Recommendation method, server, client, computer medium and device

Abstract

The present application discloses a recommendation method, a server, a client, a computer medium and a device, which can be applied to various scenarios such as artificial intelligence, machine learning, and deep learning. The method includes: obtaining target object information; obtaining a heterogeneous graph corresponding to the target object information; determining, according to the heterogeneous graph, a first network object information set corresponding to at least one first object node item among a plurality of node items, and a second network object information set corresponding to at least one second object node item among the plurality of node items; recommending a first target network object to a target object corresponding to the target object information by using the first network object information set and the second network object information set; the first object node item is an object node item corresponding to the active party object of the behavior in the heterogeneous graph, and the second object node item is an object node item corresponding to the passive party object of the behavior in the heterogeneous graph. From the perspective of the behavior type of the object, the consideration of individual differences is introduced, which improves the accuracy of network object recommendation and the efficiency of network object recommendation.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and particularly to a recommendation method, a server, a client, a computer medium and a device. Background Art

[0002] In related technologies, when recommending network objects to users, popular network objects are mostly determined based on the number of orders or purchase frequencies related to the network objects, and then recommended to users. The data types involved in the analysis by this method are relatively single, and the accuracy and efficiency of network object recommendation are relatively low. Summary of the Invention

[0003] Embodiments of this application provide a recommendation method, a server, a client, a computer medium and a device, which can recommend network objects based on the behavior levels corresponding to node items in a heterogeneous graph, and improve the efficiency of network object recommendation.

[0004] On the one hand, a recommendation method is provided. The method includes: obtaining target object information; obtaining a heterogeneous graph corresponding to the target object information, where the heterogeneous graph includes multiple node items and edges between the node items, the edges between the node items represent the association relationships between the node items, and the multiple node items include at least two types of node items; according to the heterogeneous graph, determining a first network object information set corresponding to at least one first object node item among the multiple node items, and a second network object information set corresponding to at least one second object node item among the multiple node items; using the first network object information set and the second network object information set to recommend a first target network object to a target object corresponding to the target object information; where the first object node item is an object node item corresponding to the object of the active party of the behavior in the heterogeneous graph, and the second object node item is an object node item corresponding to the object of the passive party of the behavior in the heterogeneous graph.

[0005] On the other hand, a recommendation method is provided, including: when a browsing instruction of a target object for a target page is detected, obtaining the target object information of the target object; sending the target object information to a first device; receiving and presenting the recommendation information fed back by the first device for the target object information; wherein, the recommendation information is that the first device obtains the target object information and a heterogeneous graph corresponding to the target object information, the heterogeneous graph includes a plurality of node items and edges between the node items, the edges between the node items represent the association relationships between the node items, the plurality of node items include at least two types of node items, and according to the heterogeneous graph, determining a first network object information set corresponding to at least one first object node item among the plurality of node items, and a second network object information set corresponding to at least one second object node item among the plurality of node items, and using the first network object information set and the second network object information set to determine a first target network object corresponding to the target object and then feeding back the recommendation information corresponding to the first target network object; wherein, the first object node item is an object node item corresponding to the active party object of the behavior in the heterogeneous graph, and the second object node item is an object node item corresponding to the passive party object of the behavior in the heterogeneous graph.

[0006] On the other hand, a server is provided, the server includes: a first obtaining unit, configured to obtain target object information; a second obtaining unit, configured to obtain a heterogeneous graph corresponding to the target object information, the heterogeneous graph includes a plurality of node items and edges between the node items, the edges between the node items represent the association relationships between the node items, the plurality of node items include at least two types of node items; a determining unit, configured to determine, according to the heterogeneous graph, a first network object information set corresponding to at least one first object node item among the plurality of node items, and a second network object information set corresponding to at least one second object node item among the plurality of node items; a recommending unit, configured to recommend a first target network object to the target object corresponding to the target object information by using the first network object information set and the second network object information set; wherein, the first object node item is an object node item corresponding to the active party object of the behavior in the heterogeneous graph, and the second object node item is an object node item corresponding to the passive party object of the behavior in the heterogeneous graph.

[0007] On the other hand, a client is provided, including: an acquisition unit, configured to acquire target object information of the target object when detecting a browsing instruction of the target object for a target page; a sending unit, configured to send the target object information to a first device; a receiving unit, configured to receive and display recommendation information fed back by the first device for the target object information; wherein, the recommendation information is that the first device acquires the target object information and a heterogeneous graph corresponding to the target object information, the heterogeneous graph includes a plurality of node items and edges between the node items, the edges between the node items represent the association relationships between the node items, the plurality of node items include at least two types of node items, and according to the heterogeneous graph, a first network object information set corresponding to at least one first object node item in the plurality of node items, and a second network object information set corresponding to at least one second object node item in the plurality of node items are determined, and the recommendation information corresponding to a first target network object determined by using the first network object information set and the second network object information set and fed back for the target object is obtained; wherein, the first object node item is an object node item corresponding to the active party object of the behavior in the heterogeneous graph, and the second object node item is an object node item corresponding to the passive party object of the behavior in the heterogeneous graph.

[0008] On the other hand, a computer-readable storage medium is provided, which stores a computer program, and the computer program is suitable for being loaded by a processor to execute the steps in the recommendation method described in any one of the above embodiments.

[0009] On the other hand, a computer device is provided, which includes a processor and a memory, the memory stores a computer program, and the processor is configured to execute the steps in the recommendation method described in any one of the above embodiments by calling the computer program stored in the memory.

[0010] On the other hand, a computer program product is provided, including computer instructions, and when the computer instructions are executed by a processor, the steps in the recommendation method described in any one of the above embodiments are implemented.

[0011] In an embodiment of the present application, target object information is obtained; a heterogeneous graph corresponding to the target object information is obtained, where the heterogeneous graph includes a plurality of node items and edges between the node items, the edges between the node items represent the association relationships between the node items, and the plurality of node items include at least two types of node items; according to the heterogeneous graph, a first network object information set corresponding to at least one first object node item among the plurality of node items and a second network object information set corresponding to at least one second object node item among the plurality of node items are determined; the first target network object is recommended to the target object corresponding to the target object information by using the first network object information set and the second network object information set; wherein, the first object node item is an object node item corresponding to the active party object of the behavior in the heterogeneous graph, and the second object node item is an object node item corresponding to the passive party object of the behavior in the heterogeneous graph. The solution recommends network objects by combining the behavior types of the objects corresponding to the node items in the heterogeneous graph, that is, the characteristics of being active or passive, and introduces the consideration of individual differences from the perspective of the behavior types of the objects, improving the accuracy and efficiency of network object recommendation. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative efforts.

[0013] Figure 1a It is a schematic structural diagram of a recommendation system provided by an embodiment of the present application.

[0014] Figure 1b It is a schematic diagram of a heterogeneous graph provided by an embodiment of the present application.

[0015] Figure 2a It is a first flowchart of a recommendation method provided by an embodiment of the present application.

[0016] Figure 2b It is a second flowchart of a recommendation method provided by an embodiment of the present application.

[0017] Figure 2c It is a third flowchart of a recommendation method provided by an embodiment of the present application.

[0018] Figure 3 It is a fourth flowchart of a recommendation method provided by an embodiment of the present application.

[0019] Figure 4 It is a schematic structural diagram of a server provided by an embodiment of the present application.

[0020] Figure 5 The structural schematic diagram of the client provided by the embodiment of the present application.

[0021] Figure 6 The structural schematic diagram of the computer device provided by the embodiment of the present application. Detailed implementation manners

[0022] Next, the technical solutions in the embodiments of the present application will be clearly and completely described with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative efforts shall fall within the protection scope of the present application.

[0023] The embodiments of the present application can be applied to various scenarios such as artificial intelligence, machine learning, and deep learning.

[0024] The embodiments of the present application provide a recommendation method, a server, a client, a computer medium, and a device. Specifically, the recommendation method of the embodiments of the present application can be executed by a computer device, where the computer device can be a terminal or a server and the like. The terminal can be a smart phone, a tablet computer, a notebook computer, a desktop computer, a smart TV, a smart speaker, a wearable smart device, a smart vehicle terminal, and other devices. The terminal can also include a client, and the client can be a video client, a browser client, or an instant messaging client, etc. The server can be an independent physical server, or a server cluster or a distributed system composed of multiple physical servers. It can also be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, Content Delivery Network (CDN), and big data and artificial intelligence platforms.

[0025] When the method runs on the terminal, the terminal needs to first download and install relevant application programs and store them. When the terminal actually runs the foregoing method, it is used to present relevant pages and network object recommendation content, and can interact with the user through a graphical user interface. The manner in which the terminal provides the graphical user interface to the user can include various methods. For example, it can be rendered and displayed on the display screen of the terminal, or the graphical user interface can be presented through holographic projection. Specifically, the terminal can include a touch display screen and a processor. The touch display screen is used to present the graphical user interface and receive the user's page browsing instruction. The graphical user interface includes a page for the user to browse. The processor is used to run relevant applications, generate the graphical user interface, respond to operation instructions, and control the display of the graphical user interface on the touch display screen.

[0026] When this method runs on a server, it can be cloud recommendation. Cloud recommendation refers to a recommendation method based on cloud computing. In the operation mode of cloud recommendation, the running entity of the recommendation application and the entity presenting the recommendation screen are separated. The storage and operation of the method are completed on the cloud recommendation server, while the presentation of the recommendation screen is completed on the cloud recommendation client. The cloud recommendation client is mainly used for receiving and sending recommendation data and presenting the recommendation screen. For example, the cloud recommendation client can be a display device with data transmission function near the user side, such as a mobile terminal, a TV set, a computer, a palm computer, a personal digital assistant, an intelligent vehicle terminal, etc. However, the device for processing recommendation data is the cloud recommendation server in the cloud. When making a recommendation, the user operates the cloud recommendation client to send an operation instruction to the cloud recommendation server. The cloud recommendation server runs the recommendation according to the operation instruction, encodes and compresses data such as the recommendation screen, and returns it to the cloud recommendation client through the network. Finally, the cloud recommendation client decodes it and outputs the screen corresponding to the recommendation page.

[0027] First, some nouns or terms that appear in the process of describing the embodiments of this application are explained as follows:

[0028] Artificial Intelligence (AI) is to use a digital computer or a machine controlled by a digital computer to simulate, extend and expand human intelligence, a theory, method, technology and application system that perceives the environment, acquires knowledge and uses knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology of computer science. It attempts to understand the essence of intelligence and produce a new intelligent machine that can react in a way similar to human intelligence. Artificial intelligence also studies the design principles and implementation methods of various intelligent machines to enable the machines to have the functions of perception, reasoning and decision-making.

[0029] The full name of the HITS algorithm is Hyperlink-Induced Topic Search. In the HITS algorithm, each page is given two attributes: the hub attribute and the authority attribute. At the same time, web pages are divided into two types: hub pages and authority pages. Hub means center. A hub page refers to a web page that contains many links pointing to authority pages, such as some portal websites; an authority page refers to a web page that contains substantial content. The purpose of the HITS algorithm is to return high-quality authority pages to users when they query.

[0030] According to the HITS algorithm, after the user enters keywords, the algorithm calculates two values for the returned matching pages. One is the hub score, and the other is the authority score. These two values are interdependent and influence each other. The hub score refers to the sum of the authority scores of the pages pointed to by all the outgoing links on a page, and the authority score refers to the sum of the hubs in the pages where all the incoming links are located.

[0031] Attention mechanism: Originating from the study of human vision, in cognitive science, due to the bottleneck of information processing, humans will selectively focus on a part of all information while ignoring other visible information. The above mechanism is usually called the attention mechanism. Different parts of the human retina have different degrees of information processing capabilities, that is, acuity. Only the fovea of the retina has the strongest acuity. In order to rationally utilize limited visual information processing resources, humans need to select a specific part of the visual area and then focus on it. For example, when people are reading, usually only a small number of words to be read will be focused on and processed. In summary, the attention mechanism mainly has two aspects: determining which part of the input needs to be focused on; allocating limited information processing resources to important parts.

[0032] Cold start problem: How to design a personalized recommendation system without a large amount of user data and make users satisfied with the recommendation results so that they are willing to use the recommendation system is the cold start problem. The cold start problem is mainly divided into three categories: 1) User cold start: User cold start mainly solves the problem of how to make personalized recommendations for new users; 2) System cold start: System cold start mainly solves the problem of how to design a personalized recommendation system on a newly developed website; 3) Item cold start: Item cold start mainly solves the problem of how to recommend new items to users who may be interested in them.

[0033] In the related art, in the scenario of network object recommendation, the popular network objects are mostly determined based on the order quantity or purchase frequency related to the network object, and then recommended to users. The types of data involved in the analysis by this method are relatively single, and the accuracy of network object recommendation is low, and the recommendation efficiency is low.

[0034] Moreover, the inventor found that when aggregating neighbor node information through a heterogeneous graph neural network, generally the vector similarity of node representations is used as the attention weight, without considering the influence of the object in the social network; in addition, the cold start solution relies more on the information of the network object itself and does not consider more dimensions; the efficiency of network object recommendation in the related art is low.

[0035] In addition, the inventors found through research that the essence of the Hits algorithm is to discover topics through links. The problem that the Hits algorithm aims to solve is to find high-quality authoritative pages and directory pages related to the user's query topic among a vast number of web pages, especially authoritative pages, because these pages represent high-quality content that meets the user's query, and the search engine returns them to the user as search results. Its two important assumptions are: a high-quality authoritative page will be pointed to by many high-quality directory pages. A high-quality directory page will point to many high-quality authoritative pages.

[0036] In the embodiments provided in the present application, based on the analysis of the Hits algorithm and combined with the characteristics of social relationships and the user's influence in their social circle, the inventors use it as the basis data for network object recommendation to improve the personalized recommendation effect, enhance the accuracy of network object recommendation, and improve the efficiency of network object recommendation.

[0037] Optionally, the above recommendation method can be applied but is not limited to a system such as Figure 1a shown in Figure 1a which is a schematic structural diagram of the recommendation system provided in the embodiments of the present application. The recommendation system includes multiple terminals 10 and a server 20, etc.; different terminals 10 are connected to the network through the server 20, and the terminals 10 and the server 20 are connected through the network. The network connection method can be wired or wireless, etc.

[0038] Among them, multiple terminals 10 can communicate with the server 20 based on the network. Any one of the multiple terminals 10 can be both a communication data sending device and a communication data receiving device.

[0039] The terminal 10 includes a memory for storing interaction data and a processor for processing interaction data. The server 20 includes a database and a processing engine. Among them, the database is used to store interaction data; the processing engine is used to process interaction data. The terminal 10 can be used to display a graphical user interface and interact with the user through the graphical user interface. Specifically, the terminal can display the graphical user interface by downloading and installing the corresponding client and running it, or by calling the corresponding applet and running it, or by presenting the corresponding graphical user interface based on the corresponding logged-in website.

[0040] In some alternative embodiments of the present application, when the terminal 10 detects a browsing instruction of a target object for a target page, it obtains the target object information of the target object; sends the target object information to the server 20; receives and displays the recommendation information corresponding to the first target network object feedback by the server 20 for the target object information. After receiving the target object information sent by the client 10, the server 20 obtains the heterogeneous graph corresponding to the target object information; the heterogeneous graph includes a plurality of node items and the edges between the node items, and the edges between the node items represent the association relationships between the node items; according to the heterogeneous graph, determine a first network object information set corresponding to at least one first object node item among the plurality of node items, and a second network object information set corresponding to at least one second object node item among the plurality of node items; use the first network object information set and the second network object information set to determine the first target network object corresponding to the target object, and feedback the recommendation information corresponding to the first target network object. Wherein, the first object node item is an object node item corresponding to the active party object in the behavior in the heterogeneous graph, and the second object node item is an object node item corresponding to the passive party object in the behavior in the heterogeneous graph. Specifically, the recommendation information corresponding to the first target network object for the target object information can be fed back to the client 10.

[0041] Optionally, the plurality of node items at least include two node item types: object node items and network object node items, and the node item types may further include network object category node items.

[0042] Optionally, for the foregoing heterogeneous graph, reference may be made to Figure 1b as shown, the edges between the node items can represent the association relationships between the node items on both sides of the edge. For example, the association relationships between object node items may include: following, private chat sharing, card liking, etc., and the association relationships between object node items and network object node items may include: group chat sharing, private chat sharing, card sharing, purchase, click, etc.

[0043] Optionally, the object in the present application may be a user, and the foregoing target object information may refer to object identifiers such as the object name and object account of the recommended object of the product to be recommended. Optionally, the foregoing network objects may include products, stores, network services, etc.

[0044] Optionally, the association relationship between the object node item and the network object node item may further include at least one of browsing, viewing details, following, adding to the shopping cart, and commenting. Among them, the network object category node item can solve the data sparsity problem in the heterogeneous graph.

[0045] Optionally, the above-mentioned server 20 may be a server cluster or a distributed system composed of physical servers, or may also be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The terminal 10 may be a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart speaker, a smart watch, a smart vehicle terminal, etc., but is not limited thereto.

[0046] The following will be described in detail respectively. It should be noted that the description order of the following embodiments does not limit the priority order of the embodiments.

[0047] Each embodiment of the present application provides a recommendation method. This method can be executed by a terminal or a server, or can be jointly executed by a terminal and a server. In the embodiments of the present application, the case where the recommendation method is executed by the server side is taken as an example for illustration. Figure 2a FIG. 8 is a first flowchart of the recommendation method provided by the embodiments of the present application. This method can be applied to a server, and the method includes:

[0048] S201. Obtain target object information;

[0049] S202. Obtain a heterogeneous graph corresponding to the target object information. The heterogeneous graph includes a plurality of node items and edges between the node items. The edges between the node items represent the association relationships between the node items. The plurality of node items include at least two types of node items;

[0050] S203. According to the heterogeneous graph, determine a first network object information set corresponding to at least one first object node item among the plurality of node items, and a second network object information set corresponding to at least one second object node item among the plurality of node items;

[0051] S204. Use the first network object information set and the second network object information set to recommend a first target network object to the target object corresponding to the target object information; wherein, the first object node item is an object node item corresponding to the active party object in the heterogeneous graph, and the second object node item is an object node item corresponding to the passive party object in the heterogeneous graph.

[0052] Optionally, the object in the present application may be a user, and the foregoing target object information may refer to object identifiers such as the object name and object account of the object to be recommended for the product to be recommended.

[0053] Optionally, the foregoing network object may include products, stores, network services, etc.

[0054] Optionally, the aforementioned heterogeneous graph is a heterogeneous graph constructed based on a set of object behavior information within a preset time period. The heterogeneous graph includes a plurality of node items and edges between the node items. The edges between the node items represent the association relationships between the node items. The plurality of node items include at least two types of node items; the two types of node items include object node items and network object node items, and the node item types may also include network object category node items.

[0055] Optionally, the aforementioned active behavior object is an object that makes an active behavior towards the object corresponding to other node items. For example, the active behavior object is an object that shares a network object with the object corresponding to other node items, and / or an object that initiates a chat with the object corresponding to other node items.

[0056] Optionally, the aforementioned passive behavior object is an object that is the target of an active behavior by the object corresponding to other node items. For example, the passive behavior object is an object that is shared a network object by the object corresponding to other node items, and / or an object that is initiated a chat with by the object corresponding to other node items.

[0057] Optionally, the aforementioned active behavior object also includes an object that is followed by the object corresponding to other node items. The aforementioned passive behavior object also includes an object that follows the object corresponding to other node items.

[0058] For example, in the heterogeneous graph, the object target1 corresponding to the node item n1 follows the object target12 corresponding to the node item n2. target1 is the passive behavior object, and target12 is the active behavior object.

[0059] Optionally, the aforementioned first network object information set is a set composed of the first network object information corresponding to each first object node item in at least one first object node item. The first object node item and the first network object information are in one-to-one correspondence.

[0060] Optionally, the aforementioned second network object information set is a set composed of the second network object information corresponding to each second object node item in at least one second object node item. The second object node item and the second network object information are in one-to-one correspondence.

[0061] Optionally, the first network object information corresponding to the first object node item includes: at least one first aggregated network object item of the first object node item. Each first aggregated network object item includes a first network object name and a first network object aggregation weight corresponding to the first network object name.

[0062] Optionally, the correspondence between the first object node item and its corresponding first network object information can be stored in a preset information library.

[0063] Optionally, the second network object information corresponding to the second object node item includes: at least one second aggregated network object item of the second object node item, and each second aggregated network object item includes a second network object name and a second network object aggregation weight corresponding to the second network object name.

[0064] Optionally, the correspondence between the second object node item and its corresponding second network object information can be stored in a preset information library.

[0065] In some alternative embodiments of the present application, in S203 above, determining the first network object information set corresponding to at least one first object node item among the multiple node items according to the heterogeneous graph includes:

[0066] S2031. For each first object node item among the at least one first object node item, determine the first network object association weight between the first object node item and the first neighboring network object node item of the first object node item according to the heterogeneous graph;

[0067] S2032. Determine the first object association weight between the first neighboring object node item of the first object node item and the first object node item, and the second network object association weight between the first neighboring object node item and the second neighboring network object node item of the first neighboring object node item according to the heterogeneous graph;

[0068] S2033. Based on the first neighboring network object node item of the first object node item, the first network object association weight, the second neighboring network object node item of the first neighboring object node item, the second network object association weight, and the first object association weight, determine the first network object information corresponding to the first object node item;

[0069] S2034. Use the first network object information corresponding to each first object node item to determine the first network object information set corresponding to the at least one first object node item.

[0070] Optionally, in S2031 above, determining the first network object association weight between the first object node item and the first neighboring network object node item of the first object node item according to the heterogeneous graph may include:

[0071] Determine the first network object association relationship between the first object node item and the first neighboring network object node item of the first object node item according to the edges between the node items in the heterogeneous graph;

[0072] Based on the first network object association relationship and a preset weight assignment rule, determine the first network object association weight between the first object node item and the first neighboring network object node item of the first object node item.

[0073] Optionally, in the foregoing S2032, determining the first object association weight between the first neighboring object node item of the first object node item and the first object node item, and the second network object association weight between the first neighboring object node item and the second neighboring network object node item of the first neighboring object node item according to the heterogeneous graph includes:

[0074] Determine the first object association relationship between the first object node item and the first neighboring object node item of the first object node item according to the edges between the node items in the heterogeneous graph;

[0075] Based on the first object association relationship and the preset weight assignment rule, determine the first object association weight between the first object node item and the first neighboring object node item of the first object node item;

[0076] Determine the second network object association relationship between the first neighboring object node item and the second neighboring network object node item of the first neighboring object node item according to the edges between the node items in the heterogeneous graph;

[0077] Based on the second network object association relationship and the preset weight assignment rule, determine the second network object association weight between the first neighboring object node item and the second neighboring network object node item of the first neighboring object node item.

[0078] Optionally, the number of the foregoing first neighboring object node items can be 1 or multiple, and can include the first-degree neighboring object node items of the first object node item, can also include the second-degree neighboring object node items of the first object node item, and can further include the higher-degree neighboring object node items of the first object node item.

[0079] Wherein, the degree in this solution refers to the neighboring distance. Specifically, the first degree means directly connected by one edge, the second degree means indirectly connected by two edges, and the nth degree means indirectly connected by n edges. For example, the first-degree neighboring object node item of the first object node item, that is, the object node item directly connected to the first object node item by one edge, the second-degree neighboring object node item of the first object node item refers to the object node item indirectly connected to the first object node item by two edges, and so on. The nth-degree neighboring object node item of the first object node item refers to the object node item indirectly connected to the first object node item by n edges.

[0080] Optionally, the number of the foregoing first neighboring network object node items can be 1 or multiple, and can include the first-degree neighboring network object node items of the first object node item, can also include the second-degree neighboring network object node items of the first object node item, and can further include the higher-degree neighboring network object node items of the first object node item.

[0081] Optionally, the number of the foregoing second-nearest neighbor network object node items may be 1 or more. It may include the first-degree nearest neighbor network object node items of the first-nearest neighbor object node item, may also include the second-degree nearest neighbor network object node items of the first-nearest neighbor object node item, and may further include the higher-degree nearest neighbor network object node items of the first-nearest neighbor object node item.

[0082] Optionally, one first-nearest neighbor object node item corresponds to one first object association weight.

[0083] Optionally, in the foregoing S2033, determining the first network object information corresponding to the first object node item based on the first-nearest neighbor network object node items of the first object node item, the first network object association weight, the second-nearest neighbor network object node items of the first-nearest neighbor object node item, the second network object association weight, and the first object association weight includes:

[0084] Determining the first initial network object information of the first object node item based on the first-nearest neighbor network object node items of the first object node item and the first network object association weight;

[0085] Determining the first-nearest neighbor network object information of the first-nearest neighbor object node item based on the second-nearest neighbor network object node items of the first-nearest neighbor object node item and the second network object association weight;

[0086] Using the first initial network object information, the first-nearest neighbor network object information, and the first object association weight to determine the first network object information corresponding to the first object node item.

[0087] In some alternative embodiments of the present application, in the foregoing S2033, determining the first network object information corresponding to the first object node item based on the first-nearest neighbor network object node items of the first object node item, the first network object association weight, the second-nearest neighbor network object node items of the first-nearest neighbor object node item, the second network object association weight, and the first object association weight may be implemented by the following formula:

[0088]

[0089] Wherein, i1 is the first object node item, j1 is the first neighboring object node item of the first object node item, attr1[i1] is the first initial network object information, specifically the combined information of the first neighboring network object node item of the first object node item and the association weight with the first network object, hub1[j1] is the first neighboring network object information of the first neighboring object node item of the first object node item, specifically the combined information of the second neighboring network object node item of the first neighboring object node item and the association weight with the second network object, W1[j1][i1] is the first object association weight, auth1[i1] is the first network object information corresponding to the first object node item, and n1 is the number of the first neighboring object node items of the first object node item.

[0090] Optionally, in the foregoing S2034, determining the first network object information set corresponding to the at least one first object node item by using the first network object information corresponding to each first object node item may include:

[0091] Regarding the set of the first network object information corresponding to each first object node item as the first network object information set corresponding to the at least one first object node item.

[0092] Optionally, the first aggregated network object node item is a neighboring network object node item, and the first network object aggregation weight is the association weight between the first object node item and its neighboring network object node item determined according to the association relationship between the first object node item and its neighboring network object node item.

[0093] In some alternative embodiments of the present application, one first network object information may be a teacup a1 and a weight wa1; another first network object information may be a red date a2 and a weight wa2; another first network object information is an umbrella a3 and a weight wa3; then the first network object information set is {teacup a1, weight wa1; red date a2, weight wa2; umbrella a3, weight wa3}.

[0094] In some alternative embodiments of the present application, in the foregoing S203, determining the second network object information set corresponding to at least one second object node item among the multiple node items according to the heterogeneous graph includes:

[0095] S231. For each second object node item among the at least one second object node item, determine the third network object association weight between the second object node item and the third neighboring network object node item of the second object node item according to the heterogeneous graph;

[0096] S232. Determine the second object association weight between the second nearest neighbor object node entry of the second object node entry and the second object node entry, and the fourth network object association weight between the second nearest neighbor object node entry and the fourth nearest neighbor network object node entry of the second nearest neighbor object node entry according to the heterogeneous graph;

[0097] S233. Determine the second network object information corresponding to the second object node entry based on the third nearest neighbor network object node entry of the second object node entry, the third network object association weight, the fourth nearest neighbor network object node entry of the second nearest neighbor object node entry, the fourth network object association weight, and the second object association weight;

[0098] S234. Use the second network object information corresponding to each second object node entry to determine the second network object information set corresponding to the at least one second object node entry.

[0099] Optionally, in the foregoing S231, determining the third network object association weight between the second object node entry and the third nearest neighbor network object node entry of the second object node entry according to the heterogeneous graph may include:

[0100] Determine the third network object association relationship between the second object node entry and the third nearest neighbor network object node entry of the second object node entry according to the edges between the node entries in the heterogeneous graph;

[0101] Based on the second network object association relationship and a preset weight assignment rule, determine the third network object association weight between the second object node entry and the third nearest neighbor network object node entry of the second object node entry.

[0102] Optionally, in the foregoing S232, determining the second object association weight between the second nearest neighbor object node entry of the second object node entry and the second object node entry, and the fourth network object association weight between the second nearest neighbor object node entry and the fourth nearest neighbor network object node entry of the second nearest neighbor object node entry includes:

[0103] Determine the second object association relationship between the second object node entry and the second nearest neighbor object node entry of the second object node entry according to the edges between the node entries in the heterogeneous graph;

[0104] Based on the second object association relationship and a preset weight assignment rule, determine the second object association weight between the second object node entry and the second nearest neighbor object node entry of the second object node entry;

[0105] Determine the fourth network object association relationship between the second nearest neighbor object node item and the fourth nearest neighbor network object node item of the second nearest neighbor object node item according to the edges between the node items in the heterogeneous graph;

[0106] Determine the fourth network object association weight between the second nearest neighbor object node item and the fourth nearest neighbor network object node item of the second nearest neighbor object node item based on the fourth network object association relationship and the preset weight assignment rule.

[0107] Optionally, the number of the foregoing second nearest neighbor object node items can be 1 or multiple, and can include the first-degree nearest neighbor object node items of the second object node item, the second-degree nearest neighbor object node items of the second object node item, and can also include the higher-degree nearest neighbor object node items of the second object node item.

[0108] Optionally, the number of the foregoing third nearest neighbor network object node items can be 1 or multiple, and can include the first-degree nearest neighbor network object node items of the second object node item, the second-degree nearest neighbor network object node items of the second object node item, and can also include the higher-degree nearest neighbor network object node items of the second object node item.

[0109] Optionally, the number of the foregoing fourth nearest neighbor network object node items can be 1 or multiple, and can include the first-degree nearest neighbor network object node items of the second nearest neighbor object node item, the second-degree nearest neighbor network object node items of the second nearest neighbor object node item, and can also include the higher-degree nearest neighbor network object node items of the second nearest neighbor object node item.

[0110] Optionally, one second nearest neighbor object node item corresponds to one second object association weight.

[0111] Optionally, in the foregoing 233, determining the second network object information corresponding to the second object node item based on the third nearest neighbor network object node item of the second object node item, the third network object association weight, the fourth nearest neighbor network object node item of the second nearest neighbor object node item, the fourth network object association weight, and the second object association weight includes:

[0112] Determine the second initial network object information of the second object node item based on the third nearest neighbor network object node item of the second object node item and the third network object association weight;

[0113] Determine the second nearest neighbor network object information of the second nearest neighbor object node item based on the fourth nearest neighbor network object node item of the second nearest neighbor object node item and the fourth network object association weight;

[0114] Determine the second network object information corresponding to the second object node entry by using the second initial network object information, the second neighboring network object information, and the second object association weight.

[0115] In some alternative embodiments of the present application, in the foregoing S233, based on the third neighboring network object node entry of the second object node entry, the third network object association weight, the fourth neighboring network object node entry of the second neighboring object node entry, the fourth network object association weight, and the second object association weight, determining the second network object information corresponding to the second object node entry can be implemented by the following formula:

[0116]

[0117] Wherein, i2 is the second object node entry, j2 is the second neighboring object node entry of the second object node entry, att2[i2] is the second initial network object information, specifically the combined information of the third neighboring network object node entry of the second object node entry and the third network object association weight, auth2[j] is the second neighboring network object information of the second neighboring object node entry of the second object node entry, specifically the combined information of the fourth neighboring network object node entry of the second neighboring object node entry and the fourth network object association weight, W2[j2][i2] is the second object association weight, hub2[i2] is the second network object information corresponding to the second object node entry, and n2 is the number of second neighboring object node entries of the second object node entry.

[0118] Optionally, in the foregoing S234, using the second network object information corresponding to each second object node entry to determine the second network object information set corresponding to the at least one second object node entry may include:

[0119] Taking the set of the second network object information corresponding to each second object node entry as the second network object information set corresponding to the at least one second object node entry.

[0120] Optionally, in the foregoing S204, recommending the first target network object to the target object corresponding to the target object information by using the first network object information set and the second network object information set includes:

[0121] S2041. Determine the target network object information set corresponding to the multiple node entries by using the first network object information set and the second network object information set;

[0122] S2042. Recommend the first target network object to the target object corresponding to the target object information based on the target network object information set.

[0123] Specifically, at least one of the multiple node items belongs to the first object node item and the second object node item at the same time; determining the target network object information set corresponding to the multiple node items by using the first network object information set and the second network object information set includes:

[0124] S41. Regarding the first network object information in the first network object information set, use the first network object information corresponding to the first target object node item that does not belong to the second object node item among the at least one first object node item as the network object information corresponding to the node item corresponding to the first target object node item;

[0125] S42. Regarding the second network object information in the second network object information set, use the second network object information corresponding to the second target object node item that does not belong to the first object node item among the at least one second object node item as the network object information corresponding to the node item corresponding to the second target object node item;

[0126] S43. For the duplicate node items that belong to the first object node item and the second object node item at the same time among the multiple node items, based on the first network object information corresponding to the duplicate node item in the first network object information set, the first preset weight, the second network object information corresponding to the duplicate node item in the second network object information set, and the second preset weight, determine the network object information corresponding to the node item corresponding to the duplicate node item;

[0127] S44. Use the network object information corresponding to the node item corresponding to the first target object node item, the network object information corresponding to the node item corresponding to the second target object node item, and the network object information corresponding to the node item corresponding to the duplicate node item to determine the target network object information set corresponding to the multiple node items.

[0128] Optionally, in the foregoing S43, determining the network object information corresponding to the node item corresponding to the duplicate node item based on the first network object information corresponding to the duplicate node item in the first network object information set, the first preset weight, the second network object information corresponding to the duplicate node item in the second network object information set, and the second preset weight includes:

[0129] Calculate the product of the first network object information and the first preset weight;

[0130] Calculate the product of the second network object information and the second preset weight;

[0131] Use the sum of the product of the first network object information and the first preset weight and the product of the second network object information and the second preset weight as the network object information corresponding to the node item corresponding to the duplicate node item.

[0132] Optionally, based on the first network object information corresponding to the repeated node item in the first network object information set, the first preset weight, the second network object information corresponding to the repeated node item in the second network object information set, and the second preset weight, determining the network object information corresponding to the node item corresponding to the repeated node item can be implemented through the following formula:

[0133] Result[i] = α * auth3[i3] + β * hub3[i]

[0134] Where α is the first preset weight, β is the second preset weight, the sum of α and β is 1, auth3[i3] is the first network object information corresponding to the repeated node item, hub3[i3] is the second network object information corresponding to the repeated node item, and Result[i] is the network object information corresponding to the node item corresponding to the repeated node item.

[0135] In some alternative embodiments of the present application, recommending a first target network object to the target object based on the target network object information set includes:

[0136] Querying the candidate node items that match the target object information from the multiple node items;

[0137] Determining the target network object information corresponding to the candidate node item based on the candidate node item and the target network object information set;

[0138] Recommending a first target network object to the target object according to the target network object information.

[0139] Among them, the candidate node item that matches the target object information is the node item among the multiple node items whose corresponding object information is the same as the target object information.

[0140] Determining the target network object information corresponding to the candidate node item based on the candidate node item and the target network object information set includes:

[0141] Querying the network object information in the target network object information set whose corresponding node item is the same as the candidate node item; specifically, the corresponding relationship between each network object information in the target network object information set and its corresponding node item can be included in the heterogeneous graph or stored in a preset database.

[0142] Taking the network object information whose corresponding node item is the same as the candidate node item as the target network object information.

[0143] Recommending a first target network object to the target object according to the target network object information includes:

[0144] Recommend the network object corresponding to the network object name with a network object weight greater than a preset value in the target network object information to the target object.

[0145] Optionally, in some alternative embodiments of the present application, in order to further improve the user experience, the above method further includes:

[0146] If it is detected that the operation information of the target object for the first target network object meets a preset condition, then determine the target network object category node item connected to the network object node item corresponding to the first target network object according to the heterogeneous graph;

[0147] Obtain the target neighbor network object node item of the target network object category node item;

[0148] Recommend a second target network object to the target object by using the target neighbor network object node item.

[0149] Optionally, when the operation information is any one or more of the following, it is regarded that the operation information meets the preset condition:

[0150] The operation information is a click operation, the operation information is a browsing operation, the operation information is a like operation, and the operation information is a favorite operation.

[0151] Optionally, the target neighbor network object node item of the target network object category node item is one or more network object node items other than the first target network object node item among the multiple neighbor network object node items of the target network object category node item.

[0152] Optionally, recommending a second target network object to the target object corresponding to the target object information by using the target neighbor network object node item includes:

[0153] Obtain the network object corresponding to the target neighbor network object node item;

[0154] Recommend the network object corresponding to the target neighbor network object node item to the target object as the second target network object.

[0155] Wherein, "multiple" in the present application means greater than or equal to 2.

[0156] Optionally, in order to further improve the accuracy of network object recommendation, the method further includes:

[0157] S51. Obtain the attribute information corresponding to each node item among the multiple node items;

[0158] S52. For each of the node items, determine the attribute association weight between the node item and its neighboring node item according to the attribute information corresponding to the node item and the attribute information of the neighboring node item of the node item.

[0159] S53. Recommending the first target network object to the target object corresponding to the target object information based on the target network object information set includes: recommending the first target network object to the target object corresponding to the target object information based on the target network object information set, the attribute information corresponding to each node item in the multiple node items, and the attribute association weight between each node item and its neighboring node item.

[0160] Optionally, when the node item is an object node item, the attribute information corresponding to the node item includes: the individual information of the object, or the individual information and preference information of the object. When the node item is a network object node item, the attribute information corresponding to the node item includes: network object characteristics information such as network object price and network object type.

[0161] It should be noted that the attribute information corresponding to the node item involved in this application is information obtained after obtaining the authorization of the corresponding object.

[0162] Optionally, the attribute information of the object can be obtained from the original database corresponding to the heterogeneous graph, or can be included in the heterogeneous graph and obtained from the heterogeneous graph.

[0163] In the foregoing S52, for each of the node items, determining the attribute association weight between the node item and its neighboring node item according to the attribute information corresponding to the node item and the attribute information of the neighboring node item of the node item can be implemented based on the attention mechanism.

[0164] Specifically, recommending the first target network object to the target object corresponding to the target object information based on the target network object information set, the attribute information corresponding to each node item in the multiple node items, and the attribute association weight between each node item and its neighboring node item includes:

[0165] S61. According to the heterogeneous graph, determine the first node association weight between each node item and its neighboring node item in the multiple node items.

[0166] S62. Use the target network object information set, the first node association weight between each node item and its neighboring node item, the attribute information corresponding to each node item in the multiple node items, and the attribute association weight between each node item and its neighboring node item to determine the embedding representation corresponding to each node item in the multiple node items, and obtain the embedding representation set corresponding to the multiple node items.

[0167] S63. Recommend a first target network object to the target object corresponding to the target object information based on the embedding representation set.

[0168] Among them, in the aforementioned S62, determining the embedding representation corresponding to each node item among the multiple node items by using the target network object information set, the first node association weight between each node item and its neighboring node item, the attribute information corresponding to each node item among the multiple node items, and the attribute association weight between each node item and its neighboring node item includes:

[0169] For each node item, perform the following steps:

[0170] S621. Determine the network object information corresponding to the node item according to the target network object information set;

[0171] S622. Use the network object information, the attribute information corresponding to the node item, and the attribute information of the neighboring node item of the node item to determine a first result;

[0172] S623. Determine a second result according to the attribute association weight between the node item and its neighboring node item, the attribute information corresponding to the node item, and the attribute information of the neighboring node item of the node item;

[0173] S624. Determine a third result according to the first node association weight between the node item and its neighboring node item, the attribute information corresponding to the node item, and the attribute information of the neighboring node item of the node item;

[0174] S625. Determine the embedding representation of the node item based on the first result, the second result, and the third result.

[0175] Among them, the embedding representation of the node item can be the concatenation result of the first result, the second result, and the third result, or other information related to the first result, the second result, and the third result. Among them, the embedding representation can specifically be a vector representation.

[0176] Optionally, the method for determining the embedding representation of a node item among the multiple node items can be implemented through a preset heterogeneous graph neural network.

[0177] Optionally, the input information of the preset heterogeneous graph neural network can include:

[0178] The second node association weight between the node item and the neighboring node item of the node item determined according to the network object information corresponding to the node item;

[0179] The attribute association weight between the node item and its neighboring node item;

[0180] The first node association weight between the node item and its neighboring node item;

[0181] The attribute information corresponding to the node item and the attribute information of its neighboring node items.

[0182] When processing the foregoing input information through a preset heterogeneous graph neural network based on the node-level attention mechanism, the heterogeneous graph neural network can determine the embedding representation corresponding to the node item through the following formula:

[0183]

[0184] where δ is a non-linear activation function related to the heterogeneous graph neural network, W p is the parameter of the fully connected layer in the heterogeneous graph neural network, u represents the current node item, k represents the neighboring node item of the current node item, X k is the vector representation related to the attribute information corresponding to the neighboring node item, is the second node association weight between the node item and its corresponding neighboring node item, refers to all neighboring node items of the node item with node item type t, b p is the bias term in the heterogeneous graph neural network. is the attribute association weight between the node item and its corresponding neighboring node item, is the first node association weight between the node item and its neighboring node item. denotes concatenation, is the embedding representation of the node item. Among them, in the calculation process, the node item itself can also be regarded as a neighboring node item of the node item.

[0185] In some other alternative embodiments of the present application, the foregoing input information can also be processed through a preset heterogeneous graph neural network based on the type-level attention mechanism. When processing the foregoing input information through a preset heterogeneous graph neural network based on the type-level attention mechanism, the heterogeneous graph neural network can determine the embedding representation corresponding to the node item through the following formula:

[0186]

[0187] where, is the embedding representation corresponding to the node item with node item type t, β tu is the weight information between the current node item and the node item of type t, T is the type of all node items, W h is the parameter of the fully connected layer in the heterogeneous graph neural network based on the type-level attention mechanism, h u is the embedding representation of the node item with node item type t obtained by the heterogeneous graph neural network based on the type-level attention mechanism.

[0188] Accordingly, the heterogeneous graph neural network based on the type-level attention mechanism can also determine the embedding representations of node items of other node item types, and the specific processing process is similar to the way of determining the embedding representations of node items of node item type t.

[0189] In some alternative embodiments of the present application, as shown in Figure 2b the solution of the present application can determine the foregoing first node association weight and second node association weight through a heterogeneous graph based on a weight calculation module, and calculate the attribute association weight between a node item and its neighboring node items based on the attribute information of each node item; and through a heterogeneous graph neural network, process the input information of the foregoing heterogeneous graph neural network, and execute the heterogeneous graph neural network to obtain the embedding representation corresponding to each node item.

[0190] Optionally, the foregoing heterogeneous graph neural network model is a machine learning model, and the embedding vectors corresponding to each node item form a set of embedding representations corresponding to multiple node items;

[0191] Optionally, the foregoing recommendation of the first target network object to the target object corresponding to the target object information based on the set of embedding representations includes:

[0192] Determine the target embedding representation corresponding to the target object information from the set of embedding representations;

[0193] Use, as the first target network object, the network object corresponding to the embedding representation in the set of embedding representations corresponding to the multiple node items, which has a similarity greater than a preset similarity with the target embedding representation, except for the target embedding representation, and recommend it to the target object.

[0194] Optionally, the solution of the present application can also be applied to the recommendation of network objects in social small circles. In addition, since new users have no or few information and their interest preferences cannot be known, by recommending representative or well-spread network objects in the small circle to this part of customers, the cold start problem can be effectively alleviated.

[0195] Optionally, in the heterogeneous graph, find the network object that has been interacted with by multiple groups of connected users, indicating that the network object appears frequently among users and is easy to spread in the social circle. The network object that has been interacted with by multiple groups of connected users can be used as the first target network object.

[0196] Optionally, the higher the user intimacy, the more important the network objects that have been purchased or shared with each other are, indicating that the network objects have a good spread in the small circle. Therefore, the embodiments of the present application can also calculate the average intimacy of the neighboring object node items of each network object node item, and select the first preset number of network objects with the highest object association weight value between the neighboring object node items as the first target network objects.

[0197] In some alternative embodiments of the present application, the average weight of each neighboring network object node item of the first object node item in the heterogeneous graph may also be obtained, and the second preset number of network objects with the highest average weight among the foregoing first preset number of network objects are used as the first target network objects. Finally, the first target network objects may be recommended to new users as popular, highly disseminated, and high-quality network objects in the small circle.

[0198] In the embodiments of the present application, by obtaining target object information; obtaining a heterogeneous graph corresponding to the target object information, the heterogeneous graph includes a plurality of node items and edges between the node items, and the edges between the node items represent the association relationship between the node items, and the plurality of node items include at least two types of node items; according to the heterogeneous graph, determining a first network object information set corresponding to at least one first object node item among the plurality of node items, and a second network object information set corresponding to at least one second object node item among the plurality of node items; using the first network object information set and the second network object information set to recommend a first target network object to the target object corresponding to the target object information; wherein, the first object node item is an object node item corresponding to the object of the active party in the behavior in the heterogeneous graph, and the second object node item is an object node item corresponding to the object of the passive party in the behavior in the heterogeneous graph. The solution combines the behavior types of the objects corresponding to the node items in the heterogeneous graph, that is, the characteristics of being active or passive, recommends network objects, takes into account individual differences from the perspective of the behavior types of the objects, and improves the accuracy and efficiency of network object recommendation. It can provide personalized recommendations for users, effectively improve online active users, and transaction conversion rates and other indicators.

[0199] Moreover, for the cold start problem, the embodiments of the present application use network objects that are widely disseminated and have good conversion in social small circles as the cold start network object recall pool, and utilize the unique social elements in social e-commerce to provide an innovative solution to the difficult cold start problem in e-commerce recommendation, and can ensure the high quality, diversity, accuracy, interpretability, etc. of network object recommendation, thereby improving the satisfaction and acceptance of users with the recommendation results.

[0200] The solution of the present application will be further described below in combination with scenarios:

[0201] See Figure 2c As shown, the object behavior information set may be obtained first. The object behavior information set includes: attribute information of multiple objects, attribute information of multiple network objects (which may be goods), and association relationship information between the object and the network object, and association relationship information between the object and the object;

[0202] Extract the association relationship information between the object and the network object, and the association relationship information between objects through a relationship extractor, and construct a corresponding heterogeneous graph through a heterogeneous graph generator;

[0203] Extract the attribute information of multiple objects and the attribute information of multiple network objects through a feature extractor;

[0204] Calculate the first node association weight between a node item and its neighboring node items according to the heterogeneous graph;

[0205] Calculate the second node association weight between a node item and the neighboring node items of the node item according to the type of the object;

[0206] Calculate the attribute association weight between a node item and its neighboring node items according to the attribute information of the node item;

[0207] Input the aforementioned second node association weight, attribute association weight, first node association weight, the attribute information corresponding to the node item, and the attribute information of its neighboring node items into the heterogeneous graph neural network model, and execute the heterogeneous graph neural network model to further realize network object recommendation.

[0208] Figure 3 This is the second process schematic diagram of the recommendation method provided by the embodiment of the present application. This embodiment is described by taking the recommendation method as being executed by the client as an example. The method includes:

[0209] S301. When detecting a browsing instruction of a target object for a target page, obtain the target object information of the target object;

[0210] S302. Send the target object information to the first device;

[0211] S303. Receive and display the recommendation information fed back by the first device for the target object information;

[0212] Among them, the recommended information is that the first device obtains the target object information and the heterogeneous graph corresponding to the target object information. The heterogeneous graph includes a plurality of node items and edges between the node items. The edges between the node items represent the association relationships between the node items. The plurality of node items include at least two types of node items, and according to the heterogeneous graph, determine a first network object information set corresponding to at least one first object node item among the plurality of node items, and a second network object information set corresponding to at least one second object node item among the plurality of node items, and use the first network object information set and the second network object information set to determine the recommended information corresponding to the first target network object after determining the first target network object corresponding to the target object; among them, the first object node item is the object node item corresponding to the active party object of the behavior in the heterogeneous graph, and the second object node item is the object node item corresponding to the passive party object of the behavior in the heterogeneous graph.

[0213] Optionally, the foregoing first device may be a server. For the specific implementation manners corresponding to this embodiment, reference may be made to the foregoing content, and details are not described herein again.

[0214] All the above technical solutions can be combined arbitrarily to form alternative embodiments of the present application, which will not be elaborated herein one by one.

[0215] To facilitate better implementation of the recommendation method in the embodiments of the present application, the embodiments of the present application further provide a server. Please refer to Figure 4 , Figure 4 which is the first structural schematic diagram of the server provided in the embodiments of the present application. Among them, the server 40 includes:

[0216] A first acquisition unit 41, configured to acquire target object information;

[0217] A second acquisition unit 42, configured to acquire the heterogeneous graph corresponding to the target object information. The heterogeneous graph includes a plurality of node items and edges between the node items. The edges between the node items represent the association relationships between the node items. The plurality of node items include at least two types of node items;

[0218] A determination unit 43, configured to determine, according to the heterogeneous graph, a first network object information set corresponding to at least one first object node item among the plurality of node items, and a second network object information set corresponding to at least one second object node item among the plurality of node items;

[0219] A recommendation unit 44 for recommending a first target network object to a target object corresponding to the target object information by using the first network object information set and the second network object information set; wherein, the first object node item is an object node item corresponding to the object of the active party of the behavior in the heterogeneous graph, and the second object node item is an object node item corresponding to the object of the passive party of the behavior in the heterogeneous graph.

[0220] Optionally, when the server 40 is used to determine the first network object information set corresponding to at least one first object node item among the multiple node items according to the heterogeneous graph, it is specifically used for: for each first object node item among the at least one first object node item, determining a first network object association weight between the first object node item and the first neighboring network object node item of the first object node item according to the heterogeneous graph; determining a first object association weight between the first neighboring object node item of the first object node item and the first object node item, and a second network object association weight between the first neighboring object node item and the second neighboring network object node item of the first neighboring object node item according to the heterogeneous graph; determining the first network object information corresponding to the first object node item based on the first neighboring network object node item of the first object node item, the first network object association weight, the second neighboring network object node item of the first neighboring object node item, the second network object association weight, and the first object association weight; and using the first network object information corresponding to each first object node item to determine the first network object information set corresponding to the at least one first object node item.

[0221] Optionally, when the server 40 is used to determine the first network object information corresponding to the first object node item based on the first neighboring network object node item of the first object node item, the first network object association weight, the second neighboring network object node item of the first neighboring object node item, the second network object association weight, and the first object association weight, it is specifically used for: determining the first initial network object information of the first object node item based on the first neighboring network object node item of the first object node item and the first network object association weight; determining the first neighboring network object information of the first neighboring object node item based on the second neighboring network object node item of the first neighboring object node item and the second network object association weight; and determining the first network object information corresponding to the first object node item by using the first initial network object information, the first neighboring network object information, and the first object association weight.

[0222] Optionally, when the server 40 is used to recommend a first target network object to the target object corresponding to the target object information by using the first network object information set and the second network object information set, it is specifically configured to: determine a target network object information set corresponding to the multiple node items by using the first network object information set and the second network object information set; and recommend a first target network object to the target object corresponding to the target object information based on the target network object information set.

[0223] Optionally, at least one of the multiple node items belongs to a first object node item and a second object node item at the same time; when the server 40 is used to determine a target network object information set corresponding to the multiple node items by using the first network object information set and the second network object information set, it is specifically configured to: use, as the network object information corresponding to the node item corresponding to the first target object node item, the first network object information corresponding to the first target object node item that does not belong to the second object node item in the at least one first object node item in the first network object information set; use, as the network object information corresponding to the node item corresponding to the second target object node item, the second network object information corresponding to the second target object node item that does not belong to the first object node item in the at least one second object node item in the second network object information set; for the duplicate node items that belong to the first object node item and the second object node item at the same time among the multiple node items, determine the network object information corresponding to the node item corresponding to the duplicate node items based on the first network object information corresponding to the duplicate node items in the first network object information set, the first preset weight, the second network object information corresponding to the duplicate node items in the second network object information set, and the second preset weight; and determine the target network object information set corresponding to the multiple node items by using the network object information corresponding to the node item corresponding to the first target object node item, the network object information corresponding to the node item corresponding to the second target object node item, and the network object information corresponding to the node item corresponding to the duplicate node items.

[0224] Optionally, when the server 40 is used to recommend a first target network object to the target object corresponding to the target object information based on the target network object information set, it is specifically configured to: query candidate node items that match the target object information from the multiple node items;

[0225] determine the target network object information corresponding to the candidate node items based on the candidate node items and the target network object information set; and recommend a first target network object to the target object according to the target network object information.

[0226] Optionally, the server 40 is further configured to: obtain the attribute information corresponding to each node item among the multiple node items; for each node item, determine the attribute association weight between the node item and its neighboring node item according to the attribute information corresponding to the node item and the attribute information of its neighboring node item;

[0227] Recommending a first target network object to a target object corresponding to the target object information based on the target network object information set includes: recommending a first target network object to a target object corresponding to the target object information based on the target network object information set, the attribute information corresponding to each node item among the multiple node items, and the attribute association weight between each node item and its neighboring node item.

[0228] Optionally, when the server 40 is configured to recommend a first target network object to a target object corresponding to the target object information based on the target network object information set, the attribute information corresponding to each node item among the multiple node items, and the attribute association weight between each node item and its neighboring node item, it is specifically configured to: determine the first node association weight between each node item and its neighboring node item among the multiple node items according to the heterogeneous graph; use the target network object information set, the first node association weight between each node item and its neighboring node item, the attribute information corresponding to each node item among the multiple node items, and the attribute association weight between each node item and its neighboring node item to determine the embedding representation corresponding to each node item among the multiple node items, obtaining the embedding representation set corresponding to the multiple node items; recommend a first target network object to a target object corresponding to the target object information based on the embedding representation set.

[0229] Optionally, the server 40 is further configured to: if it is detected that the operation information of the target object for the first target network object meets a preset condition, then determine the target network object category node item connected to the network object node item corresponding to the first target network object according to the heterogeneous graph; obtain the target neighboring network object node item of the target network object category node item; use the target neighboring network object node item to recommend a second target network object to the target object.

[0230] For the specific implementation manners corresponding to this embodiment, reference may be made to the foregoing content, which will not be elaborated herein.

[0231] To facilitate better implementation of the recommendation method in the embodiments of the present application, an embodiment of the present application further provides a client. Please refer to Figure 5 , Figure 5 which is the first structural schematic diagram of the client provided in the embodiments of the present application. Among them, the client 50 includes:

[0232] An obtaining unit 51, configured to obtain target object information of the target object when detecting a browsing instruction of the target object for a target page;

[0233] A sending unit 52, configured to send the target object information to a first device;

[0234] A receiving unit 53, configured to receive and display recommendation information fed back by the first device for the target object information;

[0235] Wherein, the recommendation information is that the first device obtains the target object information and a heterogeneous graph corresponding to the target object information. The heterogeneous graph includes a plurality of node items and edges between the node items. The edges between the node items represent the association relationships between the node items. The plurality of node items include at least two types of node items. According to the heterogeneous graph, a first network object information set corresponding to at least one first object node item among the plurality of node items, and a second network object information set corresponding to at least one second object node item among the plurality of node items are determined, and the recommendation information corresponding to the first target network object fed back after determining the first target network object corresponding to the target object by using the first network object information set and the second network object information set is used; wherein, the first object node item is an object node item corresponding to the active party object of the behavior in the heterogeneous graph, and the second object node item is an object node item corresponding to the passive party object of the behavior in the heterogeneous graph.

[0236] For the specific implementation manners corresponding to this embodiment, reference may be made to the foregoing content, which will not be elaborated herein.

[0237] Each of the above units may be implemented in whole or in part by software, hardware, and their combination. Each of the above units may be embedded in a processor in a computer device in a hardware form or be independent of the processor, or may be stored in a memory in the computer device in a software form, so as to facilitate the processor to call and execute the operations corresponding to each of the above units.

[0238] Optionally, the present application further provides a computer device, including a memory and a processor. A computer program is stored in the memory. When the processor executes the computer program, the steps in the above method embodiments are implemented.

[0239] Figure 6 It is a schematic structural diagram of the computer device provided by the embodiment of the present application. The computer device may be the terminal or server shown in FIG. 1. As Figure 6As shown, the computer device 600 may include: a communication interface 601, a memory 602, a processor 603, and a communication bus 604. The communication interface 601, the memory 602, and the processor 603 communicate with each other through the communication bus 604. The communication interface 601 is used for the device 700 to communicate with external devices. The memory 602 can be used to store software programs and modules. The processor 603 runs the software programs and modules stored in the memory 602, such as the software programs for the corresponding operations in the foregoing method embodiments.

[0240] Optionally, the processor 603 may also call the software programs and modules stored in the memory 602 to perform the following operations: obtain target object information; obtain a heterogeneous graph corresponding to the target object information, where the heterogeneous graph includes multiple node items and edges between the node items, the edges between the node items represent the association relationships between the node items, and the multiple node items include at least two types of node items; according to the heterogeneous graph, determine a first network object information set corresponding to at least one first object node item among the multiple node items, and a second network object information set corresponding to at least one second object node item among the multiple node items; use the first network object information set and the second network object information set to recommend a first target network object to the target object corresponding to the target object information; where the first object node item is an object node item corresponding to the active party object in the heterogeneous graph, and the second object node item is an object node item corresponding to the passive party object in the heterogeneous graph.

[0241] Optionally, the processor 603 may also call the software programs and modules stored in the memory 602 to perform the following operations: when detecting a browsing instruction of the target object for a target page, obtain the target object information of the target object; send the target object information to a first device; receive and display the recommendation information fed back by the first device for the target object information; where the recommendation information is that the first device obtains the target object information and a heterogeneous graph corresponding to the target object information, the heterogeneous graph includes multiple node items and edges between the node items, the edges between the node items represent the association relationships between the node items, the multiple node items include at least two types of node items, and according to the heterogeneous graph, determine a first network object information set corresponding to at least one first object node item among the multiple node items, and a second network object information set corresponding to at least one second object node item among the multiple node items, and use the first network object information set and the second network object information set to determine the recommendation information corresponding to the first target network object corresponding to the target object after determining the first target network object corresponding to the target object; where the first object node item is an object node item corresponding to the active party object in the heterogeneous graph, and the second object node item is an object node item corresponding to the passive party object in the heterogeneous graph.

[0242] The present application also provides a computer-readable storage medium for storing a computer program. The computer-readable storage medium can be applied to a computer device, and the computer program causes the computer device to execute the corresponding processes in the recommendation method in the embodiments of the present application. For the sake of brevity, details are not described herein again.

[0243] The present application also provides a computer program product, which includes computer instructions stored in a computer-readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, causing the computer device to execute the corresponding processes in the recommendation method in the embodiments of the present application. For the sake of brevity, details are not described herein again.

[0244] The present application also provides a computer program, which includes computer instructions stored in a computer-readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, causing the computer device to execute the corresponding processes in the recommendation method in the embodiments of the present application. For the sake of brevity, details are not described herein again.

[0245] It should be understood that the processor in the embodiments of the present application may be an integrated circuit chip with signal processing capabilities. In the implementation process, the steps of the above method embodiments can be completed by the integrated logic circuit in the hardware of the processor or the instructions in software form. The above processor may be a general-purpose processor, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components. It can implement or execute the various methods, steps and logic block diagrams disclosed in the embodiments of the present application. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The steps of the method disclosed in combination with the embodiments of the present application can be directly embodied as being executed by the hardware decoding processor, or executed by the combination of the hardware and software modules in the decoding processor. The software module may be located in a mature storage medium in the art such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory, and the processor reads the information in the memory and combines its hardware to complete the steps of the above method.

[0246] It can be understood that the memory in the embodiments of the present application can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory can be a random access memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchlink dynamic random access memory (SLDRAM), and direct rambus random access memory (DR RAM). It should be noted that the memory of the systems and methods described herein is intended to include but not be limited to these and any other suitable types of memory.

[0247] It should be understood that the above-mentioned memory is by way of example but not limitation. For example, the memory in the embodiments of the present application can also be a static random access memory (SRAM), a dynamic random access memory (DRAM), a synchronous dynamic random access memory (SDRAM), a double data rate synchronous dynamic random access memory (DDR SDRAM), an enhanced synchronous dynamic random access memory (ESDRAM), a synchlink dynamic random access memory (SLDRAM), and a direct rambus random access memory (DR RAM), etc. That is to say, the memory in the embodiments of the present application is intended to include but not be limited to these and any other suitable types of memory.

[0248] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0249] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0250] In several embodiments provided in this application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there can be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces, and the indirect couplings or communication connections of the devices or units can be in electrical, mechanical, or other forms.

[0251] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0252] In addition, the functional units in the embodiments of this application can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit.

[0253] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to enable a computer device (which can be a personal computer, a server) to execute all or part of the steps of the methods described in various embodiments of this application. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, ROM, RAM, magnetic disks, or optical discs that can store program codes.

[0254] As described above, it is only the specific implementation manner of this application. However, the protection scope of this application is not limited thereto. Any person skilled in the art within the technical scope disclosed by this application can easily think of changes or substitutions, which should all be covered within the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.

Claims

1. A recommendation method, characterized in that, Including: Obtaining target object information; Obtaining a heterogeneous graph corresponding to the target object information, where the heterogeneous graph includes multiple node items and edges between the node items, the edges between the node items represent the association relationships between the node items, and the multiple node items include at least two types of node item types; According to the heterogeneous graph, determining a first network object information set corresponding to at least one first object node item among the multiple node items, and a second network object information set corresponding to at least one second object node item among the multiple node items; Using the first network object information set and the second network object information set to recommend a first target network object to the target object corresponding to the target object information; Wherein, the first object node item is an object node item corresponding to the active party object of the behavior in the heterogeneous graph, and the second object node item is an object node item corresponding to the passive party object of the behavior in the heterogeneous graph; If it is detected that the operation information of the target object for the first target network object meets a preset condition, then according to the heterogeneous graph, determining a target network object category node item connected to the network object node item corresponding to the first target network object; Obtaining a target neighbor network object node item of the target network object category node item; Using the target neighbor network object node item to recommend a second target network object to the target object.

2. The method according to claim 1, wherein According to the heterogeneous graph, determining the first network object information set corresponding to at least one first object node item among the multiple node items includes: For each first object node item among the at least one first object node item, determining a first network object association weight between the first object node item and a first neighbor network object node item of the first object node item according to the heterogeneous graph; Determining a first object association weight between the first neighbor object node item of the first object node item and the first object node item according to the heterogeneous graph, and a second network object association weight between the first neighbor object node item and a second neighbor network object node item of the first neighbor object node item; Based on the first neighbor network object node item of the first object node item, the first network object association weight, the second neighbor network object node item of the first neighbor object node item, the second network object association weight, and the first object association weight, determining the first network object information corresponding to the first object node item; Using the first network object information corresponding to each first object node item to determine the first network object information set corresponding to the at least one first object node item.

3. The method according to claim 2, characterized in that Based on the first neighbor network object node item of the first object node item, the first network object association weight, the second neighbor network object node item of the first neighbor object node item, the second network object association weight, and the first object association weight, determining the first network object information corresponding to the first object node item includes: Based on the first neighbor network object node item of the first object node item and the first network object association weight, determining the first initial network object information of the first object node item; Determine the first neighbor network object information of the first neighbor object node item based on the second neighbor network object node item of the first neighbor object node item and the second network object association weight; Use the first initial network object information, the first neighbor network object information, and the first object association weight to determine the first network object information corresponding to the first object node item.

4. The method according to claim 1, wherein Using the first network object information set and the second network object information set to recommend a first target network object to the target object corresponding to the target object information includes: Determine the target network object information set corresponding to the multiple node items using the first network object information set and the second network object information set; Recommend a first target network object to the target object corresponding to the target object information based on the target network object information set.

5. The method according to claim 4, characterized in that At least one of the multiple node items belongs to both the first object node item and the second object node item; Determining the target network object information set corresponding to the multiple node items using the first network object information set and the second network object information set includes: Use the network object information corresponding to the first target object node item that does not belong to the second object node item among the at least one first object node item in the first network object information set as the network object information corresponding to the node item corresponding to the first target object node item; Use the network object information corresponding to the second target object node item that does not belong to the first object node item among the at least one second object node item in the second network object information set as the network object information corresponding to the node item corresponding to the second target object node item; For the duplicate node items that belong to both the first object node item and the second object node item among the multiple node items, determine the network object information corresponding to the node item corresponding to the duplicate node item based on the first network object information corresponding to the duplicate node item in the first network object information set, the first preset weight, the second network object information corresponding to the duplicate node item in the second network object information set, and the second preset weight; Use the network object information corresponding to the node item corresponding to the first target object node item, the network object information corresponding to the node item corresponding to the second target object node item, and the network object information corresponding to the node item corresponding to the duplicate node item to determine the target network object information set corresponding to the multiple node items.

6. The method according to claim 4, characterized in that, Recommending a first target network object to the target object corresponding to the target object information based on the target network object information set includes: Query the candidate node items that match the target object information from the multiple node items; Determine the target network object information corresponding to the candidate node items based on the candidate node items and the target network object information set; Recommend a first target network object to the target object according to the target network object information.

7. The method according to claim 4, wherein The method further includes: Obtain the attribute information corresponding to each node item among the multiple node items; For each of the node items, determine the attribute association weight between the node item and its neighboring node item according to the attribute information corresponding to the node item and the attribute information of the neighboring node item of the node item; Recommending a first target network object to the target object corresponding to the target object information based on the target network object information set includes: recommending a first target network object to the target object corresponding to the target object information based on the target network object information set, the attribute information corresponding to each node item in the multiple node items, and the attribute association weight between each node item and its neighboring node item.

8. The method according to claim 7, wherein Recommending a first target network object to the target object corresponding to the target object information based on the target network object information set, the attribute information corresponding to each node item in the multiple node items, and the attribute association weight between each node item and its neighboring node item includes: According to the heterogeneous graph, determine the first node association weight between each node item and its neighboring node item in the multiple node items; Use the target network object information set, the first node association weight between each node item and its neighboring node item, the attribute information corresponding to each node item in the multiple node items, and the attribute association weight between each node item and its neighboring node item to determine the embedding representation corresponding to each node item in the multiple node items, and obtain the embedding representation set corresponding to the multiple node items; Based on the embedding representation set, recommend a first target network object to the target object corresponding to the target object information.

9. A recommendation method, characterized in that, Includes: When a browsing instruction of the target object for the target page is detected, obtain the target object information of the target object; Send the target object information to the first device; Receive and display the recommendation information feedback by the first device for the target object information; Wherein, the recommendation information is that the first device obtains the target object information and the heterogeneous graph corresponding to the target object information. The heterogeneous graph includes multiple node items and edges between the node items. The edges between the node items represent the association relationship between the node items. The multiple node items include at least two types of node items. And according to the heterogeneous graph, determine the first network object information set corresponding to at least one first object node item in the multiple node items, and the second network object information set corresponding to at least one second object node item in the multiple node items, and use the first network object information set and the second network object information set to determine the recommendation information corresponding to the first target network object corresponding to the target object after determining the first target network object corresponding to the target object; wherein, the first object node item is the object node item corresponding to the active party object of the behavior in the heterogeneous graph, and the second object node item is the object node item corresponding to the passive party object of the behavior in the heterogeneous graph; The first device is further configured to: if it detects that the operation information of the target object for the first target network object meets a preset condition, determine, according to the heterogeneous graph, a target network object category node item connected to the network object node item corresponding to the first target network object; obtain the target neighboring network object node items of the target network object category node item; and recommend a second target network object to the target object by using the target neighboring network object node items.

10. A server, characterized in that, The server includes: A first obtaining unit, configured to obtain target object information; A second obtaining unit, configured to obtain a heterogeneous graph corresponding to the target object information, where the heterogeneous graph includes a plurality of node items and edges between the node items, the edges between the node items represent the association relationships between the node items, and the plurality of node items include at least two types of node items; A determining unit, configured to determine, according to the heterogeneous graph, a first network object information set corresponding to at least one first object node item among the plurality of node items, and a second network object information set corresponding to at least one second object node item among the plurality of node items; A recommending unit, configured to recommend a first target network object to the target object corresponding to the target object information by using the first network object information set and the second network object information set; wherein the first object node item is an object node item corresponding to the active party object of the behavior in the heterogeneous graph, and the second object node item is an object node item corresponding to the passive party object of the behavior in the heterogeneous graph; The server is further configured to: if it detects that the operation information of the target object for the first target network object meets a preset condition, determine, according to the heterogeneous graph, a target network object category node item connected to the network object node item corresponding to the first target network object; obtain the target neighboring network object node items of the target network object category node item; and recommend a second target network object to the target object by using the target neighboring network object node items.

11. A client, characterized in that, including: An obtaining unit, configured to obtain the target object information of the target object when detecting a browsing instruction of the target object for a target page; A sending unit, configured to send the target object information to a first device; A receiving unit, configured to receive and display the recommendation information fed back by the first device for the target object information; Among them, the recommended information is that the first device obtains the target object information and the heterogeneous graph corresponding to the target object information. The heterogeneous graph includes multiple node items and edges between the node items. The edges between the node items represent the association relationships between the node items. The multiple node items include at least two types of node items. According to the heterogeneous graph, a first network object information set corresponding to at least one first object node item among the multiple node items, a second network object information set corresponding to at least one second object node item among the multiple node items, and the recommended information corresponding to the first target network object determined by using the first network object information set and the second network object information set and fed back after determining the first target network object corresponding to the target object are obtained; among them, the first object node item is the object node item corresponding to the active party object of the behavior in the heterogeneous graph, and the second object node item is the object node item corresponding to the passive party object of the behavior in the heterogeneous graph. The first device is further configured to: if it is detected that the operation information of the target object for the first target network object meets a preset condition, determine a target network object category node item connected to the network object node item corresponding to the first target network object according to the heterogeneous graph; obtain a target neighboring network object node item of the target network object category node item; and recommend a second target network object to the target object by using the target neighboring network object node item.

12. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, and the computer program is suitable for being loaded by a processor to execute the steps in any one of claims 1-8 or the recommendation method in claim 9.

13. A computer device, characterized in that, The computer device includes a processor and a memory. A computer program is stored in the memory, and the processor is configured to execute the steps in any one of claims 1-8 or the recommendation method in claim 9 by calling the computer program stored in the memory.

14. A computer program product, comprising computer instructions, characterized in that, When the computer instruction is executed by the processor, the steps in any one of claims 1-8 or the recommendation method in claim 9 are implemented.

Images