Object distribution map generation method and apparatus, electronic device, and storage medium

Abstract

The present disclosure provides an object distribution map generation method and device, electronic equipment and storage medium. The method comprises: obtaining operation feedback data, and generating a heterogeneous graph according to the operation feedback data; obtaining an object isomorphism graph according to the object nodes of a plurality of operation objects in the heterogeneous graph; performing graph pooling processing on the object isomorphism graph based on the embedding features of each object node, to determine a plurality of target object nodes from the plurality of object nodes in the object isomorphism graph, and generate an object distribution map corresponding to the target object nodes. The method can quickly screen out operation objects with low activity, so that computer resources can be saved when generating an object distribution map; and the target object nodes with high activity can be more accurately determined from the plurality of object nodes, so that the generated object distribution map is more accurate.

Description

Technical Field

[0001] This disclosure relates to the field of computer technology, and more specifically, to a method, apparatus, electronic device, computer-readable storage medium, and computer program product for generating object distribution maps. Background Technology

[0002] With the development of computer technology, more and more users are using electronic devices to browse web pages, watch videos, and buy items.

[0003] In related technologies, statistical analysis methods are typically used to cluster operational feedback data to generate object distribution maps. However, this method does not comprehensively consider the relationships between the objects included in the operational feedback data, and the generated object distribution maps cannot well represent the activity of objects, resulting in poor accuracy. Summary of the Invention

[0004] This disclosure provides an object distribution map generation method, apparatus, electronic device, computer-readable storage medium, and computer program product. On the one hand, the method can quickly filter out low-activity operation objects, thus saving computer resources when generating the object distribution map; on the other hand, it can more accurately determine the high-activity target object nodes from multiple object nodes, making the generated object distribution map more accurate.

[0005] This disclosure provides a method for generating an object distribution graph. The method includes: acquiring operation feedback data and generating a heterogeneous graph based on the operation feedback data; obtaining an object isomorphic graph based on object nodes of multiple operation objects in the heterogeneous graph; performing graph pooling processing on the object isomorphic graph based on the embedding features of each object node to determine multiple target object nodes from the multiple object nodes in the object isomorphic graph, and generating an object distribution graph corresponding to the target object nodes.

[0006] In some exemplary embodiments of this disclosure, graph pooling is performed on the object isomorphic graph based on the embedding features of each object node to determine multiple target object nodes from multiple object nodes in the object isomorphic graph. This includes: generating an initial parameter matrix of the object isomorphic graph based on the embedding features of each object node, and obtaining the attention weights of each object node according to the initial parameter matrix; performing a filtering process on the initial parameter matrix according to the attention weights of each object node to obtain a target parameter matrix; and determining multiple target object nodes from the multiple object nodes according to the target parameter matrix.

[0007] In some exemplary embodiments of this disclosure, the initial parameter matrix is ​​filtered out according to the attention weight of each object node to obtain the target parameter matrix, including: determining the index of the node to be filtered out according to the attention weight of each operation object; and filtering out the initial parameter matrix based on the index of the node to be filtered out to obtain the target parameter matrix.

[0008] In some exemplary embodiments of this disclosure, the object distribution map includes an object distribution heatmap; wherein, generating the object distribution map corresponding to the target object node includes: determining the location information of each target object node; determining the target coordinate range corresponding to the target object node with the location information of the target object node as the center; determining the weight of each target coordinate point according to the distance of each target coordinate point in the target coordinate range from the center; and superimposing the weights of each target coordinate point corresponding to each target object node to generate the object distribution heatmap.

[0009] In some exemplary embodiments of this disclosure, obtaining an object isomorphic graph based on the object nodes of multiple operational objects in the heterogeneous graph includes: extracting each object node in the heterogeneous graph and associating each object node with the others to generate the object isomorphic graph.

[0010] In some exemplary embodiments of this disclosure, before performing graph pooling on the isomorphic graph based on the embedding features of each object node, the method further includes: performing graph convolution on the heterogeneous graph to obtain the embedding features of each object node in the heterogeneous graph.

[0011] In some exemplary embodiments of this disclosure, graph convolution processing is performed on the heterogeneous graph to obtain the embedding features of each object node in the heterogeneous graph, including: performing graph convolution processing on the heterogeneous graph so that the object node and other nodes connected to the object node can exchange information to obtain the embedding features of each object node in the heterogeneous graph.

[0012] This disclosure provides an object distribution graph generation apparatus, comprising: an acquisition module configured to acquire operation feedback data and generate a heterogeneous graph based on the operation feedback data; an obtaining module configured to obtain an object isomorphic graph based on object nodes of multiple operation objects in the heterogeneous graph; and a determining module configured to perform graph pooling processing on the object isomorphic graph based on the embedding features of each object node, so as to determine multiple target object nodes from the multiple object nodes in the object isomorphic graph and generate an object distribution graph corresponding to the target object nodes.

[0013] This disclosure provides an electronic device, including: a processor; and a memory for storing processor-executable instructions; wherein the processor is configured to execute the executable instructions to implement the object distribution map generation method as described above.

[0014] This disclosure provides a computer-readable storage medium that, when the instructions in the computer-readable storage medium are executed by a processor of an electronic device, enables the electronic device to perform the object distribution map generation method as described above.

[0015] This disclosure provides a computer program product, including a computer program that, when executed by a processor, implements the object distribution map generation method described above.

[0016] The object distribution graph generation method provided in this disclosure generates a heterogeneous graph based on operation feedback data, obtains an object isomorphic graph based on the object nodes of multiple operation objects in the heterogeneous graph, and performs graph pooling processing on the operation object isomorphic graph based on the embedding features of each object node. On the one hand, it can quickly filter out operation objects with low activity, thus saving computer resources when generating the object distribution graph; on the other hand, it can more accurately determine the target object nodes with high activity from multiple object nodes, making the generated object distribution graph more accurate.

[0017] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0018] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure. It is obvious that the drawings described below are merely some embodiments of this disclosure, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0019] Figure 1 A schematic diagram of an exemplary system architecture for which the object distribution graph generation method of embodiments of the present disclosure can be applied is shown.

[0020] Figure 2 This is a flowchart illustrating an object distribution map generation method according to an exemplary embodiment.

[0021] Figure 3 This is a flowchart illustrating another method for generating an object distribution map according to an exemplary embodiment.

[0022] Figure 4 This is a schematic diagram illustrating graph pooling of a parameter matrix, as shown in an example.

[0023] Figure 5 This is a flowchart illustrating another method for generating an object distribution map according to an exemplary embodiment.

[0024] Figure 6 It is a schematic diagram of the target coordinate range corresponding to the position information of the target object node shown in an example.

[0025] Figure 7 It is a schematic diagram based on a heat map of object distribution shown in an example.

[0026] Figure 8 This is a block diagram illustrating an object distribution map generation apparatus according to an exemplary embodiment.

[0027] Figure 9 This is a schematic diagram illustrating the structure of an electronic device suitable for implementing exemplary embodiments of the present disclosure, according to an exemplary embodiment. Detailed Implementation

[0028] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the embodiments set forth herein; rather, they are provided so that this disclosure will be thorough and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted.

[0029] The features, structures, or characteristics described in this disclosure can be combined in any suitable manner in one or more embodiments. Numerous specific details are provided in the following description to give a thorough understanding of embodiments of this disclosure. However, those skilled in the art will recognize that the technical solutions of this disclosure can be practiced with one or more specific details omitted, or other methods, components, apparatuses, steps, etc., can be employed. In other instances, well-known methods, apparatuses, implementations, or operations are not shown or described in detail to avoid obscuring various aspects of this disclosure.

[0030] The accompanying drawings are merely illustrative of this disclosure, and the same reference numerals in the drawings denote the same or similar parts, thus omitting repeated descriptions of them. Some block diagrams shown in the drawings do not necessarily correspond to physically or logically independent entities. These functional entities may be implemented in software, in at least one hardware module or integrated circuit, or in different network and / or processor devices and / or microcontroller devices.

[0031] The flowchart shown in the accompanying drawings is merely illustrative and does not necessarily include all content and steps, nor does it require execution in the described order. For example, some steps may be broken down, while others may be combined or partially combined; therefore, the actual execution order may change depending on the specific circumstances.

[0032] In this specification, the terms “a,” “an,” “the,” “the,” and “at least one” are used to indicate the presence of at least one element / component / etc.; the terms “comprising,” “including,” and “having” are used to indicate an open-ended inclusion and to mean that there may be other elements / components / etc. in addition to the listed elements / components / etc.; the terms “first,” “second,” and “third,” etc., are used only as markings and are not a limitation on the number of objects.

[0033] Figure 1 A schematic diagram of an exemplary system architecture for which the object distribution graph generation method of embodiments of the present disclosure can be applied is shown.

[0034] like Figure 1 As shown, the system architecture may include server 101, network 102, terminal device 103, terminal device 104, and terminal device 105. Network 102 serves as the medium for providing a communication link between terminal device 103, terminal device 104, or terminal device 105 and server 101. Network 102 may include various connection types, such as wired or wireless communication links or fiber optic cables, etc.

[0035] Server 101 can be a server that provides various services, such as a back-end management server that supports the devices operated by users using terminal devices 103, 104, or 105. The back-end management server can analyze and process received requests and other data, and feed back the processing results to terminal devices 103, 104, or 105.

[0036] Terminal devices 103, 104, and 105 can be smartphones, tablets, laptops, desktop computers, smart speakers, wearable smart devices, virtual reality devices, augmented reality devices, etc., but are not limited to these.

[0037] In this embodiment of the disclosure, for example, terminal devices 103 and 104 can acquire operation feedback data generated by the operation objects operating the application in the terminal devices; server 101 can acquire the operation feedback data and generate a heterogeneous graph based on the operation feedback data, and obtain an object isomorphic graph based on the object nodes of multiple operation objects in the heterogeneous graph; perform graph pooling processing on the operation object isomorphic graph based on the embedding features of each object node to determine multiple target object nodes from the multiple object nodes in the operation object isomorphic graph; and generate an object distribution graph corresponding to the target object nodes. The object distribution graph generated by server 101 can be sent to terminal device 105 for display, and the item owner corresponding to terminal device 105 can make plans based on the object distribution graph (e.g., plan the location of the item warehouse).

[0038] It should be understood that Figure 1 The number of terminal devices 103, 104, 105, network 102, and server 101 in the diagram is merely illustrative. Server 101 can be a single physical server, a server cluster consisting of multiple servers, or a cloud server. Depending on actual needs, it can have any number of terminal devices, networks, and servers.

[0039] The following will describe in more detail the steps of the object distribution map generation method in the example embodiments of this disclosure, with reference to the accompanying drawings and embodiments. The method provided in the embodiments of this disclosure can be executed by any electronic device, such as the one described above. Figure 1 The server and / or terminal equipment in the process, but this disclosure does not limit this.

[0040] Figure 2 This is a flowchart illustrating an object distribution map generation method according to an exemplary embodiment.

[0041] like Figure 2 As shown, the method provided in this disclosure embodiment may include the following steps.

[0042] In step S210, operation feedback data is acquired, and a heterogeneous graph is generated based on the operation feedback data.

[0043] In this embodiment of the disclosure, operation feedback data refers to the data generated by the user's operations on the application software when using the application software. The operations performed may include, but are not limited to, watching short videos, liking, and posting comments. Operation feedback data may be, for example, the data generated by the user's operations when browsing short videos.

[0044] It should be noted that all information and data involved in this disclosure have been fully authorized by the user and all parties, and the acquisition, storage, processing and transmission of information comply with the requirements of relevant laws and regulations.

[0045] The operation feedback data in this embodiment can be multi-dimensional, meaning it can include various forms of operation feedback data. For example, multi-dimensional operation feedback data can include operation feedback data generated when browsing short videos and operation feedback data generated when purchasing items. In this embodiment, the heterogeneous graph can include object nodes of multiple operation objects.

[0046] Because operational feedback data is multi-dimensional, heterogeneous graphs can be generated based on this data. A heterogeneous graph is a network graph composed of multiple nodes of different types, and nodes connected by edges recording different types of connection relationships. For example, node types can include object nodes (e.g., user nodes), content nodes, attribute nodes (e.g., user attribute nodes), item nodes (e.g., product nodes), item attribute nodes, and item owner nodes (e.g., merchant nodes).

[0047] Among them, the object node of the operation object can be, for example, user A, user B, user C, etc.; the content node can be, for example, short video A, short video B, short video C, etc.; the attribute node of the operation object can be, for example, gender, etc.; the item node can be, for example, daily necessities, clothing, home appliances, etc.; the item attribute node can be, for example, brand, price, rating, etc.; and the item owner node can be, for example, merchant Zhang San, merchant Li Si, etc.

[0048] In an exemplary embodiment, generating a heterogeneous graph based on operation feedback data includes: generating object nodes of an operation object, attribute nodes of an operation object, item nodes, and attribute nodes of items based on the operation feedback data; and constructing edges between object nodes and attribute nodes of an operation object, between object nodes and item nodes, and between item nodes and attribute nodes of items based on the operation feedback data to generate a heterogeneous graph.

[0049] In this embodiment of the disclosure, object nodes, attribute nodes, item nodes, and item attribute nodes of an operation object can be extracted from multi-dimensional operation feedback data. For example, object nodes, item nodes, and item attribute nodes of an operation object that purchases an item can be extracted from operation feedback data generated from purchasing an item. An edge (the relationship can be purchase) between the object nodes and item nodes of the operation object can be constructed based on the operation feedback data generated from purchasing an item, as well as an edge between the item nodes and item attribute nodes. For example, object nodes and attribute nodes of an operation object that watches short videos can be extracted from operation feedback data generated from browsing short videos (e.g., liking to watch a certain type of short videos). Content nodes can also be extracted, and an edge between the object nodes and attribute nodes of the operation object can be constructed based on the operation feedback data generated from browsing short videos, as well as an edge between the object nodes and content nodes of the operation object.

[0050] In step S220, an isomorphic graph of operation objects is obtained based on the object nodes of multiple operation objects in the heterogeneous graph.

[0051] In this context, an isomorphic graph refers to a graph that includes nodes of one type, while an operation object isomorphic graph refers to a graph that only includes object nodes of the operation object type.

[0052] In this embodiment of the disclosure, there are multiple types of nodes in the heterogeneous graph. All object nodes of the operation object can be extracted from the heterogeneous graph, and an operation object isomorphic graph can be generated based on the relationship between the object nodes of the operation object. Alternatively, object nodes of all operation objects that are associated with a certain target node can be extracted from the heterogeneous graph, and an operation object isomorphic graph can be generated based on the relationship between the object nodes of the operation object. For example, user nodes that have actual purchasing behavior with the merchant node to be analyzed can be extracted to generate an operation object isomorphic graph.

[0053] In an exemplary embodiment, generating an operation object isomorphism graph based on each object node in the heterogeneous graph includes: extracting each object node in the heterogeneous graph and associating each object node with the others to generate an operation object isomorphism graph.

[0054] In this embodiment of the disclosure, the isomorphic graph of the operation objects can be a complete graph. After extracting the object nodes of each operation object in the heteromorphic graph, the object nodes of two operation objects that were not originally connected are connected. This can prevent the number of connected branches of the graph from being greater than 1 (i.e., a graph becomes multiple independent graphs) due to the deletion of some nodes during the subsequent graph pooling process.

[0055] In step S230, graph pooling is performed on the isomorphic graph of the operation object based on the embedding features of each object node, so as to determine multiple target object nodes from multiple object nodes in the isomorphic graph of the operation object and generate an object distribution graph corresponding to the target object nodes.

[0056] The target object node can be a relatively active object node.

[0057] By performing graph pooling on the isomorphic graph of the operation object based on the embedding features of each object node obtained by graph convolution operation on the heterogeneous graph, multiple target object nodes that can represent the key information of the isomorphic graph of the operation object can be determined from multiple object nodes. For example, the number of target object nodes can be set to K, where K is an integer greater than or equal to 1. The value of K can be set according to the actual situation. For example, K can be the number of warehouses set by a merchant.

[0058] In an exemplary embodiment, before performing graph pooling on the isomorphic graph based on the embedding features of each object node, the method may further include: performing graph convolution on the heterogeneous graph to obtain the embedding features of each object node in the heterogeneous graph.

[0059] In this embodiment of the disclosure, the heterogeneous graph can be processed by multi-round graph convolution, and after convergence, the embedding features of each node in the heterogeneous graph are obtained, wherein each node includes each object node.

[0060] Because of the differences in node embedding and adjacency structure, during graph convolution, non-adjacent nodes can transmit information to other nodes through their adjacency relationships, while adjacent nodes tend to have similar information. Therefore, node embedding features representing the node adjacency structure can be obtained in this way.

[0061] In an exemplary embodiment, graph convolution processing is performed on the heterogeneous graph to obtain the embedding features of each object node in the heterogeneous graph. This includes performing graph convolution processing on the heterogeneous graph so that the object node and other nodes connected to the object node can exchange information to obtain the embedding features of each object node in the heterogeneous graph.

[0062] Specifically, graph convolution processing is performed on the heterogeneous graph to enable information transfer between object nodes and attribute nodes of operation objects connected to the object nodes, between item nodes and attribute nodes of connected items, and between object nodes of operation objects and connected item nodes, so as to obtain the embedding features of each object node in the heterogeneous graph.

[0063] In this embodiment of the disclosure, the adjacency matrix of the heterogeneous graph can be determined first, and multiple graph convolution operations can be performed on the heterogeneous graph based on the adjacency matrix to obtain the embedding matrix of the heterogeneous graph; based on the position of each object node in the embedding matrix, the embedding features of each object node can be extracted from the embedding matrix.

[0064] In an exemplary embodiment, an initial parameter matrix of an object isomorphic graph can be generated based on the embedding features of each object node, and the attention weights of each object node can be obtained based on the initial parameter matrix; the initial parameter matrix can be filtered based on the attention weights of each object node to obtain a target parameter matrix; and multiple target object nodes can be determined from multiple object nodes based on the target parameter matrix.

[0065] In this embodiment of the disclosure, after determining multiple target object nodes, the location information of each target operation object node can be determined, and an object distribution map corresponding to the target object node can be generated based on the location information.

[0066] Location information can refer to the geographical location of the object being operated on; the object distribution map refers to a distribution map used to characterize the geographical distribution of the object being operated on. The object distribution map can be a distribution map of various forms, such as a distribution heat map or a distribution histogram.

[0067] In this embodiment of the disclosure, the index and location information of each object node can be preset. After the target object node is determined, the location information corresponding to the index of the target object node can be queried, and an operation object distribution map can be generated based on the location information of each target object node.

[0068] The object distribution graph generation method provided in this disclosure generates a heterogeneous graph based on operation feedback data, obtains an object isomorphic graph based on the object nodes of multiple operation objects in the heterogeneous graph, and performs graph pooling processing on the operation object isomorphic graph based on the embedding features of each object node. On the one hand, it can quickly filter out operation objects with low activity, thus saving computer resources when generating the object distribution graph; on the other hand, it can more accurately determine the target object nodes with high activity from multiple object nodes, making the generated object distribution graph more accurate.

[0069] In some embodiments, a heterogeneous graph is generated based on multi-dimensional operation feedback data, and graph convolution processing is performed on the heterogeneous graph to obtain the embedding features of each object node. By comprehensively considering the multi-dimensional data, the obtained embedding features of each object node can more comprehensively and accurately represent the activity of each object node, thereby generating an object distribution graph representing comprehensive activity information.

[0070] Furthermore, the object distribution map generation method provided in this disclosure allows item owners to lay out their warehouses based on the object distribution map, thereby saving logistics resources and ensuring timely delivery. For video platforms, it can recommend similar videos for operation objects that are close in location on the object distribution map, making the recommended videos more accurate.

[0071] Figure 3 This is a flowchart illustrating another method for generating an object distribution map according to an exemplary embodiment. Figure 3 The paper illustrates the specific steps of performing graph pooling on an object isomorphic graph based on the embedding features of each object node to determine multiple target object nodes from multiple object nodes in the object isomorphic graph.

[0072] exist Figure 3 In the embodiments, the above Figure 2 Step S230 in the embodiment may further include the following steps.

[0073] In step S231, an initial parameter matrix of the object isomorphism graph is generated based on the embedding features of each object node, and the attention weight of each object node is obtained based on the initial parameter matrix.

[0074] In this embodiment of the disclosure, reference is made to Figure 4 The first parameter matrix X can be generated based on the embedding features of each object node obtained by performing graph convolution on the heterogeneous graph. l Its corresponding adjacency matrix is ​​A l , where l is an integer greater than or equal to 1, l can represent the number of graph pooling operations, and the first parameter matrix can be the initial parameter matrix (i.e., when l = 1).

[0075] Specifically, the initial parameter matrix is ​​subjected to self-attention processing to obtain the attention weights of each object node.

[0076] In this embodiment of the disclosure, the first parameter matrix X can be adjusted using a self-attention mechanism. l (e.g., initial parameter matrix X) 1 The matrix is ​​processed to obtain a matrix representing the mutual information between each object node; this matrix is ​​multiplied by a learnable parameter matrix p, and then normalized using the sigmoid function to obtain the attention weights of each object node.

[0077] refer to Figure 4 For the first parameter matrix X l (e.g., initial parameter matrix X) 1 After self-attention processing, the attention weights of each object node are obtained as W1, W2, W3, and W4, respectively.

[0078] In step S232, the initial parameter matrix is ​​filtered out according to the attention weight of each object node to obtain the target parameter matrix.

[0079] In this embodiment of the disclosure, for each graph pooling operation, the initial parameter matrix can be filtered out based on a pre-set numerical or percentage value and the attention weight of each object node to obtain the target parameter matrix.

[0080] For example, refer to Figure 4 K can be set to 50% or 2. Based on the attention weight of the operation objects, retain the top 50% (or top 2) of operation object nodes with higher attention weights, and filter out object nodes with lower attention weights. For example, if W1 and W3 are high, then retain the object nodes with attention weights of W1 and W3, thus obtaining the target parameter matrix X. l +1 .

[0081] In this embodiment of the disclosure, the parameter matrix obtained by performing graph pooling on the isomorphic graph of the operation object once can be used as the target parameter matrix, or the target parameter matrix can be obtained by performing graph pooling on the isomorphic graph of the operation object multiple times.

[0082] In an exemplary embodiment, the initial parameter matrix is ​​sieved based on the attention weights of each object node to obtain the target parameter matrix. This includes: determining the index of the node to be sieved based on the attention weights of each operation object; and sieving the initial parameter matrix based on the index of the node to be sieved to obtain the target parameter matrix.

[0083] In this embodiment of the disclosure, an initial adjacency matrix corresponding to an initial parameter matrix can be determined; the index of the node to be screened can be determined according to the attention weight of each object node; the initial parameter matrix can be screened based on the index of the node to be screened to obtain an intermediate parameter matrix, and the initial adjacency matrix can be screened to obtain an intermediate adjacency matrix; the intermediate parameter matrix can be graph convolutional based on the intermediate adjacency matrix to obtain a target parameter matrix.

[0084] Specifically, refer to Figure 4 Determine the initial parameter matrix X l The corresponding initial adjacency matrix is ​​A l Based on the attention weights of each object node, the indices of the nodes to be removed are determined to be 2 and 4. Based on these indices, the initial parameter matrix X is then... l Perform a sieving process to delete the initial parameter matrix X. l The vectors in rows 2 and 4 are used to obtain the intermediate parameter matrix X. l +1Based on the indices 2 and 4 of the nodes to be filtered out, the initial adjacency matrix A is... l Perform a filtering process to delete the initial adjacency matrix A. l The vectors in row 2, row 4, column 2, and column 4 are used to obtain the intermediate adjacency matrix A. l+1 Based on the intermediate adjacency matrix A l+1 For the intermediate parameter matrix X l+1 Perform graph convolution to obtain the target parameter matrix; after obtaining the target parameter matrix, continue to perform graph pooling on the target parameter matrix using the same method as above until the final target parameter matrix is ​​obtained.

[0085] In step S233, multiple target object nodes are determined from multiple object nodes based on the target parameter matrix.

[0086] In this embodiment of the disclosure, the object node corresponding to the vector in the target parameter matrix is ​​determined as the target object node.

[0087] refer to Figure 4 For example, if X l+1 If the target parameter matrix is ​​defined, then the 2nd and 4th object nodes are determined as the target object nodes.

[0088] It should be noted that, Figure 4 The example given is that the initial parameter matrix is ​​4-dimensional (i.e., there are 4 operation object nodes). However, in practical applications, there can be tens of thousands or hundreds of thousands of operation objects, i.e., there can be tens of thousands or hundreds of thousands of object nodes. This disclosure does not limit this.

[0089] The object distribution map generation method provided in this disclosure can extract key information from the isomorphic graph of the operation objects using graph pooling, while removing object nodes with low information contribution. On the one hand, it can quickly filter out operation objects with low activity, thus saving computer resources when generating the operation object distribution map; on the other hand, it can more accurately determine the target object nodes with high activity from multiple object nodes, making the generated operation object distribution map more accurate.

[0090] Figure 5 This is a flowchart illustrating another method for generating an object distribution map according to an exemplary embodiment. Figure 5 The specific steps for generating the object distribution map corresponding to the target operation object are shown.

[0091] In an exemplary embodiment, the object distribution map includes an object distribution heatmap.

[0092] exist Figure 5 In the embodiments, the above Figure 2 Step S230 in the embodiment may further include the following steps.

[0093] In step S234, the location information of each target object node is determined.

[0094] Location information can refer to the geographical location of the object being operated on.

[0095] In this embodiment of the disclosure, the index and location information of each object node can be preset. After the target object node is determined, the location information corresponding to the index of the target object node can be queried.

[0096] In step S235, the target coordinate range corresponding to the target object node is determined with the position information of the target object node as the center.

[0097] In this embodiment of the disclosure, the target coordinate range corresponding to the position information of each target object node can be obtained by taking the position information of each target object node as the center and the preset influence range R as the radius.

[0098] Figure 6 This is a schematic diagram illustrating the target coordinate range corresponding to the position information of a target object node, as shown in the example. (Reference) Figure 6 Each circle can represent the target coordinate range corresponding to the position information of a target object node.

[0099] In step S236, the weight of each target coordinate point is determined based on the distance of each target coordinate point from the center within the target coordinate range.

[0100] In this embodiment of the disclosure, for each target coordinate range, the target coordinate points that are closer to the center of the target coordinate range can be assigned a larger weight, and the target coordinate points that are farther away from the center of the target coordinate range can be assigned a smaller weight.

[0101] In step S237, the weights of each target coordinate point corresponding to each target object node are superimposed to generate an object distribution heatmap.

[0102] Among them, the operation distribution heatmap refers to a graph that intuitively shows the number of operation objects in a region through the shades of color. Generally, the darker the color, the more operation objects there are, and the lighter the color, the fewer operation objects there are.

[0103] In this embodiment of the disclosure, all target coordinate points can be distributed in the graph, and the weights of each target coordinate point can be superimposed. Obviously, the more superimposed layers there are, the greater the weight of the superimposed points that are closer to the thermal center.

[0104] In this embodiment of the disclosure, different RGB (Red Green Blue, the three primary colors of light) color information can be configured for different weight ranges to generate an object distribution heatmap that can represent the comprehensive activity information of the operation object.

[0105] If the weight of the superimposed part of the heatmap is too large, overflow may occur during the mapping to RGB. Therefore, a buffer can be set up to save the value of each superimposed part. Finally, the value of this part is mapped to a floating-point number in [0,1] as a scaling factor and multiplied with the color representation range of RGB, thereby avoiding the loss of heatmap information due to numerical overflow.

[0106] Figure 7 This is a schematic diagram based on an example of an object distribution heatmap. (Reference) Figure 7 The left side shows the generated object distribution heatmap, and the right side shows a schematic diagram after overlaying the generated object distribution heatmap with the actual map.

[0107] In this embodiment of the disclosure, an object heat map is generated based on the location information of each target object node, which can more intuitively represent the distribution of active operation objects.

[0108] It should also be understood that the above description is only intended to help those skilled in the art better understand the embodiments of this disclosure, and is not intended to limit the scope of the embodiments of this disclosure. Those skilled in the art can obviously make various equivalent modifications or changes based on the examples given above. For example, some steps in the above methods may be unnecessary, or new steps may be added, etc. Alternatively, any combination of two or more of the above embodiments may be used. Such modifications, changes, or combinations also fall within the scope of the embodiments of this disclosure.

[0109] It should also be understood that the above description of the embodiments of this disclosure focuses on highlighting the differences between the various embodiments. Similarities or differences not mentioned can be referred to each other, and for the sake of brevity, they will not be repeated here.

[0110] It should also be understood that the sequence number of each process does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this disclosure.

[0111] It should also be understood that, in the various embodiments of this disclosure, unless otherwise specified or in case of logical conflict, the terminology and / or descriptions between different embodiments are consistent and can be referenced by each other, and the technical features in different embodiments can be combined to form new embodiments according to their inherent logical relationships.

[0112] The foregoing section detailed examples of object distribution map generation methods provided in this disclosure. It is understood that, in order to implement the aforementioned functions, computer devices include corresponding hardware structures and / or software modules for performing each function. Those skilled in the art should readily recognize that, based on the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein, this disclosure can be implemented in hardware or a combination of hardware and computer software. Whether a function is implemented in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this disclosure.

[0113] The following are embodiments of the apparatus disclosed herein, which can be used to execute embodiments of the method disclosed herein. For details not disclosed in the apparatus embodiments of this disclosure, please refer to the embodiments of the method disclosed herein.

[0114] Figure 8 This is a block diagram illustrating an object distribution map generation apparatus according to an exemplary embodiment. (Refer to...) Figure 8 The object distribution map generation device 800 may include an acquisition module 810, an obtaining module 820, and a determining module 830.

[0115] The acquisition module 810 is configured to acquire operation feedback data and generate a heterogeneous graph based on the operation feedback data; the acquisition module 820 is configured to obtain an object isomorphic graph based on the object nodes of multiple operation objects in the heterogeneous graph; and the determination module 830 is configured to perform graph pooling processing on the object isomorphic graph based on the embedding features of each object node, so as to determine multiple target object nodes from the multiple object nodes in the object isomorphic graph and generate an object distribution graph corresponding to the target object nodes.

[0116] In some exemplary embodiments of this disclosure, the determining module 830 is configured to perform: generating an initial parameter matrix of the object isomorphism graph based on the embedding features of each object node, and obtaining the attention weights of each object node according to the initial parameter matrix; performing a filtering process on the initial parameter matrix according to the attention weights of each object node to obtain a target parameter matrix; and determining a plurality of target object nodes from the plurality of object nodes according to the target parameter matrix.

[0117] In some exemplary embodiments of this disclosure, the determining module 830 is configured to perform: determining the index of the node to be screened based on the attention weight of each operation object; and performing a screening process on the initial parameter matrix based on the index of the node to be screened to obtain the target parameter matrix.

[0118] In some exemplary embodiments of this disclosure, the object distribution map includes an object distribution heatmap; wherein, the determining module 830 is configured to perform: determining the location information of each target object node; determining the target coordinate range corresponding to the target object node with the location information of the target object node as the center; determining the weight of each target coordinate point according to the distance of each target coordinate point in the target coordinate range from the center; and superimposing the weights of each target coordinate point corresponding to each target object node to generate the object distribution heatmap.

[0119] In some exemplary embodiments of this disclosure, the obtaining module 820 is configured to perform: extracting each object node in the heterogeneous graph and associating the object nodes with each other to generate the object isogeneous graph.

[0120] In some exemplary embodiments of this disclosure, the obtaining module 820 is configured to perform graph convolution processing on the heterogeneous graph to obtain the embedding features of each object node in the heterogeneous graph.

[0121] In some exemplary embodiments of this disclosure, the obtaining module 820 is configured to perform graph convolution processing on the heterogeneous graph, such that the object node and other nodes connected to the object node exchange information to obtain the embedding features of each object node in the heterogeneous graph.

[0122] It should be noted that the block diagrams shown in the above figures are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor terminal devices and / or microcontroller terminal devices.

[0123] Regarding the apparatus in the above embodiments, the specific manner in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated upon here.

[0124] The following reference Figure 9 To describe an electronic device 900 according to such an embodiment of the present disclosure. Figure 9 The electronic device 900 shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments disclosed herein.

[0125] like Figure 9As shown, the electronic device 900 is presented in the form of a general-purpose computing device. The components of the electronic device 900 may include, but are not limited to: at least one processing unit 910, at least one storage unit 920, a bus 930 connecting different system components (including storage unit 920 and processing unit 910), and a display unit 940.

[0126] The storage unit stores program code, which can be executed by the processing unit 910 to perform the steps described in the "Exemplary Methods" section of this specification according to various exemplary embodiments of this disclosure. For example, the processing unit 910 can perform actions such as... Figure 2 The steps shown are as follows.

[0127] For example, electronic devices can achieve such Figure 2 The steps shown.

[0128] Storage unit 920 may include readable media in the form of volatile storage units, such as random access memory (RAM) 921 and / or cache memory 922, and may further include read-only memory (ROM) 923.

[0129] The storage unit 920 may also include a program / utility 924 having a set (at least one) of program modules 925, including but not limited to: an operating system, one or more application programs, other program modules, and program data, each or some combination of these examples may include an implementation of a network environment.

[0130] Bus 930 can represent one or more of several types of bus structures, including a memory cell bus or memory cell controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of the various bus structures.

[0131] Electronic device 900 can also communicate with one or more external devices 970 (e.g., keyboard, pointing device, Bluetooth device, etc.), and with one or more devices that enable a user to interact with electronic device 900, and / or with any device that enables electronic device 900 to communicate with one or more other computing devices (e.g., router, modem, etc.). This communication can be performed via input / output (I / O) interface 950. Furthermore, electronic device 900 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 960. As shown, network adapter 960 communicates with other modules of electronic device 900 via bus 930. It should be understood that, although not shown in the figures, other hardware and / or software modules can be used in conjunction with electronic device 900, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0132] From the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, terminal device, or network device, etc.) to execute the methods according to the embodiments of this disclosure.

[0133] In an exemplary embodiment, a computer-readable storage medium including instructions is also provided, such as a memory including instructions that can be executed by a processor of the device to perform the described method. Optionally, the computer-readable storage medium may be a ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc.

[0134] In an exemplary embodiment, a computer program product is also provided, including a computer program / instructions that, when executed by a processor, implement the object distribution map generation method described above.

[0135] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the following claims.

[0136] It should be understood that this disclosure is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this disclosure is limited only by the appended claims.

Claims

1. A method for generating an object distribution map, characterized in that, include: Obtain operation feedback data and generate a heterogeneous graph based on the operation feedback data; An object isomorphism graph is obtained based on the object nodes of multiple operational objects in the heterogeneous graph; The object isomorphic graph is subjected to graph pooling based on the embedding features of each object node, so as to determine multiple target object nodes from multiple object nodes in the object isomorphic graph and generate an object distribution graph corresponding to the target object nodes. The step of performing graph pooling on the object isomorphic graph based on the embedding features of each object node to determine multiple target object nodes from multiple object nodes in the object isomorphic graph includes: The initial parameter matrix and the corresponding initial adjacency matrix of the object isomorphism graph are generated based on the embedding features of each object node. The attention weights of each object node are obtained based on the initial parameter matrix. The index of the node to be removed is determined based on the attention weight of each operation object. Based on the index of the node to be filtered out, the initial parameter matrix is ​​filtered to obtain the intermediate parameter matrix; Based on the index of the node to be filtered out, the initial adjacency matrix is ​​filtered to obtain the intermediate adjacency matrix; The target parameter matrix is ​​obtained by performing graph convolution on the intermediate parameter matrix based on the intermediate adjacency matrix. Multiple target object nodes are determined from the multiple object nodes based on the target parameter matrix; The object distribution map includes an object distribution heatmap; generating the object distribution map corresponding to the target object node includes: Determine the location information of each target object node; Using the location information of the target object node as the center, determine the target coordinate range corresponding to the target object node; The weight of each target coordinate point is determined based on the distance of each target coordinate point from the center within the target coordinate range; The weights of each target coordinate point corresponding to each target object node are superimposed to generate the object distribution heatmap.

2. The method according to claim 1, characterized in that, Based on the object nodes of multiple operational objects in the heterogeneous graph, an object isomorphic graph is obtained, including: Extract each object node from the heterogeneous graph and associate each object node with another to generate the isogeneous graph of objects.

3. The method according to claim 1, characterized in that, Before performing graph pooling on the object isomorphic graph based on the embedding features of each object node, the method further includes: The heterogeneous graph is subjected to graph convolution processing to obtain the embedding features of each object node in the heterogeneous graph.

4. The method according to claim 3, characterized in that, The heterogeneous graph is subjected to graph convolution processing to obtain the embedding features of each object node in the heterogeneous graph, including: The heterogeneous graph is subjected to graph convolution processing, which enables the object node and other nodes connected to the object node to exchange information, so as to obtain the embedding features of each object node in the heterogeneous graph.

5. An object distribution map generation device, characterized in that, include: The acquisition module is configured to acquire operation feedback data and generate a heterogeneous graph based on the operation feedback data; The module is configured to obtain an object isomorphic graph based on the object nodes of multiple operational objects in the heterogeneous graph; The determination module is configured to perform graph pooling processing on the object isomorphism graph based on the embedding features of each object node, so as to determine multiple target object nodes from multiple object nodes in the object isomorphism graph and generate an object distribution graph corresponding to the target object nodes. The determining module is further configured to generate an initial parameter matrix and a corresponding initial adjacency matrix of the object isomorphism graph based on the embedding features of each object node; and to obtain the attention weights of each object node based on the initial parameter matrix. The index of the node to be removed is determined based on the attention weight of each operation object. Based on the index of the node to be filtered out, the initial parameter matrix is ​​filtered to obtain the intermediate parameter matrix; Based on the index of the node to be filtered out, the initial adjacency matrix is ​​filtered to obtain the intermediate adjacency matrix; The intermediate parameter matrix is ​​subjected to graph convolution based on the intermediate adjacency matrix to obtain the target parameter matrix; multiple target object nodes are determined from the multiple object nodes based on the target parameter matrix. The object distribution map includes an object distribution heatmap; the determining module is further configured to determine the location information of each target object node. Using the location information of the target object node as the center, determine the target coordinate range corresponding to the target object node; determine the weight of each target coordinate point according to the distance of each target coordinate point in the target coordinate range from the center; The weights of each target coordinate point corresponding to each target object node are superimposed to generate the object distribution heatmap.

6. An electronic device, characterized in that, include: processor; Memory used to store the processor's executable instructions; The processor is configured to execute the executable instructions to implement the method as described in any one of claims 1 to 4.

7. A computer-readable storage medium, wherein instructions in the computer-readable storage medium, when executed by a processor of an electronic device, enable the electronic device to perform the method as described in any one of claims 1 to 4.

8. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 4.

Images