Method, device, storage medium and program product for recommending item information to users

Abstract

The method, device, storage medium, and program product provided by this application for recommending item information to users relate to data processing technology, including: obtaining a preset topological graph used to represent the relationship between users and items, and update data used to update the preset topological graph; updating the edges in the preset topological graph based on the update data to obtain an updated topological graph; obtaining a pre-trained vector model corresponding to the preset topological graph, and updating the vector of any node in the topological graph based on the updated topological graph and the obtained vector model, where the vector is used to represent the characteristics of the node; determining similar item nodes to the user node based on the vector of each node, and recommending item information of the similar item nodes to the user corresponding to the user node. The solution provided by this application can quickly update the node vector based on the update data, and thus can more accurately recommend information to users based on the updated node vector.

Description

Technical Field

[0001] The present application relates to data processing technology, and in particular to a method, device, storage medium, and program product for recommending item information to users. Background Art

[0002] In order to provide users with more abundant network information, information can be actively pushed to users, for example, product information and information that users may be interested in.

[0003] In order to push network information of interest to users, there are personalized recommendation methods in the prior art. By vectorizing users and information, the information of interest to each user can be determined and then recommended to the user.

[0004] To record the relationship between users and information, a network topology graph can be used to store this relationship. However, the complexity of this graph often makes real-time recommendations difficult. Therefore, improving the timeliness of personalized recommendations is a technical problem that technicians in this field urgently need to solve. Summary of the Invention

[0005] The present application provides a method, device, storage medium and program product for recommending item information to users, so as to solve the problem of poor timeliness of personalized recommendation in the prior art.

[0006] A first aspect of the present application is to provide a method for recommending item information to a user, comprising:

[0007] Obtaining a preset topological graph for representing the association relationship between users and items, and update data for updating the preset topological graph; wherein the preset topological graph includes nodes, the nodes include user nodes and item nodes, and the user nodes and item nodes connected by edges have an association relationship;

[0008] updating the edges in the preset topology graph according to the update data to obtain an updated topology graph;

[0009] Obtaining a vector model pre-trained using an initial topological graph corresponding to the preset topological graph, and updating a vector of any node in the topological graph according to the updated topological graph and the obtained vector model, wherein the vector is used to represent a feature of the node;

[0010] Based on the vectors of each of the nodes, similar item nodes of the user node are determined, and item information of the similar item nodes is recommended to the user corresponding to the user node.

[0011] A second aspect of the present application is to provide a device for personalized recommendation information, comprising:

[0012] an acquisition unit, configured to acquire a preset topological graph representing the association relationship between users and items, and update data for updating the preset topological graph; wherein the preset topological graph includes nodes, the nodes including user nodes and item nodes, and the user nodes and item nodes connected by edges have an association relationship;

[0013] A graph updating unit, configured to update edges in the preset topology graph according to the update data to obtain an updated topology graph;

[0014] a vector updating unit, configured to obtain a pre-trained vector model corresponding to the preset topology map, and update a vector of any node in the topology map based on the updated topology map and the obtained vector model, wherein the vector is used to characterize the characteristics of the node;

[0015] A recommendation unit is configured to determine similar item nodes of a user node based on the vectors of each node, and recommend item information of the similar item nodes to a user corresponding to the user node. Another aspect of the present application is to provide an electronic device comprising:

[0016] Memory;

[0017] processor; and

[0018] computer programs;

[0019] The computer program is stored in the memory and is configured to be executed by the processor to implement the method for recommending item information to a user as described in the first aspect above.

[0020] Yet another aspect of the present application is to provide a computer-readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to implement the method for recommending item information to a user as described in the first aspect above.

[0021] Yet another aspect of the present application is to provide a computer program product, comprising a computer program, which, when executed by a processor, implements the method for recommending item information to a user as described in the first aspect.

[0022] The technical effects of the method, device, storage medium, and program product provided by this application for recommending item information to users are:

[0023] The present application provides a method, device, storage medium, and program product for recommending item information to users, including: obtaining a preset topological map for representing the relationship between users and items, and update data for updating the preset topological map; wherein the preset topological map includes nodes, including user nodes and item nodes, and the user nodes and item nodes connected by edges have an association relationship; updating the edges in the preset topological map based on the update data to obtain an updated topological map; obtaining a pre-trained vector model corresponding to the preset topological map, and updating the vector of any node in the topological map based on the updated topological map and the obtained vector model, wherein the vector is used to represent the characteristics of the node; determining similar item nodes to the user node based on the vector of each node, and recommending item information of the similar item nodes to the user corresponding to the user node. In the solution provided by the present application, a preset topological map and a vector model corresponding to the preset topological map can be pre-set. When updated data is available, the preset topological map can be directly updated based on the updated data, and the updated topological map can be processed using the vector model to obtain the updated node vector. This method can quickly update the node vector based on the updated data, and thus can more accurately recommend information to users based on the updated node vector. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 A schematic diagram of a transductive graph embedding algorithm architecture is shown as an exemplary embodiment;

[0025] Figure 2 A schematic diagram of an inductive graph embedding algorithm architecture is shown as an exemplary embodiment;

[0026] Figure 3 This is a flowchart of a method for personalized recommendation information shown as an exemplary embodiment of the present application;

[0027] Figure 4 This is a schematic diagram of a collection behavior subgraph shown in an exemplary embodiment of the present application;

[0028] Figure 5 This is a flowchart of a method for personalized recommendation information shown in another exemplary embodiment of the present application;

[0029] Figure 6 A schematic diagram of an initial topology diagram shown in an exemplary embodiment of the present application;

[0030] Figure 7 A schematic diagram of a behavior subgraph shown as an exemplary embodiment of the present application;

[0031] Figure 8 This is a structural diagram of a device for personalized recommendation information shown as an exemplary embodiment of the present application;

[0032] Figure 9This is a structural diagram of an apparatus for personalized recommendation information shown in another exemplary embodiment of the present application;

[0033] Figure 10 This is a structural diagram of an electronic device shown as an exemplary embodiment of the present application. DETAILED DESCRIPTION

[0034] In personalized recommendation solutions, in order to achieve near real-time recommendation results, based on full training, it is necessary to promptly update the user's vector results according to the user's new interactive behavior, so that when making recommendations, the user's recent behavior can be combined to provide minute-level effect feedback.

[0035] There are currently two types of graph embedding methods: one is the direct graph embedding algorithm, such as GCN (Graph Convolutional Network), and the other is the inductive graph embedding algorithm, such as GRAPHSAGE (Graph SAmple and aggreGatE).

[0036] Figure 1 The figure is a schematic diagram of the architecture of a transductive graph embedding algorithm, showing an exemplary embodiment.

[0037] As shown in 1, first collect all offline data, which includes user behavior, user attributes, and item attributes. Through graph construction, model users and items as nodes in the graph.

[0038] User behavior is used to build connections between nodes, while user and item attributes serve as features of nodes in the graph. After constructing the full graph, algorithms such as GCN utilize the entire adjacency matrix and graph convolution operations to fuse information from adjacent nodes and generate vector representations of the nodes.

[0039] When new user behavior occurs, the graph must be rebuilt based on the full offline data and incremental data, and vector representations of each node must be generated based on the graph. This approach is slow in updating user vector representations and cannot quickly adjust the graph based on incremental data. Therefore, it cannot support near-real-time architecture updates.

[0040] Figure 2 The figure is a schematic diagram of the inductive graph embedding algorithm architecture shown as an exemplary embodiment.

[0041] like Figure 2 As shown, first, the full amount of offline data is collected, where the full data includes data such as user behavior, user attributes, and item attributes. Users and items are modeled as nodes in the graph through graph construction.

[0042] Taking GRAPHSAGE as an example, training is performed on the constructed graph to store the mapping relationships between nodes. A node's neighbors can be processed based on the mapping function to determine the node's vector representation. However, inductive graph embedding algorithms are primarily used for offline training of all nodes and cannot be used directly for personalized information recommendations.

[0043] In order to solve the above technical problems, the present application provides a solution for recommending item information to users. In this solution, a topological map can be constructed in advance, and a vector model can be pre-trained. When there is updated data, the topological map can be directly updated using the updated data, and the updated map can be processed based on the vector model to determine the updated vector of the node. This method can quickly update the vector of the node according to the updated data, and then personalized recommendation information can be provided to the user based on the updated vector, thereby improving the timeliness of personalized recommendations.

[0044] Figure 3 The flowchart of a method for recommending item information to a user is shown as an exemplary embodiment of the present application.

[0045] like Figure 3 As shown, the method for recommending item information to a user provided by this application includes:

[0046] Step 301: Obtain a preset topology graph for representing the association relationship between users and items, and update data for updating the preset topology graph; wherein the preset topology graph includes nodes, the nodes include user nodes and item nodes, and there is an association relationship between the user nodes and the item nodes connected by edges.

[0047] The method provided in this application can be executed by one or more electronic devices with computing capabilities, such as a server, such as a cloud server or a distributed server. Specifically, the server can provide information recommendation functions, such as recommending products to users.

[0048] Furthermore, the electronic device may be triggered to execute the method of this embodiment when update data is generated, or the method of this embodiment may be executed once every preset time period.

[0049] In actual application, the electronic device can obtain a preset topological map for characterizing the relationship between users and items. The preset topological map includes multiple nodes, including user nodes and item nodes. There are edges between the user nodes and the item nodes, thereby connecting the user nodes and the item nodes through the edges.

[0050] A topology graph consisting of multiple nodes can be pre-built based on offline data. For example, if a piece of offline data includes user A accessing information B, nodes A and B can be constructed, and an edge between A and B can be constructed to indicate the relationship between A and B.

[0051] Specifically, the weight value of the edge can be set according to needs.

[0052] Furthermore, the electronic device can also obtain update data, which may include user behavior data. For example, if a user browses a certain product or accesses a certain piece of information, the electronic device can obtain the relevant information.

[0053] In actual application, the updated data obtained is data used to update the preset topology map.

[0054] Step 302: Update the edges in the preset topology graph according to the updated data to obtain an updated topology graph.

[0055] In actual application, the electronic device can update the acquired preset topology map according to the acquired update data to obtain an updated topology map.

[0056] In one embodiment, the updated topology map may be used as a preset topology map and acquired again.

[0057] The electronic device can add edges to the preset topology graph based on the updated data to update the topology graph, and can also update the weight values ​​of the edges to further update the topology graph. For example, if the updated data includes that user A browsed product C, an edge between A and C can be constructed, thereby updating the preset topology graph.

[0058] Step 303: obtain a pre-trained vector model corresponding to a preset topology map, and update the vector of any node in the topology map according to the updated topology map and the obtained vector model. The vector is used to represent the characteristics of the node.

[0059] Specifically, the vector model may be trained in advance using an initial topology map corresponding to a preset topology map. The preset topology map may be the same as the initial topology map, or may be obtained by updating the initial topology map once or multiple times.

[0060] For example, after constructing a topology map based on offline data, training can be performed based on the topology map to obtain a vector model. Since the preset topology map is also obtained by processing offline data, the preset topology map has a corresponding relationship with the above-mentioned vector model.

[0061] Furthermore, the GRAPHSAGE algorithm can be used to train the initial topology graph to generate a vector model. Specifically, a behavioral subgraph can be collected for each node. This subgraph includes the node, its neighboring nodes, and the relationships between these nodes. For example, corresponding behavioral subgraphs can be collected for nodes P1, P2, and P3. The electronic device can then learn the relationships between neighboring nodes and the central node in these subgraphs to generate a vector model.

[0062] In practical applications, after updating a preset topology, the electronic device can also obtain a vector model corresponding to the preset topology. Based on the updated topology and the obtained vector model, the vector of any node in the topology can be updated. A node's vector can characterize the node and can be generated based on information about the node itself and its neighboring nodes.

[0063] Specifically, the relationship between nodes in the updated topology graph has changed. Therefore, the behavior subgraph of each node may be collected again, or only the behavior subgraph of the node with which the edge between the node and the neighboring node has changed may be collected.

[0064] Figure 4 This is a schematic diagram of a collection behavior subgraph shown in an exemplary embodiment of the present application.

[0065] like Figure 4 As shown, for example, there is a topology structure as shown in FIG41, wherein the behavior subgraph of node 42 is shown as 43. For example, if the updated topology is 44, then the behavior subgraph of node 42 is shown as 45.

[0066] In one implementation, the behavior subgraph of another node 46 does not change, and therefore, the vector of the node 46 does not need to be updated, and further, the behavior subgraph of the node 46 does not need to be collected.

[0067] Furthermore, the updated vector of the node can be determined based on the updated behavior subgraph of the node and the vector model. For example, the node and its neighboring node information can be input into the vector model so that the vector model outputs the updated vector of the node.

[0068] Step 304: Determine similar item nodes to the user node based on the vectors of each node, and recommend item information of the similar item nodes to the user corresponding to the user node.

[0069] In practical applications, electronic devices can determine similar nodes between nodes based on the vectors of each node. For example, when recommending information to user A, the similarity between user A's node and other nodes can be determined to obtain similar nodes to user A's node.

[0070] The electronic device can also recommend item information to any node based on its similar nodes. For example, if the node of user A is highly similar to the node of item D, item D's information can be recommended to user A. For another example, if the node of user A is highly similar to the node of user E, item information can be recommended to user A based on user E's information.

[0071] The method provided in this embodiment is used to recommend item information to a user. The method is executed by a device equipped with the method provided in this embodiment, and the device is generally implemented in hardware and / or software.

[0072] The method for recommending item information to a user provided in this embodiment includes: obtaining a preset topological graph for representing the relationship between users and items, and update data for updating the preset topological graph; wherein the preset topological graph includes nodes, including user nodes and item nodes, and the user nodes and item nodes connected by edges have an association relationship; updating the edges in the preset topological graph based on the update data to obtain an updated topological graph; obtaining a pre-trained vector model corresponding to the preset topological graph, and updating the vector of any node in the topological graph based on the updated topological graph and the obtained vector model, wherein the vector is used to represent the characteristics of the node; determining similar item nodes to the user node based on the vector of each node, and recommending item information of the similar item nodes to the user corresponding to the user node. In the method provided in this application, a preset topological graph and a vector model corresponding to the preset topological graph can be pre-set. When updated data is available, the preset topological graph can be directly updated based on the updated data, and the updated topological graph can be processed using the vector model to obtain the updated node vector. This method can quickly update the node vector based on the updated data, and thus can more accurately recommend item information to users based on the updated node vector.

[0073] Figure 5 The flowchart of a method for recommending item information to a user is shown as another exemplary embodiment of the present application.

[0074] like Figure 5 As shown, the method for recommending item information to a user provided by this application includes:

[0075] Step 501: Obtain preset offline data and construct an initial topology map based on the offline data.

[0076] Offline data can be prepared in advance, for example, the last seven days of offline data, and an initial topology graph can be constructed based on this offline data. The constructed initial topology graph includes multiple nodes, such as user nodes and item nodes, with edges between these nodes, representing an association between the two nodes. For example, an edge between user node A and user node B can represent an association between user A and user B. For another example, an edge between user node A and item node C represents an association between user A and item C.

[0077] Specifically, each piece of offline data can be parsed to extract behavioral data. For example, if user A browses item C, the electronic device can construct a node for user A, a node for item C, and an edge between user A and item C. In this way, the electronic device can construct an initial topological graph consisting of multiple nodes and the relationships between them.

[0078] Furthermore, by using the most recent offline data to construct the initial topology map, the initial topology map can include the most recent information of each user, such as information about items recently clicked by the user, thereby making the initial topology map more timely.

[0079] Figure 6 This is a schematic diagram of an initial topology diagram shown in an exemplary embodiment of the present application.

[0080] like Figure 6 As shown, the initial topology graph includes multiple nodes, and the nodes are connected by edges, thereby representing that there is an association relationship between two nodes.

[0081] Step 502: Determine a behavior subgraph of each node in the initial topology graph.

[0082] Step 503: Perform training based on the behavior subgraph to obtain a vector model corresponding to the preset topology graph.

[0083] In practical applications, after constructing an initial topology map, the electronic device can train the initial topology map to obtain a vector model. Specifically, the electronic device can learn the relationship between nodes and further obtain the relationship between a node and its neighboring nodes.

[0084] For each node in the initial topology graph, the electronic device can obtain a corresponding behavior subgraph. The behavior subgraph may include a central node, its adjacent nodes, and the relationships between these nodes.

[0085] Figure 7 This is a schematic diagram of a behavior subgraph according to an exemplary embodiment of the present application.

[0086] like Figure 7As shown, the initial topology graph can be processed to collect a behavior subgraph of the central node 71. The behavior subgraph includes multiple neighboring nodes of the central node 71.

[0087] Specifically, the order k can be pre-set to determine the neighboring nodes of each central node based on k. For example, if k = 1, the nodes directly connected to the central node are determined as the neighboring nodes of the central node. If k = 2, the first node directly connected to the central node and the second node directly connected to the first node are both determined as the neighboring nodes of the central node.

[0088] Furthermore, the number of adjacent points n can be set. Each time a behavior subgraph is collected, n adjacent points of the central node can be collected, thereby reducing the size of the behavior subgraph collected in a single time.

[0089] In practical applications, each time a behavior subgraph is collected, a behavior subgraph can be collected for each node, thereby obtaining each behavior subgraph of each node.

[0090] The electronic device can train each collected behavior subgraph to obtain a vector model corresponding to the preset topology graph. The obtained vector model can learn the relationship between nodes in each behavior subgraph, specifically the relationship between adjacent nodes and the central node.

[0091] Specifically, steps 502 and 503 may be iteratively performed to sample behavior subgraphs multiple times and perform training, thereby gradually optimizing the vector model.

[0092] Step 504: Acquire a preset topology map for representing the association relationship between users and items, and update data for updating the preset topology map.

[0093] The implementation of step 504 is similar to that of step 301 .

[0094] The obtained preset topology map may be the initial topology map constructed in step 501, or may be a topology map obtained by updating the initial topology map.

[0095] Step 505: Obtain a vocabulary table including vocabulary corresponding to each node in the preset topology graph.

[0096] Specifically, a vocabulary table can be pre-set. For example, after constructing an initial topology map, the electronic device can determine a vocabulary table based on each node in the initial topology map, thereby obtaining a vocabulary table that includes vocabulary corresponding to each node. For example, if the nodes include a node for item A, the electronic device can include information about item A as an entry in the vocabulary table.

[0097] Step 506: Filter out data in the updated data that does not belong to the vocabulary in the vocabulary according to the vocabulary.

[0098] Furthermore, the electronic device can perform preliminary filtering on the update data, specifically analyzing the subject information included in each update data, such as the operation behavior of subject A on subject B, and determining whether these subjects belong to the vocabulary in the vocabulary. If not, the corresponding update data is filtered out.

[0099] For example, if the updated data includes subject M, but the vocabulary does not include the vocabulary corresponding to M, the updated data will be filtered out.

[0100] Step 507: Update the edges in the preset topology graph according to the remaining update data to obtain an updated topology graph.

[0101] In actual application, the subjects included in the remaining update data after filtering all exist in the vocabulary table. Therefore, the subjects included in the remaining update data have corresponding nodes in the preset topology map. Therefore, the preset topology map can be updated according to the update data.

[0102] The electronic device may determine the relationship between the incremental nodes according to the remaining updated data, and update the edges in the preset topology graph according to the relationship between the incremental nodes to obtain an updated topology graph.

[0103] Specifically, an edge may be added to the preset topology graph based on the updated data, thereby obtaining an updated topology graph. For example, if the updated data includes detailed information about user B accessing item D, the electronic device may add an edge between user B's node and item D's node in the preset topology graph.

[0104] Step 508: Obtain a pre-trained vector model corresponding to a preset topology graph.

[0105] Step 508 is similar to the method of obtaining the vector model in step 302 and will not be described in detail.

[0106] Step 509 : Determine a behavior subgraph of any node based on the updated topology graph; wherein the behavior subgraph includes any node and its corresponding neighboring nodes, and the behavior subgraph is used to represent the relationship between any node and its neighboring nodes.

[0107] Furthermore, the edge relationship between nodes is added to the updated topology graph. Therefore, the behavior subgraph of any node can be determined based on the updated topology graph, thereby collecting the updated behavior subgraph of any node.

[0108] In practical applications, the behavior subgraph of each node can be updated, and the behavior subgraph of the node whose relationship with the neighboring node has changed can also be updated.

[0109] The updated behavior subgraph may include intermediate nodes and neighboring nodes of the intermediate nodes. The updated behavior subgraph of the same node may include richer information than the behavior subgraph before the update.

[0110] Specifically, an order can be pre-set, and the neighboring nodes of any node can be determined based on the preset order. Furthermore, a behavioral subgraph of any node can be determined based on the corresponding relationship between any node and its neighboring nodes. For example, an order of i can be set, and the electronic device can consider the i-th node connected to any node, starting from the node, and the nodes between the node and the i-th node as the neighboring nodes of the node.

[0111] Step 510: Determine an update vector of any node based on the behavior subgraph and vector model of any node.

[0112] Specifically, the vector model is obtained by collecting behavioral subgraphs based on the initial topology graph and then training these behavioral subgraphs. Therefore, adding edges to the initial topology graph based on updated data will not significantly affect the structure of the vector model itself. Based on this, the method provided in this application uses the previously trained vector model to process the behavioral subgraphs, thereby updating the update vector of the central node in each behavioral subgraph.

[0113] Furthermore, the electronic device can encode the behavior subgraph of any node based on the vector model to obtain the vector code of any node. The vector model includes multiple network structure parameters. The electronic device can use the attributes of any node and the attributes of its neighboring nodes as input information, and process the input information based on the network structure parameters included in the vector model to obtain the vector code of any node. Any node can be used as a central node, so that the neighboring nodes of the central node can be aggregated according to the vector model to obtain the vector code of the central node.

[0114] In practical applications, electronic devices encode the attributes of any node based on a vector model to obtain the attribute code for the node. The vector model may include an attribute coding network and parameters within the attribute coding network. The attribute coding network and its parameters can then be used to encode the attributes of any node to obtain the attribute code.

[0115] The electronic device can also use a vector model to encode the attribute code of any node based on the behavior subgraph of any node to obtain the vector code of any node. The vector model can include a topological coding network and parameters of the topological coding network. The topological coding network and its parameters can then be used in combination with the behavior subgraph of any node to encode the vector code of any node, thereby obtaining the vector code of any node.

[0116] In this way, a vector encoding including node attributes and the node topology information can be obtained.

[0117] In practical applications, the electronic device can process the vector code of any node based on the vector model to obtain the update vector of any node. The vector model can also include parameters of a mapping function, and the mapping function can process the vector code of any node to obtain the update vector of the node.

[0118] The electronic device can determine a mapping function based on a vector model. The vector model can include mapping function parameters, allowing the electronic device to restore the mapping function based on the vector model. The electronic device can determine an update vector for any node based on the mapping function and the vector code of any node. Specifically, the vector code can be used as input data for the mapping function, and the mapping function can then map the vector code to the update vector for the node.

[0119] Step 511: Determine the similarities between any user node and other item nodes based on the updated vectors of each node; and determine the similar item nodes of the user node based on the similarities between the user node and each item node.

[0120] Specifically, when information needs to be recommended to a user corresponding to a node, the electronic device may determine the similarity between the node and other nodes based on the updated vector of the node and the updated vectors of other nodes.

[0121] For example, if information needs to be recommended to user A, the electronic device may determine the similarity between the vector of user A's node and the vectors of other nodes, and further determine similar nodes to user A's node.

[0122] Furthermore, the electronic device may determine a cosine value between the updated vectors of two nodes and use it as the similarity between the two nodes.

[0123] Step 512: recommend item information of similar item nodes to the user corresponding to the user node.

[0124] The electronic device may recommend item information of an item to a user represented by any node based on similar nodes of the node.

[0125] For example, item information can be recommended to user A based on similar nodes to user A's node. For example, if a similar node to user A's node is an item node, the item information of that item node can be directly recommended to user A. For another example, if a similar node to user A's node is another user node, the electronic device can recommend item information to user A based on the information of the other user node, such as recommending item information of interest to the other user.

[0126] Figure 8 This is a structural diagram of an apparatus for recommending item information to a user, shown as an exemplary embodiment of the present application.

[0127] like Figure 8 As shown, the apparatus 800 for recommending item information to a user provided in this embodiment includes:

[0128] An acquisition unit 810 is configured to acquire a preset topological graph representing the association relationship between users and items, and update data for updating the preset topological graph; wherein the preset topological graph includes nodes, the nodes including user nodes and item nodes, and the user nodes and item nodes connected by edges have an association relationship;

[0129] A graph updating unit 820, configured to update edges in the preset topology graph according to the update data to obtain an updated topology graph;

[0130] A vector updating unit 830 is configured to obtain a pre-trained vector model corresponding to the preset topology map, and update a vector of any node in the topology map based on the updated topology map and the obtained vector model, wherein the vector is used to represent the characteristics of the node;

[0131] The recommendation unit 840 is configured to determine similar item nodes of the user node based on the vectors of the nodes, and recommend item information of the similar item nodes to the user corresponding to the user node.

[0132] The implementation method, principle, effect and other aspects of the device for personalized recommendation information provided by this application are as follows: Figure 3 The examples shown are similar and will not be described in detail.

[0133] Figure 9 This is a structural diagram of an apparatus for recommending item information to a user, shown as another exemplary embodiment of the present application.

[0134] like Figure 9 As shown, the apparatus 900 for recommending item information to a user provided by the present application, the graph updating unit 820, includes:

[0135] A vocabulary acquisition module 821 is used to acquire a vocabulary corresponding to each node in the preset topology graph;

[0136] A filtering module 822 is configured to filter out data included in the update data that does not belong to a vocabulary in the vocabulary according to the vocabulary;

[0137] The updating module 823 is configured to update the edges in the preset topology graph according to the remaining update data to obtain an updated topology graph.

[0138] In an optional implementation, the updating module 823 is specifically configured to:

[0139] The incremental inter-node relationship is determined according to the remaining updated data, and the edges in the preset topology graph are updated according to the incremental inter-node relationship to obtain an updated topology graph.

[0140] In an optional implementation, the vector updating unit 830 includes:

[0141] The subgraph determination module 831 is configured to determine a behavior subgraph of any node based on the updated topology graph; wherein the behavior subgraph includes any node and its corresponding neighboring nodes, and the behavior subgraph is used to represent the relationship between the node and its neighboring nodes;

[0142] The vector updating module 832 is configured to determine an update vector of any node according to the behavior subgraph of any node and the vector model.

[0143] In an optional implementation, the subgraph determination module 831 is specifically configured to determine neighboring nodes of any node based on a preset order, and determine a behavior subgraph of any node according to a corresponding relationship between any node and its neighboring nodes.

[0144] In an optional implementation, the vector updating module 832 is specifically configured to:

[0145] Encoding the behavior subgraph of any node based on the vector model to obtain a vector code of any node;

[0146] The vector code of any node is processed based on the vector model to obtain an updated vector of any node.

[0147] In an optional implementation, the vector updating module 832 is specifically configured to:

[0148] Encoding the attributes of any node based on the vector model to obtain the attribute code of any node;

[0149] By utilizing the vector model, the attribute code of any node is encoded according to the behavior subgraph of any node to obtain the vector code of any node.

[0150] In an optional implementation, the vector updating module 832 is specifically configured to:

[0151] determining a mapping function according to the vector model;

[0152] An update vector of any node is determined according to the mapping function and the vector code of any node.

[0153] In an optional implementation, the recommendation unit 840 includes:

[0154] A similarity determination module 841 is used to determine the similarities between any user node and other item nodes based on the updated vectors of each node;

[0155] The similar node determination module 842 is configured to determine similar item nodes of the user node based on the similarities between the user node and the item nodes.

[0156] In an optional embodiment, the device further includes a preset unit 850, configured to:

[0157] Acquire preset offline data, and construct an initial topology map based on the offline data;

[0158] Accordingly, the vector updating unit 830 is specifically configured to:

[0159] Acquire the initial topology map.

[0160] In an optional implementation, the preset unit 850 is further configured to:

[0161] Determine a behavior subgraph for each node in the initial topology graph;

[0162] Training is performed according to the behavior subgraph to obtain a vector model corresponding to the preset topology graph.

[0163] Figure 10 This is a structural diagram of an electronic device shown as an exemplary embodiment of the present application.

[0164] like Figure 10 As shown, the electronic device provided in this embodiment includes:

[0165] Memory 101;

[0166] Processor 102; and

[0167] computer programs;

[0168] The computer program is stored in the memory 101 and is configured to be executed by the processor 102 to implement any of the above-mentioned methods for recommending item information to a user.

[0169] This embodiment also provides a computer-readable storage medium having a computer program stored thereon.

[0170] The computer program is executed by a processor to implement any of the above-mentioned methods for recommending item information to a user.

[0171] This embodiment further provides a computer program, including program code. When a computer runs the computer program, the program code executes any one of the above-described methods for recommending item information to a user.

[0172] Those skilled in the art will appreciate that all or part of the steps in the above-described method embodiments can be implemented using hardware associated with program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0173] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some or all of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for recommending item information to a user, characterized in that: include: Obtaining a preset topological graph for representing associations between users and items, and update data for updating the preset topological graph; wherein the preset topological graph includes nodes, the nodes including user nodes and item nodes, and the user nodes and item nodes connected by edges have associations; the preset topological graph is a pre-constructed topological graph or a pre-updated topological graph; updating the edges in the preset topology graph according to the update data to obtain an updated topology graph; Obtaining a pre-trained vector model corresponding to the preset topology map, and updating a vector of any node in the topology map based on the updated topology map and the obtained vector model, wherein the vector is used to characterize the characteristics of the node; Based on the vectors of each of the nodes, similar item nodes of the user node are determined, and item information of the similar item nodes is recommended to the user corresponding to the user node.

2. The method according to claim 1, characterized in that The updating of the edges in the preset topology graph according to the update data to obtain an updated topology graph includes: Obtaining a vocabulary list of words corresponding to each node in the preset topology graph; According to the vocabulary, filtering out data included in the update data that does not belong to the vocabulary in the vocabulary; The edges in the preset topology graph are updated according to the remaining update data to obtain an updated topology graph.

3. The method according to claim 2, characterized in that Updating the edges in the preset topology graph according to the remaining update data to obtain an updated topology graph includes: The incremental inter-node relationship is determined according to the remaining updated data, and the edges in the preset topology graph are updated according to the incremental inter-node relationship to obtain an updated topology graph.

4. The method according to claim 1, wherein The updating of the vector of any node in the topology map according to the updated topology map and the obtained vector model includes: Determine a behavior subgraph of any node according to the updated topology graph; wherein the behavior subgraph includes any node and its corresponding neighboring nodes, and the behavior subgraph is used to represent the relationship between the any node and its neighboring nodes; An update vector of any node is determined according to the behavior subgraph of any node and the vector model.

5. The method according to claim 4, characterized in that Determining a behavior subgraph corresponding to any node according to the updated topology graph includes: The neighboring nodes of any node are determined based on a preset order, and the behavior subgraph of any node is determined according to the corresponding relationship between any node and its neighboring nodes.

6. The method according to claim 4, characterized in that The step of determining an update vector of any node according to the behavior subgraph of any node and the vector model includes: Encoding the behavior subgraph of any node based on the vector model to obtain a vector code of any node; The vector code of any node is processed based on the vector model to obtain an updated vector of any node.

7. The method according to claim 6, characterized in that The encoding process of the behavior subgraph of any node based on the vector model to obtain the vector encoding of any node includes: Encoding the attributes of any node based on the vector model to obtain the attribute code of any node; By utilizing the vector model, the attribute code of any node is encoded according to the behavior subgraph of any node to obtain the vector code of any node.

8. The method according to claim 6, characterized in that The processing of the vector code of any node based on the vector model to obtain an updated vector of any node includes: determining a mapping function according to the vector model; An update vector of any node is determined according to the mapping function and the vector code of any node.

9. The method according to claim 1, characterized in that Determining similar item nodes of the user node based on the updated vectors of each node includes: Based on the updated vectors of each node, determine the similarities between any user node and other item nodes; Determine similar item nodes to the user node based on the similarities between the user node and each item node.

10. The method according to any one of claims 1 to 9, characterized in that The preset topology map includes an initial topology map; The method further comprises: Acquire preset offline data, and construct an initial topology map based on the offline data; Correspondingly, the obtaining of the preset topology map includes: obtaining the initial topology map.

11. The method according to claim 10, characterized in that Also includes: Determine a behavior subgraph for each node in the initial topology graph; Training is performed according to the behavior subgraph to obtain a vector model corresponding to the preset topology graph.

12. A device for recommending item information to a user, characterized in that: include: an acquisition unit, configured to acquire a preset topological graph representing the association relationship between users and items, and update data for updating the preset topological graph; wherein the preset topological graph includes nodes, the nodes including user nodes and item nodes, and the user nodes and item nodes connected by edges have an association relationship; the preset topological graph is a pre-constructed topological graph or a pre-updated topological graph; A graph updating unit, configured to update edges in the preset topology graph according to the update data to obtain an updated topology graph; a vector updating unit, configured to obtain a pre-trained vector model corresponding to the preset topology map, and update a vector of any node in the topology map based on the updated topology map and the obtained vector model, wherein the vector is used to characterize the characteristics of the node; The recommendation unit is used to determine similar item nodes of the user node based on the vectors of each node, and recommend item information of the similar item nodes to the user corresponding to the user node.

13. An electronic device, characterized in that: include: Memory; processor; as well as computer programs; The computer program is stored in the memory and configured to be executed by the processor to implement the method according to any one of claims 1 to 11.

14. A computer-readable storage medium, characterized in that A computer program is stored thereon, The computer program is executed by a processor to implement the method according to any one of claims 1 to 11.

15. A computer program product comprising a computer program, which, when executed by a processor, implements the method according to any one of claims 1 to 11.

Images