value between the feature vector and the attribute vector, use the cosine similarity value as the second similarity value of the first candidate matching node, sort the second similarity values ​​of all first candidate matching nodes, and determine the first matching node that represents the same entity as the target data.

12. The computer device according to claim 9, wherein, Also includes: Obtain the inverted index constructed based on all nodes and feature edges in the feature paths of all nodes in the graph data; Obtain a matching data structure and a candidate data structure, wherein the matching data structure represents the mapping relationship between all target data and their corresponding first matching nodes, and the candidate data structure represents the mapping relationship between all target data and their corresponding first candidate nodes; After adding or deleting target feature edges in the graph data to obtain updated graph data, the associated nodes are determined based on the target feature edges and the inverted index, and the associated feature path starting from the associated node is determined based on the associated nodes. When performing missing attribute matching on the associated feature path corresponding to the associated node, the target data corresponding to the associated node found in the matching data structure is determined as the first data to be modified. Based on the associated feature path, the incremental attribute value corresponding to the missing attribute in the first data to be modified is determined. When performing existing attribute matching on the associated feature path corresponding to the associated node, the target data corresponding to the associated node found in the candidate data structure is determined to be the second data to be modified. The attribute values ​​corresponding to the missing attributes in the second data to be modified are recalculated according to the updated graph data to obtain the incremental attribute values ​​corresponding to the missing attributes in the second data to be modified.

13. The computer device according to claim 12, wherein, After adding or deleting target feature edges in the graph data to obtain updated graph data, the method further includes: If the target feature edge is added to the graph data, the target feature edge is written into a blank key value of the inverted index, and the associated node is written into the array corresponding to the key value; If the target feature edge is deleted from the graph data, then the key value and data corresponding to the target feature edge are removed from the inverted index.

14. The computer device according to claim 12, wherein, The step of determining the incremental attribute value corresponding to the missing attribute in the first data to be modified based on the associated feature path includes: Obtain the feature value of the last node in the associated feature path, and update the first candidate attribute value corresponding to the first data to be modified based on the feature value of the last node in the associated feature path to obtain the second candidate attribute value. The second candidate attribute values ​​are sorted to obtain a second sorting result. Based on the second sorting result, the incremental attribute values ​​corresponding to the missing attributes in the first data to be modified are determined.

15. The computer device according to claim 12, wherein, The step of recalculating the attribute values ​​corresponding to the missing attributes in the second data to be modified based on the updated graph data to obtain the incremental attribute values ​​corresponding to the missing attributes in the second data to be modified includes: Obtain updated graph data, determine the feature value of each node in the updated graph data, and determine the attributes and corresponding attribute values ​​of the second data to be modified; Based on the attribute values ​​of the second data to be modified and the feature values ​​of each node in the updated graph data, a third similarity value is calculated to characterize the syntactic similarity between each node in the updated graph data and the second data to be modified. Based on the third similarity value, a second candidate node in the updated graph data is determined. Based on the second candidate node, the first candidate node corresponding to the second data to be modified in the candidate data structure is updated to obtain the updated candidate data structure; For any second candidate node, a third feature path starting from the second candidate node is determined. The existing attributes in the second data to be modified are matched with the third feature path, and the second candidate node corresponding to the third feature path that matches the existing attributes in the second data to be modified is determined as the second candidate matching node. For any second candidate matching node, a fourth similarity value representing the semantic similarity between the second candidate matching node and the second data to be modified is calculated based on the feature vector of the second candidate matching node and the attribute vector of the second data to be modified. Based on the fourth similarity value, a second matching node representing the same entity as the second data to be modified is determined. Based on the second matching node, the first matching node corresponding to the second data to be modified in the matching data structure is updated to obtain the updated matching data structure; For any second matching node, a fourth feature path starting from the second matching node is determined. The missing attributes in the second data to be modified are matched with the fourth feature path. The feature value corresponding to the fourth feature path that matches the missing attributes in the second data to be modified is determined as the third candidate attribute value. All the third candidate attribute values ​​are sorted to obtain a third sorting result. Based on the third sorting result, the incremental attribute value corresponding to the missing attributes in the second data to be modified is determined.

16. One or more readable storage media storing computer-readable instructions, wherein, When the computer-readable instructions are executed by one or more processors, the one or more processors cause the processors to perform the following steps: Obtain the graph data corresponding to the graph structure and the dataset corresponding to the table structure, determine the attributes and corresponding attribute values ​​of the target data in the dataset, and determine the feature values ​​of each node in the graph data, wherein the attributes include existing attributes and missing attributes; Based on the attribute values ​​of the target data and the feature values ​​of each node in the graph data, a first similarity value is calculated to characterize the syntactic similarity between each node in the graph data and the target data. Based on the first similarity value, a first candidate node in the graph data is determined. For any first candidate node, a first feature path starting from the first candidate node is determined, and the existing attributes in the target data are matched with the first feature path. The first candidate node corresponding to the first feature path that matches the existing attributes in the target data is determined as the first candidate matching node. For any first candidate matching node, a second similarity value representing the semantic similarity between the first candidate matching node and the target data is calculated based on the feature vector of the first candidate matching node and the attribute vector of the target data. Based on the second similarity value, a first matching node representing the same entity as the target data is determined. For any matching node, a second feature path starting from the first matching node is determined. The missing attributes in the target data are matched with the second feature path. The feature value corresponding to the second path that matches the missing attributes in the target data is determined as the first candidate attribute value. All the first candidate attribute values ​​are sorted to obtain a first sorting result. Based on the first sorting result, the batch attribute values ​​corresponding to the missing attributes in the target data are determined.

17. The readable storage medium according to claim 16, wherein, The step of calculating a first similarity value, representing the syntactic similarity between each node in the graph data and the target data, based on the attribute values ​​of the target data and the feature values ​​of each node in the graph data, and determining a first candidate node in the graph data based on the first similarity value, includes: The attribute values ​​corresponding to all existing attributes in the target data are formed into an attribute value set; For any node in the graph data, determine the first subgraph data corresponding to the node, and form a feature value set by taking the feature values ​​of all nodes in the first subgraph data; Calculate the intersection of the attribute value set and the feature value set, and calculate the union of the attribute value set and the feature value set; The ratio of the intersection to the union is used as the first similarity value of the node. Based on the first similarity values ​​of all nodes in the graph data and a preset threshold, the first candidate node is determined.

18. The readable storage medium according to claim 16, wherein, For any first candidate matching node, calculating a second similarity value representing the semantic similarity between the first candidate matching node and the target data based on the feature vector of the first candidate matching node and the attribute vector of the target data, and determining a first matching node representing the same entity as the target data based on the second similarity value, includes: For any first candidate matching node, determine the second subgraph data corresponding to the first candidate matching node, and vectorize the feature values ​​of all nodes in the second subgraph data to obtain the feature vector. Vectorize the attribute values ​​corresponding to all existing attributes in the target data to obtain attribute vectors; Calculate the cosine similarity value between the feature vector and the attribute vector, use the cosine similarity value as the second similarity value of the first candidate matching node, sort the second similarity values ​​of all first candidate matching nodes, and determine the first matching node that represents the same entity as the target data.

19. The readable storage medium according to claim 16, wherein, Also includes: Obtain the inverted index constructed based on all nodes and feature edges in the feature paths of all nodes in the graph data; Obtain a matching data structure and a candidate data structure, wherein the matching data structure represents the mapping relationship between all target data and their corresponding first matching nodes, and the candidate data structure represents the mapping relationship between all target data and their corresponding first candidate nodes; After adding or deleting target feature edges in the graph data to obtain updated graph data, the associated nodes are determined based on the target feature edges and the inverted index, and the associated feature path starting from the associated node is determined based on the associated nodes. When performing missing attribute matching on the associated feature path corresponding to the associated node, the target data corresponding to the associated node found in the matching data structure is determined as the first data to be modified. Based on the associated feature path, the incremental attribute value corresponding to the missing attribute in the first data to be modified is determined. When performing existing attribute matching on the associated feature path corresponding to the associated node, the target data corresponding to the associated node found in the candidate data structure is determined to be the second data to be modified. The attribute values ​​corresponding to the missing attributes in the second data to be modified are recalculated according to the updated graph data to obtain the incremental attribute values ​​corresponding to the missing attributes in the second data to be modified.

20. The readable storage medium according to claim 19, wherein, After adding or deleting target feature edges in the graph data to obtain updated graph data, the method further includes: If the target feature edge is added to the graph data, the target feature edge is written into a blank key value of the inverted index, and the associated node is written into the array corresponding to the key value; If the target feature edge is deleted from the graph data, then the key value and data corresponding to the target feature edge are removed from the inverted index.

Images