A logistics entity alignment method, device, equipment and storage medium

Abstract

The present invention discloses a logistics entity alignment method, apparatus, device, and storage medium. The method comprises: clustering nodes with geographic information in a knowledge graph containing logistics data to obtain a plurality of subgraphs; wherein the knowledge graph comprises nodes and edges, the nodes being logistics entity data including enterprises and contacts, and the edges being relationship data; for each of the subgraphs, assigning a community label to each enterprise node that identifies the community to which it belongs, and utilizing a label propagation algorithm that incorporates temporal information of the logistics entity data to iteratively propagate the community label to each node to obtain a final community label for each node; for each final community label, performing entity alignment according to the corresponding node to obtain an entity alignment result including enterprises and contacts belonging to the same community. Utilizing the method disclosed in the present invention, substantially related enterprises and contacts can be associated through entity alignment.

Description

Technical Field

[0001] The present invention relates to the field of logistics technology, and in particular to a logistics entity alignment method, device, equipment and storage medium. Background Art

[0002] Building a logistics customer knowledge base requires processing a large amount of entity data from multiple source knowledge bases. Due to the varying sources of entity data, as well as differences in manual definition and proofreading, semantically identical entity data can be represented differently in different knowledge bases. Entity data with different names and semantics may represent or belong to the same entity, while entity data with the same entry name may represent two semantically different things. Regarding business needs, customer records are typically created based on contacts, and a company often has multiple contacts. If each contact is treated as a customer, business services and statistics for the company and the contact will be fragmented, complicating maintenance and data analysis.

[0003] Existing entity alignment solutions can only align identical entities. Furthermore, these solutions only consider the attributes of each entity, not the relationships between them. This makes it impossible to align actually related objects. Therefore, a method for aligning logistics entity data, including companies and contacts, is urgently needed. Summary of the Invention

[0004] The present invention provides a logistics entity alignment method, device, equipment and storage medium to solve the technical problem that the existing technology cannot align logistics entity data including enterprises and contacts and associate substantially related enterprises and contacts.

[0005] In order to solve the above technical problems, in a first aspect, the present invention provides a logistics entity alignment method, the method comprising:

[0006] Clustering nodes with geographic information in a knowledge graph containing logistics data to obtain several subgraphs; wherein the knowledge graph includes nodes and edges, the nodes are logistics entity data including enterprises and contacts, and the edges are relationship data;

[0007] For each of the subgraphs, a community label is assigned to each enterprise node, indicating the community to which it belongs. The community label is iteratively propagated to each node using a label propagation algorithm that incorporates the temporal information of the logistics entity data to obtain the final community label for each node.

[0008] For each final community label, entity alignment is performed according to the corresponding nodes to obtain entity alignment results including enterprises and contacts belonging to the same community.

[0009] Optionally, the iterative propagation of community labels to each node using a label propagation algorithm that integrates the time information of the logistics entity data includes:

[0010] Determine the edge weight of each edge based on the out-degree, in-degree, and number of paths of different lengths of the nodes of each edge, and determine the propagation probability of the node based on the edge weight and the time information of the logistics entity data;

[0011] For each community label of each node, obtain the label distribution probability of the node in round (n+1) based on the propagation probability and the label distribution probability of the node and its neighboring nodes in round n;

[0012] The propagation probability and label distribution probability are repeatedly updated until a preset stopping condition is reached, and the final label distribution probability of the node is obtained. The community label is then propagated to each node according to the final label distribution probability.

[0013] Optionally, the edge weight of each edge is determined based on the out-degree and in-degree of the nodes of each edge and the number of paths of different lengths, including:

[0014] Determine the edge weight of each edge using a weight calculation formula that includes the out-degree, in-degree, and number of paths of length 2 and 3 of each edge's node.

[0015] The weight calculation formula is:

[0016] w ij =min(max(tmp, 0), 1)

[0017]

[0018] Among them, w ij is the weight of the i-node of the edge to the j-node of the edge, I is the basic weight affected by the type of the edge, inD i is the in-degree of node i, outD i is the out-degree of node i, inD j is the in-degree of node j, outD j is the out-degree of node j, oneNodePath ij is the number of paths of length 2 between node i and node j, twoNodePath ij is the number of paths of length 3 between node i and node j.

[0019] Optionally, determining the propagation probability of the node according to the edge weight and the time information of the logistics entity data includes:

[0020] Determining the propagation probability of the node according to a propagation probability formula including the edge weight and the time information of the logistics entity data;

[0021] Among them, the propagation probability formula is:

[0022] Among them, T ij is the propagation probability of node i to node j, Δt ij The number of months from the current time when the logistics entity data was recorded.

[0023] Optionally, obtaining the label distribution probability of the node in round (n+1) according to the propagation probability and the label distribution probability of the node and its neighboring nodes in round n includes:

[0024] Obtaining a temporary label distribution probability of the node in round (n+1) according to a probability update formula including the propagation probability and the label distribution probability of the node and its neighboring nodes in round n;

[0025] Filter the temporary label distribution probability of the node in the n+1th round to obtain the label distribution probability of the node in the n+1th round;

[0026] Wherein, the probability update formula is:

[0027]

[0028]

[0029] in, is the label distribution probability of node j and label k in round n, is the temporary label distribution probability of node i k label in round n+1, T ji is the propagation probability of node j to node i, N(i) is the set of node i and its neighboring nodes, K(i) is the set of all labels of node i and its neighboring nodes, z i is the normalization coefficient.

[0030] Optionally, filtering the temporary label distribution probability of the node in the n+1th round to obtain the label distribution probability of the node in the n+1th round includes:

[0031] According to the probability filtering formula, the temporary label distribution probability of the node in the n+1th round is filtered to obtain the label distribution probability of the node in the n+1th round;

[0032] Among them, the probability filtering formula is:

[0033]

[0034]

[0035] in, is the label distribution probability of node i k label in round n+1, L(i) is the label set of node i, is the maximum value of the temporary label distribution probability of node i and label k in the n+1th round, S is the probability filtering coefficient, which is a positive real number greater than 1.

[0036] Optionally, clustering the nodes with geographic information in the knowledge graph containing logistics data to obtain several subgraphs, including:

[0037] Clustering algorithms are used to cluster nodes with geographic information in the knowledge graph containing logistics data;

[0038] Determine whether the average value of the clustering coefficient of each subgraph in the clustering result reaches the maximum; if not, adjust the radius parameter of the clustering algorithm by using the radius parameter adjustment formula until the judgment condition is met, thereby obtaining a plurality of subgraphs;

[0039] Among them, the clustering coefficient of each subgraph is

[0040] The radius parameter adjustment formula is:

[0041]

[0042]

[0043]

[0044] Among them, v is the node, r v is the number of triangles of v node, d v is the number of first-order neighbor nodes of node v, r m 、r m+1 are the radius parameters of the mth round and the m+1th round respectively, G is the full graph of the knowledge graph, g is the subgraph, GCC g is the clustering coefficient of the subgraph, is the average clustering coefficient of the mth round, is the change in the average clustering coefficient in the mth round, and M is the number of g in G.

[0045] Optionally, for each final community label, entity alignment is performed based on the corresponding node, including:

[0046] If the community corresponding to the final community label includes an enterprise, output all nodes of the community and the corresponding logistics entity data;

[0047] If the community corresponding to the final community label includes multiple enterprises, determine whether the multiple enterprises belong to the same group. If so, output all nodes of the community and the corresponding logistics entity data. Otherwise, perform derivative segmentation on the community according to the standard of one enterprise corresponding to one community, and output all nodes of several communities and the corresponding logistics entity data.

[0048] If the community corresponding to the final community tag does not include an enterprise, a third-party interface is called to query enterprise information, and the enterprise information is used as the enterprise node of the community, and all nodes of the community and the corresponding logistics entity data are output.

[0049] Optionally, after obtaining entity alignment results including enterprises and contacts belonging to the same community, further comprising:

[0050] If several enterprises in the historical community in the historical entity alignment result have multiple community labels in the entity alignment result, the enterprise with the smallest number among the enterprises uses the historical community label, and the other enterprises in the enterprises use the community labels in the entity alignment result;

[0051] If several enterprises in the newly added community in the entity alignment result have multiple historical community labels in the historical entity alignment result, the historical community label with the smallest number is used as the community label of the several enterprises.

[0052] In a second aspect, the present invention provides a logistics entity alignment device, comprising a clustering module, a label updating module, and an entity alignment module;

[0053] The clustering module is used to cluster nodes with geographic information in a knowledge graph containing logistics data to obtain a plurality of subgraphs; wherein the knowledge graph includes nodes and edges, the nodes are logistics entity data including enterprises and contacts, and the edges are relationship data;

[0054] The label updating module is configured to assign a community label indicating the community to which each enterprise node belongs to each subgraph, and iteratively propagate the community label to each node using a label propagation algorithm that incorporates the time information of the logistics entity data to obtain a final community label for each node;

[0055] The entity alignment module is used to perform entity alignment according to the corresponding nodes for each final community label, and obtain an entity alignment result including enterprises and contacts belonging to the same community.

[0056] In a third aspect, the present invention provides a logistics entity alignment device, comprising a memory and a processor, wherein:

[0057] The memory is used to store computer programs;

[0058] The processor is used to read the program in the memory and execute the steps of the logistics entity alignment method provided in the first aspect above.

[0059] In a fourth aspect, the present invention provides a computer-readable storage medium having a readable computer program stored thereon, which, when executed by a processor, implements the steps of the logistics entity alignment method provided in the first aspect above.

[0060] Compared with the prior art, the logistics entity alignment method, device, equipment and storage medium provided by the present invention have the following beneficial effects:

[0061] It can meet the demand for alignment of logistics entity data including enterprises and contacts, reduce the manpower consumption of logistics entity data alignment, and improve the effect of logistics entity data alignment; by clustering the nodes with geographic information in the knowledge graph containing logistics data, several subgraphs are obtained to improve the efficiency of subsequent calculations and the efficiency of subsequent community label propagation; for each of the subgraphs, a community label that marks the community to which it belongs is assigned to each enterprise node, and a label propagation algorithm that integrates the time information of the logistics entity data is used to iteratively propagate the community label to each node to obtain the final community label of each node, which can improve the efficiency and accuracy of the final community label; for each final community label, entity alignment is performed according to the corresponding node to obtain an entity alignment result including enterprises and contacts belonging to the same community, and substantially related enterprises and contacts can be associated through entity alignment. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only part of the embodiments of the present invention, rather than all the embodiments. For ordinary technicians in this field, without paying any creative work, other drawings obtained based on these drawings are all within the scope of protection of this application.

[0063] Figure 1 This is a flow chart of a logistics entity alignment method provided by an embodiment of the present invention;

[0064] Figure 2 1 is a structural diagram of a logistics entity alignment device provided by an embodiment of the present invention;

[0065] Figure 3 This is a structural diagram of a logistics entity alignment device provided by an embodiment of the present invention;

[0066] Figure 4 It is a structural diagram of a computer-readable storage medium provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0067] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0068] In order to make the description of the present disclosure more detailed and complete, the following is an illustrative description of the implementation methods and specific examples of the present invention; however, this is not the only form of implementing or using the specific embodiments of the present invention. The implementation methods cover the features of multiple specific embodiments and the method steps and their sequence for constructing and operating these specific embodiments. However, other specific embodiments can also be used to achieve the same or equal functions and step sequences. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0069] It should be noted that the terms "first," "second," and the like in the description and claims of the present invention and the accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate, such that the embodiments of the present invention described herein can be practiced in an order other than that illustrated or described herein.

[0070] In the description of the embodiments of the present application, "multiple" refers to two or more than two, and other quantifiers should be understood similarly. The preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention. In addition, the embodiments of the present application and the features in the embodiments may be combined with each other if there is no conflict.

[0071] Example 1

[0072] like Figure 1 The above is a flow chart of a logistics entity alignment method provided by an embodiment of the present invention, and the logistics entity alignment method includes the following steps:

[0073] Step S101: Clustering nodes with geographic information in a knowledge graph containing logistics data to obtain a plurality of subgraphs; wherein the knowledge graph includes nodes and edges, the nodes are logistics entity data including enterprises and contacts, and the edges are relationship data;

[0074] The above-mentioned logistics data includes logistics entity data and relationship data. The logistics entity data is the entity data that needs to be aligned. The embodiment of the present invention does not impose any limitation on the specific implementation of the above-mentioned logistics data, such as format and integrity.

[0075] The above-mentioned logistics entity data includes enterprises and contacts, and may also include other logistics entity data, such as sales personnel, introducers, senders, recipients, payers, mobile phone numbers, ID numbers, bank card numbers, addresses, buildings and other entity data.

[0076] The above nodes correspond to logistics entity data; the above edges are relational data representing the relationship between different logistics entity data. The edges may have directions and edge weights, wherein a node includes a logistics entity data, for example, a node only includes an enterprise.

[0077] Relationship data represents the relationship between different logistics entity data. For example, if a customer and customer resources have a churn or conversion relationship, the corresponding relationship data is churn or conversion. If a customer and a company have an affiliation relationship, the corresponding relationship data is affiliation. If an introducer and a customer resource have an introduction relationship, the corresponding relationship data is introduction. If a customer and a salesperson have a maintenance relationship, the corresponding relationship data is maintenance. If a sender and a recipient have a shipping or receiving relationship, the corresponding relationship data is shipping or receiving. If a mobile phone number has a mobile phone relationship with the salesperson, introducer, sender, recipient, and payee, the corresponding relationship data is mobile phone. If an address, building, and company have a registered place relationship, the corresponding relationship data is registered place. If an ID card number, bank card number, and payee have an ownership relationship, the corresponding relationship data is ownership.

[0078] The above-mentioned knowledge graph containing logistics data is pre-constructed. Its construction method includes designing a knowledge description system about logistics data, that is, raw data, and then processing the logistics data based on the above-mentioned knowledge description system to obtain logistics entity data and relationship data, and then importing the logistics entity data and relationship data into the graph database.

[0079] When clustering nodes with geographic information in the knowledge graph containing logistics data, the DBSCAN clustering algorithm, K-Means algorithm, hierarchical clustering algorithm, spectral clustering algorithm, random walk algorithm, partitioning algorithm or modularity maximization algorithm can be used.

[0080] Step S102: For each of the subgraphs, a community label is assigned to each enterprise node, indicating the community to which it belongs. The community label is iteratively propagated to each node using a label propagation algorithm that incorporates the time information of the logistics entity data to obtain a final community label for each node.

[0081] When assigning community labels to nodes, only the community label indicating the community to which each enterprise node belongs may be assigned. No label may be assigned to other types of nodes, such as contact nodes.

[0082] The label propagation algorithm is a local community division algorithm based on label propagation. The time information of logistics entity data can be integrated into the label propagation algorithm to iteratively propagate the community label to each node to obtain the final community label of each node. The community label can be represented by a string, which can be randomly generated.

[0083] Step S103 : For each final community label, entity alignment is performed according to the corresponding node to obtain an entity alignment result including enterprises and contacts belonging to the same community.

[0084] Since the above-mentioned logistics entity data may include other logistics entity data in addition to enterprises and contacts, the entity alignment result may include other logistics entity data in addition to enterprises and contacts.

[0085] The embodiments of the present invention can meet the demand for aligning logistics entity data including enterprises and contacts, reduce the manpower consumption for aligning logistics entity data, and improve the effect of aligning logistics entity data.

[0086] As an optional implementation, the nodes with geographic information in the knowledge graph containing logistics data are clustered to obtain several subgraphs, including:

[0087] Clustering algorithms are used to cluster nodes with geographic information in the knowledge graph containing logistics data;

[0088] Determine whether the average value of the clustering coefficient of each subgraph in the clustering result reaches the maximum; if not, adjust the radius parameter of the clustering algorithm by using the radius parameter adjustment formula until the judgment condition is met, thereby obtaining a plurality of subgraphs;

[0089] Among them, the clustering coefficient of each subgraph is

[0090] The radius parameter adjustment formula is:

[0091]

[0092]

[0093]

[0094] Among them, v is the node, r v is the number of triangles of v node, d v is the number of first-order neighbor nodes of node v, r m 、r m+1 are the radius parameters of the mth round and the m+1th round respectively, G is the full graph of the knowledge graph, g is the subgraph, GCC g is the clustering coefficient of the subgraph, is the average clustering coefficient of the mth round, is the change in the average clustering coefficient in the mth round, and M is the number of g in G.

[0095] The clustering coefficient of the above subgraph can reflect the degree of aggregation of the nodes inside the subgraph. If the average value of the clustering coefficients of each subgraph in a clustering result reaches the maximum, it means that the clustering result is optimal.

[0096] The DBSCAN clustering algorithm is used to cluster nodes with geographic information in the knowledge graph containing logistics data, thereby dividing a large graph into several small subgraphs, thereby improving the efficiency of subsequent calculations and the efficiency of subsequent community label propagation; among them, nodes in the large graph that are not clustered or associated will be retained and not involved in subsequent calculations. Geographic information refers to the latitude and longitude coordinate information on the map or specific address information.

[0097] It should be noted that the K-Means algorithm requires specifying the number of clusters K and is more sensitive to noisy data. Hierarchical clustering algorithms and spectral clustering algorithms may have computing time and memory limitations for large-scale graphs. Although the random walk algorithm is suitable for segmenting graphs, the segmentation results are too unstable and are particularly susceptible to abnormal data. The segmentation algorithm is suitable for dividing graphs into non-intersecting subgraphs, but due to data quality issues, an abnormal relationship may cause the subgraphs to intersect and become inseparable, seriously affecting the clustering effect. The modularity maximization algorithm cannot achieve the goal of repeated community detection; the DBSCAN clustering algorithm does not have the above shortcomings and can be used for very densely clustered areas; and the DBSCAN clustering algorithm can segment large graphs based on the characteristics of enterprise objects and geographical characteristics, while using graph semantic features, such as centrality features and triangle counts, to evaluate the effectiveness of segmentation.

[0098] As an optional implementation, the method of iteratively propagating community labels to each node using a label propagation algorithm that integrates the time information of the logistics entity data includes:

[0099] Determine the edge weight of each edge based on the out-degree, in-degree, and number of paths of different lengths of the nodes of each edge, and determine the propagation probability of the node based on the edge weight and the time information of the logistics entity data;

[0100] For each community label of each node, obtain the label distribution probability of the node in round (n+1) based on the propagation probability and the label distribution probability of the node and its neighboring nodes in round n;

[0101] The propagation probability and label distribution probability are repeatedly updated until a preset stopping condition is reached, and the final label distribution probability of the node is obtained. The community label is then propagated to each node according to the final label distribution probability.

[0102] In some embodiments, the above-mentioned preset stopping condition can be a preset number of iterations, for example, 1000 times, or a preset degree of change in the community label, that is, repeatedly updating the propagation probability and the label distribution probability until the preset number of iterations is reached to obtain the final label distribution probability of the node, or repeatedly updating the propagation probability and the label distribution probability until the preset degree of change in the community label is reached to obtain the final label distribution probability of the node.

[0103] The degree of change of the community label is the ratio of the number of nodes whose community labels have changed to the number of all nodes. The degree of change of the community label is preset to be a real number less than 1. The specific value thereof may not be limited, for example, it is 0.001 or a real number less than 0.001.

[0104] In an embodiment of the present invention, the propagation probability of the node is determined based on the edge weight and the time information of the logistics entity data, and the label distribution probability of the node in the n+1th round is obtained based on the propagation probability and the label distribution probability of the node and its neighboring nodes in the nth round. In this way, the final label distribution probability of the node can be accurately obtained. According to the final label distribution probability, the community label is propagated to each node, which can improve the entity alignment effect.

[0105] As an optional implementation, the edge weight of each edge is determined based on the out-degree and in-degree of the nodes of each edge and the number of paths of different lengths, including:

[0106] Determine the edge weight of each edge using a weight calculation formula that includes the out-degree, in-degree, and number of paths of length 2 and 3 of each edge's nodes.

[0107] The weight calculation formula is:

[0108] w ij =min(max(tmp, 0), 1)

[0109]

[0110] Among them, w ij is the weight of the i-node of the edge to the j-node of the edge, I is the basic weight affected by the type of the edge, inD i is the in-degree of node i, outD i is the out-degree of node i, inD j is the in-degree of node j, outD jis the out-degree of node j, oneNodePath ij is the number of paths of length 2 between node i and node j, twoNodePath ij is the number of paths of length 3 between node i and node j.

[0111] In some embodiments, the basic weight I is affected by the type of connection, which may include mobile phone, ID, affiliation, workplace, registration location, and follow-up notes, etc. The basic weights corresponding to different connection types may be different. For example, if the connection type includes mobile phone, ID, or affiliation, the corresponding basic weight I may be the first weight, such as 0.9; if the connection type is workplace, registration location, or follow-up notes, the corresponding basic weight I may be the second weight, such as 0.5; if the connection type is other types, such as corporate address, etc., the corresponding basic weight I may be the third weight, such as 0.1.

[0112] As an optional implementation, determining the propagation probability of the node according to the edge weight and the time information of the logistics entity data includes:

[0113] Determining the propagation probability of the node according to a propagation probability formula including the edge weight and the time information of the logistics entity data;

[0114] Among them, the propagation probability formula is

[0115] Among them, T ij is the propagation probability of node i to node j, Δt ij The number of months from the current time when the logistics entity data was recorded.

[0116] It should be noted that the number of months between the record of the logistics entity data and the current time is the time information of the logistics entity data. The propagation probability of the node is not only related to the edge weight, but also related to the time information of the logistics entity data.

[0117] In the embodiment of the present invention, the propagation probability of each node can be quickly obtained according to the propagation probability formula including the edge weight and the time information of the logistics entity data.

[0118] As an optional implementation, obtaining the label distribution probability of the node in the (n+1)th round according to the propagation probability and the label distribution probability of the node and its neighboring nodes in the nth round includes:

[0119] Obtaining a temporary label distribution probability of the node in round (n+1) according to a probability update formula including the propagation probability and the label distribution probability of the node and its neighboring nodes in round n;

[0120] Filter the temporary label distribution probability of the node in the n+1th round to obtain the label distribution probability of the node in the n+1th round;

[0121] Wherein, the probability update formula is:

[0122]

[0123]

[0124] in, is the label distribution probability of node j and label k in round n, is the temporary label distribution probability of node i k label in round n+1, T ji is the propagation probability of node j to node i, N(i) is the set of node i and its neighboring nodes, K(i) is the set of all labels of node i and its neighboring nodes, z i is the normalization coefficient.

[0125] It should be noted that the probability of temporary label distribution of node i k label in round n+1 is the same as T ji It is related to the propagation probability of node j to node i and the label distribution probability of node j’s k label in the nth round.

[0126] In the embodiment of the present invention, the temporary label distribution probability of the node in the (n+1) round can be accurately obtained according to the probability update formula including the propagation probability and the label distribution probability of the node and its neighboring nodes in the nth round.

[0127] As an optional implementation, filtering the temporary label distribution probability of the node in the n+1th round to obtain the label distribution probability of the node in the n+1th round includes:

[0128] According to the probability filtering formula, the temporary label distribution probability of the node in the n+1th round is filtered to obtain the label distribution probability of the node in the n+1th round;

[0129] Among them, the probability filtering formula is:

[0130]

[0131]

[0132] in, is the label distribution probability of node i k label in round n+1, L(i) is the label set of node i, is the maximum value of the temporary label distribution probability of node i and label k in the n+1th round, S is the probability filtering coefficient, which is a positive real number greater than 1.

[0133] In some embodiments, the temporary label distribution probability of the node in the n+1 round is filtered according to the probability filtering formula, and the community labels with a maximum probability value of less than 1 / S in the temporary label distribution probability of the node in the n+1 round can be filtered out. After filtering the temporary label distribution probability of the node in the n+1 round, the remaining community labels, that is, the community labels corresponding to the temporary label distribution probability greater than or equal to 1 / S of the maximum probability value, are normalized to obtain the label distribution probability of the node in the n+1 round; wherein, the probability filtering coefficient S is a positive real number greater than 1, which can be pre-set according to actual conditions, for example, S is 3.

[0134] In the embodiment of the present invention, the temporary label distribution probability of the node in the n+1th round is filtered through a probability filtering formula, which can improve the calculation efficiency of the label distribution probability and the efficiency of community label propagation, thereby improving the efficiency of entity alignment.

[0135] As an optional implementation, for each final community label, entity alignment is performed based on the corresponding node, including:

[0136] If the community corresponding to the final community label includes an enterprise, output all nodes of the community and the corresponding logistics entity data;

[0137] If the community corresponding to the final community label includes multiple enterprises, determine whether the multiple enterprises belong to the same group. If so, output all nodes of the community and the corresponding logistics entity data. Otherwise, perform derivative segmentation on the community according to the standard of one enterprise corresponding to one community, and output all nodes of several communities and the corresponding logistics entity data.

[0138] If the community corresponding to the final community tag does not include an enterprise, a third-party interface is called to query enterprise information, and the enterprise information is used as the enterprise node of the community, and all nodes of the community and the corresponding logistics entity data are output.

[0139] In some embodiments, after obtaining the final community label, the nodes can be merged according to the community label, that is, the data structure of the community label corresponding to the node can be transformed into the data structure of the community label corresponding to the node; for example, {"Enterprise 1":["Community 1","Community 2"],"Customer 1":["Community 1","Community 3"]} is transformed into {"Community 1":["Enterprise 1","Customer 1"],"Community 2":["Enterprise 1"],"Community 3":["Customer 1"]}, so as to facilitate the judgment of how many enterprise nodes a community corresponds to.

[0140] In some embodiments, if the community corresponding to the final community label includes an enterprise, all nodes of the community and the logistics entity data corresponding to the nodes are output; wherein all nodes may include enterprise nodes, customer nodes, contact nodes, and mobile phone number nodes, etc., and the corresponding logistics entity data may be enterprises, customers, contacts, and mobile phone numbers, etc.; if the community corresponding to the final community label includes multiple enterprises, the multiple enterprises are input into the enterprise group discrimination model to determine whether the multiple enterprises belong to the same group. If the judgment result output by the enterprise group discrimination model is that the multiple enterprises belong to the same group, all nodes of the community and the corresponding logistics entity data are output, and the corresponding community label is output. If the judgment result output by the enterprise group discrimination model is that the multiple enterprises do not belong to the same group, If there is a group, the community is derived and segmented according to the standard that one enterprise corresponds to one community, and all nodes of several communities and the corresponding logistics entity data are output. When the community is derived and segmented, different communities may have the same nodes, that is, overlapping nodes. The above-mentioned enterprise family discrimination model is a classification model trained based on some enterprise family data in public industrial and commercial information and the correspondence between company names. If the community corresponding to the final community label does not include an enterprise, the enterprise information can be queried by calling a third-party interface based on the enterprise abbreviation or other descriptive text of the customer object. The enterprise information is used as the enterprise node of the community, and all nodes of the community and the corresponding logistics entity data are output. Finally, the entity alignment result needs to include enterprises and contacts belonging to the same community.

[0141] In an embodiment of the present invention, for each final community tag, different processing is performed depending on whether the community corresponding to the final community tag includes one enterprise, multiple enterprises, or no enterprises. When the community corresponding to the final community tag includes one enterprise, all nodes of the community and the corresponding logistics entity data are directly output. When the community corresponding to the final community tag includes multiple enterprises, the community is derived and split, and all nodes of several communities and the corresponding logistics entity data are output. A third-party interface is only called if the community corresponding to the final community tag does not include an enterprise, thereby reducing the number of calls to the third-party interface and lowering the cost of entity alignment.

[0142] As an optional implementation, after obtaining the entity alignment result including the enterprises and contacts belonging to the same community, further comprising:

[0143] If several enterprises in the historical community in the historical entity alignment result have multiple community labels in the entity alignment result, the enterprise with the smallest number among the enterprises uses the historical community label, and the other enterprises in the enterprises use the community labels in the entity alignment result;

[0144] If several enterprises in the newly added community in the entity alignment result have multiple historical community labels in the historical entity alignment result, the historical community label with the smallest number is used as the community label of the several enterprises.

[0145] In some embodiments, after obtaining entity alignment results including enterprises and contacts belonging to the same community, the community tags corresponding to the enterprises need to be stored in a cache system. If there are historical community tags in the historical community in the cache system, the historical community tags are updated.

[0146] In a specific embodiment, if there are two enterprises in a historical community with a community label of 1 in the historical entity alignment result, namely, Enterprise 1 and Enterprise 2, and there are two community labels in the entity alignment result, namely, Enterprise 1 corresponds to community label 2 and Enterprise 2 corresponds to community label 3, then the enterprise with the smallest number among the two enterprises uses the historical community label, namely, Enterprise 1 uses community label 1 and Enterprise 2 uses community label 3.

[0147] In another specific embodiment, if several enterprises in the newly added community with the community label 6 in the entity alignment result, namely, Enterprise 3 and Enterprise 4, have multiple historical community labels in the historical entity alignment results, that is, Enterprise 3 corresponds to community label 4, and Enterprise 4 corresponds to community label 5, then the historical community label with the smallest number, namely, community label 4, is used as the community label of the several enterprises, namely, Enterprise 3 and Enterprise 4.

[0148] In some embodiments, after determining the community tag of an enterprise, the community tag is written into a graph database, and the relationship between customers, leads and enterprises can also be obtained and written into a query engine to facilitate subsequent queries; wherein, the clues are data including "name": xxx, "company name": xxx, "mobile phone number": xxx, "referrer": xxx and other content.

[0149] In some embodiments, after completing the logistics entity alignment, whenever new logistics entity data or relationship data is input, the graph database is updated with the new logistics entity data and relationship data, and then the entity alignment is performed, specifically including:

[0150] A new knowledge graph is formed based on the new logistics entity data and relationship data, and the label propagation algorithm is used to propagate community labels from nodes with community labels to nodes without community labels.

[0151] After the label propagation is completed, if the node corresponding to the new logistics entity data has only one community label, then the community label is output; if the node corresponding to the new logistics entity data has multiple community labels, the community is derived and split according to the standard of one enterprise corresponding to one community; if the node corresponding to the new logistics entity data has no community label, then the new logistics entity data is retained.

[0152] The modularity of the entire graph is calculated. If the modularity of the entire graph is lower than a preset value, the logistics entity alignment method according to an embodiment of the present invention is executed.

[0153] Some companies currently divide customer maintenance into two categories: service customers and potential customers. Successfully developed potential customers are transferred to service customers, while service customers who churn are transferred to potential customers. However, this results in the same company being imported into the potential customer list by multiple contacts, leading to an increasing number of records in the potential customer list, creating a data maze that is difficult to analyze. Logistics-related data analysis often requires repeated identification logic for identical companies, but this is limited to determining equivalence in the company name field. This is far from sufficient, as the company name field is manually entered and often contains abbreviations and dirty data. Without the ability to link objects belonging to the same company, the business database data collected by the data warehouse cannot support customer lifecycle analysis.

[0154] Therefore, it is essential to align the logistics entity data, including companies and contacts. By aligning the logistics entity data, including companies and contacts, objects from the same company can be linked, solving the aforementioned data silo problem. Traditional entity alignment schemes only consider the attributes of each entity, ignoring the relationships between entities. This approach has the following drawbacks: first, it is labor-intensive; second, it is difficult to migrate across multiple domains; third, because the expression of attributes is discrete, this entity alignment scheme ignores the semantic similarity of attributes, limiting the effectiveness of entity alignment; and fourth, it is unable to identify nodes with dissimilar field content but are actually related. Furthermore, large-scale knowledge fusion technology requires a rich contextual environment, but the predicate types between entity data in business systems are few, making it difficult for existing technical solutions to utilize the semantics of predicates for fusion.

[0155] An embodiment of the present invention provides a logistics entity alignment method, which clusters nodes with geographic information in a knowledge graph containing logistics data to obtain several subgraphs to improve the efficiency of subsequent calculations and the efficiency of subsequent community label propagation; for each of the subgraphs, a community label that indicates the community to which it belongs is assigned to each enterprise node, and a label propagation algorithm that integrates the time information of the logistics entity data is used to iteratively propagate the community label to each node to obtain the final community label of each node, which can improve the efficiency and accuracy of the final community label; for each final community label, entity alignment is performed according to the corresponding node to obtain an entity alignment result including enterprises and contacts belonging to the same community. Through entity alignment, substantially related enterprises and contacts are associated, thereby meeting the demand for alignment of logistics entity data including enterprises and contacts, reducing the manpower consumption of logistics entity data alignment, and improving the effect of logistics entity data alignment.

[0156] Example 2

[0157] Based on the above logistics entity alignment method, the embodiment of the present invention provides a logistics entity alignment device, the structural diagram of which is as follows: Figure 2 As shown, the logistics entity alignment device 20 includes a clustering module 21, a label updating module 22 and an entity alignment module 23;

[0158] The clustering module 21 is used to cluster nodes with geographic information in the knowledge graph containing logistics data to obtain a plurality of subgraphs; wherein the knowledge graph includes nodes and edges, the nodes are logistics entity data including enterprises and contacts, and the edges are relationship data;

[0159] The label updating module 22 is configured to assign a community label indicating the community to which each enterprise node belongs to each subgraph, and iteratively propagate the community label to each node using a label propagation algorithm that incorporates the time information of the logistics entity data to obtain a final community label for each node;

[0160] The entity alignment module 23 is configured to perform entity alignment according to the corresponding nodes for each final community label, and obtain an entity alignment result including enterprises and contacts belonging to the same community.

[0161] For other details about how the modules in the logistics entity alignment device implement the above technical solution, please refer to the description of the logistics entity alignment method provided in the above invention embodiment, which will not be repeated here.

[0162] Example 3

[0163] Based on the above logistics entity alignment method, such as Figure 3 As shown, an embodiment of the present invention further provides a schematic structural diagram of a logistics entity alignment device, which includes a processor 31 and a memory 32 coupled to the processor 31. The memory 32 stores a computer program, which, when executed by the processor 31, causes the processor 31 to perform the steps of the logistics entity alignment method in the above embodiment.

[0164] For other details about how the processor 31 in the logistics entity alignment device implements the above technical solution, please refer to the description of the logistics entity alignment method provided in the above invention embodiment, which will not be repeated here.

[0165] Among them, the processor 31 can also be called a CPU (Central Processing Unit), and the processor 31 may be an integrated circuit chip with signal processing capabilities; the processor 31 can also be a general-purpose processor, DSP (Digital Signal Processor), ASIC (Application Specific Integrated Circuit), FPGA (Field Programmable Gate Array) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, among which the general-purpose processor can be a microprocessor or the processor 31 can also be any conventional processor, etc.

[0166] Example 4

[0167] like Figure 4 As shown, an embodiment of the present invention further provides a schematic diagram of the structure of a computer-readable storage medium, wherein a readable computer program 41 is stored on the storage medium 40; wherein the computer program 41 can be stored in the above-mentioned storage medium in the form of a software product, including a number of instructions for causing a computer device (which can be a personal computer, server, or network device, etc.) or a processor to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: a USB flash drive, a mobile hard disk, a magnetic disk or an optical disk, a ROM (Read-Only Memory), a RAM (Random Access Memory), and other media that can store program code, or a terminal device such as a computer, server, mobile phone, or tablet.

[0168] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the modules is merely a logical function division. In actual implementation, there may be other division methods, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0169] The modules described as separate components may or may not be physically separate, and the components shown as modules may or may not be physical modules, that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules may be selected to achieve the purpose of the present embodiment according to actual needs.

[0170] In addition, the functional modules in the various embodiments of the present application may be integrated into a single processing module, or each module may exist physically separately, or two or more modules may be integrated into a single module. The aforementioned integrated modules may be implemented in the form of hardware or in the form of software functional modules. If the integrated modules are implemented in the form of software functional modules and sold or used as independent products, they may be stored in a computer-readable storage medium.

[0171] In the above embodiments, all or part of the embodiments may be implemented by software, hardware, firmware, or any combination thereof. When implemented by software, all or part of the embodiments may be implemented in the form of a computer program product.

[0172] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website, a computer, a server, or a data center by wired (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, microwave, etc.) mode to another website, a computer, a server, or a data center. The computer-readable storage medium can be any available medium that a computer can store or a data storage device such as a server or a data center that includes one or more available media integrations. The available medium can be a magnetic medium, (such as a floppy disk, a hard disk, a magnetic tape), an optical medium (such as a DVD), or a semiconductor medium (such as a solid-state drive (SSD)).

[0173] The above is a detailed introduction to the technical solution provided by the present application. Specific examples are used in the present application to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea. At the same time, for those skilled in the art, according to the ideas of the present application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

[0174] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0175] The present application is described with reference to the flowcharts and / or block diagrams of the methods, apparatuses and computer program products according to the present application. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as the combination of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0176] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0177] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0178] Obviously, those skilled in the art may make various changes and modifications to this application without departing from the spirit and scope of this application. Thus, if these modifications and variations of this application fall within the scope of the claims of this application and their equivalents, this application is intended to include these modifications and variations.

Claims

1. A logistics entity alignment method, characterized in that: include: Clustering nodes with geographic information in a knowledge graph containing logistics data to obtain several subgraphs; wherein the knowledge graph includes nodes and edges, the nodes are logistics entity data including enterprises and contacts, and the edges are relationship data; For each of the subgraphs, a community label is assigned to each enterprise node, indicating the community to which it belongs. The community label is iteratively propagated to each node using a label propagation algorithm that incorporates the temporal information of the logistics entity data to obtain the final community label for each node. For each final community label, entity alignment is performed according to the corresponding node, including, If the community corresponding to the final community label includes an enterprise, all nodes of the community and the corresponding logistics entity data are output. If the community corresponding to the final community label includes multiple enterprises, determine whether the multiple enterprises belong to the same group. If so, output all nodes of the community and the corresponding logistics entity data. Otherwise, perform derivative segmentation on the community according to the standard of one enterprise corresponding to one community, and output all nodes of several communities and the corresponding logistics entity data. If the community corresponding to the final community tag does not include an enterprise, call a third-party interface to query enterprise information, use the enterprise information as the enterprise node of the community, and output all nodes of the community and the corresponding logistics entity data. Entity alignment results are obtained that include businesses and contacts belonging to the same community.

2. The logistics entity alignment method according to claim 1, characterized in that: The method of iteratively propagating community labels to each node by utilizing a label propagation algorithm that integrates the time information of the logistics entity data includes: Determine the edge weight of each edge based on the out-degree, in-degree, and number of paths of different lengths of the nodes of each edge, and determine the propagation probability of the node based on the edge weight and the time information of the logistics entity data; For each community label of each node, obtain the label distribution probability of the node in round (n+1) based on the propagation probability and the label distribution probability of the node and its neighboring nodes in round n; The propagation probability and label distribution probability are repeatedly updated until a preset stopping condition is reached, and the final label distribution probability of the node is obtained. The community label is then propagated to each node according to the final label distribution probability.

3. The logistics entity alignment method according to claim 2, characterized in that: The edge weight of each edge is determined based on the out-degree and in-degree of the nodes on each edge and the number of paths of different lengths, including: Determine the edge weight of each edge using a weight calculation formula that includes the out-degree, in-degree, and number of paths of length 2 and 3 of each edge's nodes. The weight calculation formula is: w ij =min(max(tmp,0),1) Among them, w ij is the weight of the i-node of the edge to the j-node of the edge, I is the basic weight affected by the type of the edge, inD i is the in-degree of node i, outD i is the out-degree of node i, inD j is the in-degree of node j, outD j is the out-degree of node j, oneNodePath ij is the number of paths of length 2 between node i and node j, twoNodePath ij is the number of paths of length 3 between node i and node j.

4. The logistics entity alignment method according to claim 2, characterized in that: Determining the propagation probability of the node according to the edge weight and the time information of the logistics entity data includes: Determining the propagation probability of the node according to a propagation probability formula including the edge weight and the time information of the logistics entity data; Among them, the propagation probability formula is: Among them, T ij is the propagation probability of node i to node j, Δt ij The number of months from the current time when the logistics entity data was recorded.

5. The logistics entity alignment method according to claim 2, characterized in that: Obtaining the label distribution probability of the node in round (n+1) based on the propagation probability and the label distribution probability of the node and its neighboring nodes in round n includes: Obtaining a temporary label distribution probability of the node in round (n+1) according to a probability update formula including the propagation probability and the label distribution probability of the node and its neighboring nodes in round n; Filter the temporary label distribution probability of the node in the n+1th round to obtain the label distribution probability of the node in the n+1th round; Wherein, the probability update formula is: in, is the label distribution probability of node j and label k in round n, is the temporary label distribution probability of node i k label in round n+1, T ji is the propagation probability of node j to node i, N(i) is the set of node i and its neighboring nodes, K(i) is the set of all labels of node i and its neighboring nodes, z i is the normalization coefficient.

6. The logistics entity alignment method according to claim 5, characterized in that: Filtering the temporary label distribution probability of the node in the n+1th round to obtain the label distribution probability of the node in the n+1th round includes: According to the probability filtering formula, the temporary label distribution probability of the node in the n+1th round is filtered to obtain the label distribution probability of the node in the n+1th round; Among them, the probability filtering formula is: in, is the label distribution probability of node i k label in round n+1, L(i) is the label set of node i, is the maximum value of the temporary label distribution probability of node i and label k in the n+1th round, S is the probability filtering coefficient, which is a positive real number greater than 1.

7. The logistics entity alignment method according to claim 1, characterized in that: The nodes with geographic information in the knowledge graph containing logistics data are clustered to obtain several subgraphs, including: Clustering algorithms are used to cluster nodes with geographic information in the knowledge graph containing logistics data; Determine whether the average value of the clustering coefficient of each subgraph in the clustering result reaches the maximum; if not, adjust the radius parameter of the clustering algorithm by using the radius parameter adjustment formula until the judgment condition is met, thereby obtaining a plurality of subgraphs; Among them, the clustering coefficient of each subgraph is The radius parameter adjustment formula is: Among them, v is the node, r v is the number of triangles of v node, d v is the number of first-order neighbor nodes of node v, r m 、r m+1 are the radius parameters of the mth round and the m+1th round respectively, G is the full graph of the knowledge graph, g is the subgraph, GCC g is the clustering coefficient of the subgraph, is the average clustering coefficient of the mth round, is the change in the average clustering coefficient in the mth round, and M is the number of g in G.

8. The logistics entity alignment method according to claim 1, characterized in that: After obtaining entity alignment results including businesses and contacts belonging to the same community, it also includes; If several enterprises in the historical community in the historical entity alignment result have multiple community labels in the entity alignment result, the enterprise with the smallest number among the enterprises uses the historical community label, and the other enterprises in the enterprises use the community labels in the entity alignment result; If several enterprises in the newly added community in the entity alignment result have multiple historical community labels in the historical entity alignment result, the historical community label with the smallest number is used as the community label of the several enterprises.

9. A logistics entity alignment device, characterized in that: Includes clustering module, label update module and entity alignment module; The clustering module is used to cluster nodes with geographic information in a knowledge graph containing logistics data to obtain a plurality of subgraphs; wherein the knowledge graph includes nodes and edges, the nodes are logistics entity data including enterprises and contacts, and the edges are relationship data; The label updating module is configured to assign a community label indicating the community to which each enterprise node belongs to each subgraph, and iteratively propagate the community label to each node using a label propagation algorithm that incorporates the time information of the logistics entity data to obtain a final community label for each node; The entity alignment module is used to perform entity alignment according to the corresponding nodes for each final community label, including: If the community corresponding to the final community label includes an enterprise, all nodes of the community and the corresponding logistics entity data are output. If the community corresponding to the final community label includes multiple enterprises, determine whether the multiple enterprises belong to the same group. If so, output all nodes of the community and the corresponding logistics entity data. Otherwise, perform derivative segmentation on the community according to the standard of one enterprise corresponding to one community, and output all nodes of several communities and the corresponding logistics entity data. If the community corresponding to the final community tag does not include an enterprise, call a third-party interface to query enterprise information, use the enterprise information as the enterprise node of the community, and output all nodes of the community and the corresponding logistics entity data. Entity alignment results are obtained that include businesses and contacts belonging to the same community.

10. A logistics entity alignment device, characterized in that: comprising a memory and a processor, wherein: The memory is used to store computer programs; The processor is configured to read the computer program in the memory and execute the steps of any one of the logistics entity alignment methods according to claims 1 to 8.

11. A computer-readable storage medium, characterized in that A readable computer program is stored thereon, and when the program is executed by a processor, the steps of any logistics entity alignment method as described in claims 1 to 8 are implemented.

Images