Information sharing processing method and device, equipment and storage medium

Abstract

Embodiments of the present application provide a kind of information sharing processing method, device, equipment and storage medium, in the embodiments of the present application, asymmetric sharing interaction relationship between users is modeled first, time information and asymmetric sharing relationship can be coupled simultaneously, respectively aggregate sharing user and candidate shared user respectively in sharing and receiving two directions of historical neighbor user information, to generate the sharing representation information of corresponding user at this time, and the sharing representation information of shared object, then, based on these sharing representation information, predict the sharing probability of sharing user to candidate shared user sharing target object at sharing time, finally, information sharing is carried out based on sharing probability.Therefore, it automatically recommends the shared user who is more interested in the target object to be shared for sharing user, reduces the degree of complexity of information sharing, improves information sharing efficiency and user experience.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to an information sharing processing method, apparatus, device, and storage medium. Background Art

[0002] In recent years, e-commerce platforms have gradually integrated social features, such as product sharing. This allows users to not only browse and purchase products through e-commerce platforms, but also share their favorite products with friends, enabling information interaction with friends. Currently, e-commerce platforms provide a sharing portal on the product details page. Users click on this portal to enter the sharing page, which displays a third-party application icon. Users click on the third-party application icon to jump to the third-party application's user list page. From this page, users can select the target users to share with. The e-commerce platform then shares the product information with these selected users.

[0003] The existing information sharing operations on e-commerce platforms are relatively cumbersome and the path is long. It also takes a certain amount of time to select target users on the user list page. The information sharing efficiency is low and the user experience is poor. Summary of the Invention

[0004] Multiple aspects of the present application provide an information sharing processing method, apparatus, device and storage medium to reduce the complexity of information sharing and improve information sharing efficiency and user experience.

[0005] An embodiment of the present application provides an information sharing processing method, including: responding to an information sharing request initiated by a target sharing user for a target object at time t, and aggregating the first historical neighbor information of the target sharing user and the candidate shared user in the sharing dimension and the second historical neighbor information in the receiving dimension, as well as the historical neighbor information of the target object based on historical sharing behavior data; generating sharing representation information of the target sharing user and the candidate shared user at time t based on the first historical neighbor information of the target sharing user and the candidate shared user in the sharing dimension and the second historical neighbor information of the candidate shared user, and generating sharing representation information of the target object at time t based on the historical neighbor information of the target object; predicting the sharing probability of the target sharing user sharing the target object with the candidate shared user at time t based on the sharing representation information of the target sharing user, the candidate shared user and the target object at time t; and providing the information of the target shared user among the candidate shared users to the target sharing user based on the sharing probability, so that the target sharing user shares the target object with the target shared user.

[0006] An embodiment of the present application also provides an information sharing processing device, including: a neighbor information acquisition module, used to respond to an information sharing request initiated by a target sharing user for a target object at time t, and aggregate the first historical neighbor information of the target sharing user and the candidate shared user in the sharing dimension and the second historical neighbor information in the receiving dimension, and aggregate the historical neighbor information of the target object based on historical sharing behavior data; a representation information generation module, used to generate sharing representation information of the target sharing user and the candidate shared user at time t based on the first historical neighbor information of the target sharing user and the candidate shared user in the sharing dimension and the second historical neighbor information of the target shared user in the receiving dimension, and generate sharing representation information of the target object at time t based on the historical neighbor information of the target object; a sharing probability prediction module, used to predict the sharing probability of the target sharing user sharing the target object with the candidate shared user at time t based on the sharing representation information of the target sharing user, the candidate shared user and the target object at time t; a user information recommendation module, used to provide the information of the target shared user among the candidate shared users to the target sharing user based on the sharing probability, so that the target sharing user shares the target object with the target shared user.

[0007] An embodiment of the present application also provides a computer device, including: a memory and a processor; the memory is used to store a computer program, and the processor is coupled to the memory and is used to execute the computer program to implement the steps in the information sharing processing method.

[0008] An embodiment of the present application also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the processor is enabled to implement the steps in the information sharing processing method.

[0009] In the embodiment of the present application, an asymmetric sharing interaction relationship between users is innovatively modeled, which can simultaneously couple time information and asymmetric sharing relationships, and aggregate the historical neighbor user information of the sharing user and the candidate shared user in both the sharing and receiving directions, respectively, to generate the sharing representation information of the corresponding user at that time, as well as the sharing representation information of the shared object. Then, based on these sharing representation information, the probability of the sharing user sharing the target object to the candidate shared user at the sharing moment is predicted, and finally, based on the sharing probability, the target shared user is locked and information is shared. In this way, the characteristics of the user's sharing preference changing over time are fully considered, and the user's current sharing willingness and receiving willingness are taken into account, and the shared user who is more interested in the target object to be shared is automatically recommended to the sharing user, which is conducive to shortening the time of searching for the shared user, reducing the tediousness of information sharing, and improving information sharing efficiency and user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0011] Figure 1 A schematic diagram of an application scenario provided in an embodiment of the present application;

[0012] Figure 2 A flowchart of an information sharing processing method provided in an embodiment of the present application;

[0013] Figure 3 A schematic diagram of another application scenario provided by an embodiment of the present application;

[0014] Figure 4 A flowchart of another information sharing processing method provided in an embodiment of the present application;

[0015] Figure 5 A schematic diagram of the structure of an information sharing processing device provided in an embodiment of the present application;

[0016] Figure 6 A schematic diagram of the structure of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0017] To make the purpose, technical solutions, and advantages of this application more clear, the technical solutions of this application will be clearly and completely described below in conjunction with the specific embodiments of this application and the corresponding drawings. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0018] In the embodiments of the present application, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the access relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent three situations: A exists alone, A and B exist at the same time, and B exists alone, where A and B can be singular or plural. In the text description of the present application, the character " / " generally indicates that the previous and next associated objects are in an "or" relationship. In addition, in the embodiments of the present application, "first", "second", "third", "fourth", "fifth" and "sixth" are only used to distinguish the contents of different objects and have no other special meanings.

[0019] First, some terms involved in the embodiments of this application are explained:

[0020] TDG (Time Dynamic Graph) is a dynamic graph with time attributes that change over time. The nodes and edges in the graph have time attributes, and the edges and nodes may appear and disappear. Assume that the graph is defined as G = (V, E), V = {(v,t s ,t e )}. v is a node in the graph, E is an edge in the graph, t s is the time point when node v appears, t e is the time point when node v disappears, t s ≤t e E={(u,v,t s * ,t e * )}, where u∈V, v∈V, t s * is the time point when the edge connecting node u and node v appears, t e * is the time point when the edge connecting node u and node v disappears.

[0021] CTDG (Continuous Time Dynamic Graph): A time dynamic graph that can model fine-grained time information. Assume that the graph is defined as G = (V, E, T), where V is a node set consisting of multiple nodes, E is an edge set consisting of multiple edges, and T: E→R + ,T is a function that maps each edge in E to the corresponding timestamp. Each edge in E can be assigned a timestamp t∈R+ with the finest time granularity, where R+ is a real number. Such a graph G is called CTDG.

[0022] TTP (Temporal Point Process): A random process consisting of a series of (labeled) events in the continuous time domain.

[0023] GAT (Graph Attention Network): uses the attention mechanism to assign different weights to different neighbor nodes, and then aggregates neighbor information based on this weight.

[0024] GRU (Gate Recurrent Unit) is a recurrent neural network proposed to solve the problem of gradient disappearance in long-term memory and back propagation.

[0025] The temporal attention network is an attention network based on a temporal mechanism, such as the temporal-based GAT.

[0026] At present, the existing information sharing operations on e-commerce platforms are relatively cumbersome, with long paths, and it takes a certain amount of time to select the target user in the user list page. The information sharing efficiency is low and the user experience is poor. To this end, in some embodiments of the present application, an asymmetric sharing interaction relationship between users is innovatively modeled, which can simultaneously couple time information and asymmetric sharing relationships, and aggregate the historical neighbor user information of the sharing user and the candidate shared user in both the sharing and receiving directions, respectively, to generate the sharing representation information of the corresponding user at this time, as well as the sharing representation information of the shared object. Then, based on these sharing representation information, the sharing probability of the sharing user sharing the target object to the candidate shared user at the sharing moment is predicted, and finally, based on the sharing probability, the target shared user is locked and information sharing is performed. Thus, the characteristics of the user's sharing preference changing over time are fully considered, and the shared user who is more interested in the target object to be shared is automatically recommended to the sharing user, which is conducive to shortening the time of searching for the shared user, reducing the tediousness of information sharing, and improving information sharing efficiency and user experience.

[0027] Figure 1 This is a schematic diagram of an application scenario provided by the embodiment of this application. Please refer to Figure 1 In this application scenario, the terminal device 10 can interact with the server 30 through a wired network or a wireless network. For example, a wired network may include coaxial cables, twisted pairs, and optical fibers, and a wireless network may be a 2G network, a 3G network, a 4G network, or a 5G network, a Wireless Fidelity (WIFI) network, or the like. This application does not limit the specific type or form of interaction, as long as it can realize the function of interaction between the terminal device 10 and the server 30. Optionally, terminal devices include, but are not limited to, mobile phones, tablet computers, laptops, wearable devices, and vehicle-mounted devices. Servers include, but are not limited to, conventional servers, cloud servers, various server instances with computing and processing capabilities (such as virtual machines, containers, etc.), server arrays, or distributed server clusters composed of multiple servers. It should be understood that Figure 1 The number of terminal devices and servers is only for reference.

[0028] exist Figure 1In the application scenario shown, a user shares the product objects he browses with other users through an e-commerce application as an example. The user opens the shopping app (application) provided by the e-commerce platform installed on his terminal device, enters the product information page and browses the product's pictures, texts and other related information. The product information page can be the home page of the shopping app, the product details page, the store page, the shopping cart page, the comment page, etc. There is no limitation on this. Any page that can display product information is applicable to the embodiment of this application. For the product objects of interest, see Figure 1 As shown in Figure 1, the sharing user clicks the share control on the product information page to trigger an information sharing operation that shares the product object's access link (such as a Uniform Resource Locator (URL)) or other information such as images and text with other users. The terminal device responds to the user-initiated information sharing operation and sends an information sharing request to the server. In response to the information sharing request, the server selects some users with a higher sharing probability from the candidate shared users as target shared users and returns the target shared users' information to the terminal device. The terminal device then displays the target shared users' information to the sharing user, so that the sharing user can share the link to the shared product with these target shared users. Optionally, the candidate shared users can be users in the sharing user's friend list. The server can obtain the sharing user's friend list and select all or some users in the friend list as candidate shared users. Further, the server can analyze the sharing user's historical behavior data to determine which users in the friend list have similar historical behavior or interests to the sharing user and select these users as candidate shared users. Alternatively, the server can select some users in the friend list who have recently interacted frequently with the sharing user as candidate shared users, or directly select all users in the friend list as candidate shared users. Alternatively, candidate users can also be users who have a sharing relationship with the sharing user and are obtained from the sharing user's historical behavior data. This can include users who have shared product information with the sharing user as sharers, as well as users who have had product information shared with the sharing user as recipients. The method for obtaining candidate users from the sharing user's historical behavior data can be found in the following embodiments and will not be described in detail here. These target users are selected from the candidate users who are more interested in the shared product.

[0029] Specifically, see Figure 1As shown in ②, the server first performs the aggregation operation of historical neighbor information, and can obtain the first historical neighbor information on the sharing dimension and the second historical neighbor information on the receiving dimension corresponding to the sharing user, the candidate shared user, and the commodity object to be shared. It should be noted that for commodity objects, their historical neighbor information includes both users who have shared the commodity object as sharers (i.e., historical neighbor information on the sharing dimension) and users who have received the commodity object or its information as sharees (i.e., historical neighbor information on the receiving dimension); however, for commodity objects, their historical neighbor information includes both historical neighbor information on the sharing dimension and historical neighbor information on the receiving dimension, but there is no need to distinguish between the two, and they can be unified as the historical neighbor information of the commodity object. Next, see Figure 1 As shown in ③, the server performs the generation operation of sharing representation information to obtain the sharing representation information of the sharing user, the candidate shared user, and the product to be shared; then, see Figure 1 As shown in ④, the server performs a sharing probability prediction operation to predict the probability of the sharing user sharing the desired product object with each candidate shared user. Figure 1 As shown in step ⑤, the server determines the target user from the candidate users based on the sharing probability and provides the target user's information to the terminal device. Finally, the terminal device displays the target user's information provided by the server in the form of a list. When the sharing user triggers the sharing control associated with one or more users in the "shared user list", the terminal device sends the access entry link or other information such as images and text of the product object to be shared to the corresponding user, thereby achieving the purpose of sharing the product object with other users, thus completing the process of product sharing between users.

[0030] Of course, there is no limitation on the application scenarios to which the technical solutions provided in the embodiments of the present application are applicable. Figure 1 This is merely an exemplary application scenario. The technical solution of the embodiments of this application can share not only commodity objects, but also other objects, such as recruitment information, rental information, travel information, media information, and media social information, etc. Different information objects in these different application scenarios can be shared using the methods provided by the embodiments of this application.

[0031] In addition, Figure 1 In the application scenario shown, the server executes the information processing logic in the information sharing process (i.e. Figure 1The following description will focus on the processing logic shown in ②-⑤ in the figure, but is not limited to this. For example, as terminal devices become more powerful, terminal devices may replace the server to perform one or more of the following operations: historical neighbor information aggregation, shared representation information generation, and shared probability prediction. This is not a limitation.

[0032] The following describes in detail the technical solutions provided by various embodiments of the present application in conjunction with the accompanying drawings.

[0033] Figure 2 This is a flow chart of an information sharing method provided in an embodiment of the present application. Figure 2 As shown, the method may include the following steps:

[0034] 201. Respond to the information sharing request initiated by the target sharing user for the target object at time t, and aggregate the first historical neighbor information of the target sharing user and the candidate shared users in the sharing dimension and the second historical neighbor information in the receiving dimension based on the historical sharing behavior data, as well as the historical neighbor information of the target object.

[0035] 202. Generate sharing representation information of the target sharing user and the candidate shared user at time t based on their respective first historical neighbor information in the sharing dimension and second historical neighbor information in the receiving dimension, and generate sharing representation information of the target object at time t based on the historical neighbor information of the target object.

[0036] 203. Predict the sharing probability of the target sharing user sharing the target object with the candidate sharing users at time t based on the sharing representation information of the target sharing user, the candidate sharing users, and the target object at time t.

[0037] 204. Provide information of a target user among candidate users to the target sharing user based on the sharing probability, so that the target sharing user shares the target object with the target user.

[0038] Among them, step 201 can be simply referred to as the aggregation operation of historical neighbor information, step 202 can be simply referred to as the generation operation of sharing representation information, and step 203 can be simply referred to as the prediction operation of sharing probability, which can be understood as the implementation of ②, ③ and ④ in the above application scenario embodiment.

[0039] Specifically, the target sharing user is any user who currently has a sharing demand for the target object. In order to distinguish them from the sharing users involved in historical sharing behaviors, the user who currently has a sharing demand is referred to as the target sharing user. The target object can be, for example, a commodity object, electronic document, recruitment information, rental information, travel information, media information, media social information, or instant messaging information, etc., without limitation. The target sharing user can initiate an information sharing request for the target object at any time. For ease of understanding and distinction, the sharing moment when the user initiates the information sharing request for the target object is recorded as time t. Time t is the information sharing moment, which can be any time.

[0040] In this embodiment, in response to an information sharing request initiated by a target sharing user for a target object at time t, an aggregation operation of historical neighbor information is performed, i.e., step 201 is executed. Specifically, based on the historical sharing behavior data, the first historical neighbor information of the target sharing user and the candidate shared users in the sharing dimension and the second historical neighbor information in the receiving dimension can be aggregated, as well as the historical neighbor information of the target object.

[0041] Further optionally, a temporal attention network based on CTDG can be used to aggregate the first historical neighbor information of the target sharing user and the candidate shared user in the sharing dimension and the second historical neighbor information in the receiving dimension, as well as the historical neighbor information of the target object, based on the historical sharing behavior data. First, a CTDG describing the sharing behavior between different users is constructed based on the historical sharing behavior data, and a CTDG describing the sharing behavior between the target object and the user is constructed. Then, on the one hand, the CTDG describing the sharing behavior between different users is processed using the temporal attention network to obtain the first historical neighbor information of the target sharing user and the candidate shared user in the sharing dimension and the second historical neighbor information in the receiving dimension. On the other hand, the historical neighbor information of the target object is obtained by processing the CTDG constructed to describe the sharing behavior between the target object and the user.

[0042] Based on the above, further optionally, when performing the aggregation operation of historical neighbor information, the first CTDG of the target sharing user and the candidate shared user in the sharing dimension and the second CTDG in the receiving dimension, as well as the third CTDG of the target object can be obtained based on the historical sharing behavior data; the first CTDG of the target sharing user and the candidate shared user in the sharing dimension and the second CTDG in the receiving dimension are respectively input into the temporal attention network for feature learning to obtain the first historical neighbor information of the target sharing user and the candidate shared user in the sharing dimension and the second historical neighbor information in the receiving dimension; the third CTDG of the target object is input into the temporal attention network for feature learning to obtain the historical neighbor information of the target object.

[0043] In actual applications, the sharing behavior data generated in the historical time period before time t in the application scenario to which the information sharing belongs can be collected, that is, multiple historical sharing behavior data can be collected. For example, if the information sharing belongs to the sharing of commodity objects in the e-commerce scenario, the sharing behavior data of commodity objects generated in the historical time period before time t on the entire e-commerce platform can be obtained. For example, if the information sharing belongs to the sharing of media information information in the media information scenario, the sharing behavior data of media information information generated in the historical time period before time t on the entire media information platform can be obtained. For another example, if the information sharing belongs to the sharing of instant messaging information in the instant messaging scenario, the sharing behavior data of instant messaging information generated in the historical time period before time t on the entire instant messaging platform can be obtained. Regardless of the application scenario, each historical sharing behavior data includes, for example, but is not limited to: user information of the sharing user who initiated the sharing behavior, user information of the shared user who accepted the sharing behavior, sharing time information when the sharing behavior was initiated, and information about the sharing objects involved in the sharing behavior, etc. The sharing behavior in this embodiment is an asymmetric sharing behavior. On the one hand, it is reflected in the fact that the sharing behavior is a three-way interactive behavior, which involves the interactive relationship between the sharing user, the shared user, and the sharing object. On the other hand, the sharing behavior is also asymmetric between users, that is, user A initiates a sharing behavior to user B, but user B may not necessarily initiate a sharing behavior to user A.

[0044] Next, multiple historical sharing behavior data are analyzed to construct a global CTDG describing sharing behavior between different users from the dimensions of sharing users, shared users, and shared objects. This global CTDG includes user nodes, each representing a sharing user or a shared user, and edges from the user node corresponding to the sharing user to the user node corresponding to the shared user. Information such as the shared object and sharing time in the sharing behavior corresponding to these two user nodes is stored as attribute information of each edge. Based on this global CTDG, the first CTDG of the target sharing user and the candidate shared user in the sharing dimension and the second CTDG in the receiving dimension can be obtained. In other words, on the one hand, the first CTDG of the target sharing user in the sharing dimension and the second CTDG in the receiving dimension can be obtained from the global CTDG, and on the other hand, the first CTDG of each candidate shared user in the sharing dimension and the second CTDG in the receiving dimension can be obtained from the global CTDG. Each candidate shared user corresponds to a first CTDG in the sharing dimension and a second CTDG in the receiving dimension.

[0045] Based on the above, as an example, according to the historical sharing behavior data, one implementation method of obtaining the first CTDG of the target sharing user and the candidate shared user in the sharing dimension and the second CTDG in the receiving dimension is: from the historical sharing behavior data, determine the sharing user, shared user, sharing object information and sharing time information involved in each historical sharing behavior; for each historical sharing behavior, abstract the sharing user and the shared user as user nodes, create a directed edge from the sharing user to the shared user between the user nodes, and use the sharing object information and sharing time information as the attribute information of the directed edge to obtain the global CTDG; take the target sharing user and the candidate shared user as query objects respectively, obtain the local CTDG when the query object is the sharing user from the global CTDG as its first CTDG in the sharing dimension, and obtain the local CTDG when the query object is the shared user as its second CTDG in the receiving dimension.

[0046] Specifically, historical sharing behavior data is analyzed to determine at least one sharing user and at least one shared user associated with each sharing user; at least one sharing user and at least one shared user associated with each sharing user are abstracted as a user node respectively; for each sharing user and its associated shared user, a directed edge connecting the corresponding user nodes is created, the directed edge represents the sharing behavior between the corresponding users, and the direction of the directed edge is from the user node corresponding to the sharing user to the user node corresponding to the shared user associated with it, and the sharing object information and sharing time information are obtained from the corresponding historical sharing behavior, and the sharing object information and sharing time information are used as attribute information of the directed edge, thereby constructing a global CTDG.

[0047] Specifically, when the query object is the target sharing user, the local CTDG of the target sharing user as a sharing user (i.e., sharer) can be obtained from the global CTDG as its first CTDG in the sharing dimension, and the local CTDG of the target sharing user as a shared user (i.e., sharee) can be obtained as its second CTDG in the receiving dimension.

[0048] Among them, for the first CTDG of the target sharing user in the sharing dimension obtained from the global CTDG, the first CTDG includes the user node corresponding to the target sharing user and the user node corresponding to at least one shared user who received the sharing behavior initiated by the target sharing user. The directed edges in the first CTDG are from the user node corresponding to the target sharing user to the user node corresponding to the shared user. In other words, the first CTDG of the target sharing user in the sharing dimension describes the sharing behavior between the target sharing user as a sharing user and its corresponding shared user. For the second CTDG of the target sharing user in the receiving dimension obtained from the global CTDG, the second CTDG includes the user node corresponding to the target sharing user and the user node corresponding to at least one sharing user who initiated the sharing behavior toward the target sharing user. The directed edges in the second CTDG are from the user node that initiated the sharing behavior toward the target sharing user to the user node corresponding to the target sharing user. In other words, the second CTDG of the target sharing user in the receiving dimension describes the sharing behavior between the target sharing user as a shared user and its corresponding sharing user.

[0049] Specifically, when the query object is a candidate shared user, the local CTDG of the candidate shared user as a sharing user (i.e., sharer) can be obtained from the global CTDG as its first CTDG in the sharing dimension, and the local CTDG of the candidate shared user as a shared user (i.e., sharee) can be obtained as its second CTDG in the receiving dimension.

[0050] For the first CTDG of a candidate shared user in the sharing dimension obtained from the global CTDG, the first CTDG includes the user node corresponding to the candidate shared user and the user node corresponding to at least one shared user that received the sharing action initiated by the candidate shared user. Directed edges in the first CTDG point from the user node corresponding to the candidate shared user to the user node corresponding to the corresponding shared user. In other words, the first CTDG of a candidate shared user in the sharing dimension describes the sharing action between the candidate shared user as a sharing user and its corresponding shared user. For the second CTDG of a candidate shared user in the receiving dimension obtained from the global CTDG, the second CTDG includes the user node corresponding to the candidate shared user and the user node corresponding to at least one sharing user that initiated the sharing action toward the candidate shared user. Directed edges in the second CTDG point from the user node that initiated the sharing action toward the candidate shared user to the user node corresponding to the candidate shared user. In other words, the second CTDG of a candidate shared user in the receiving dimension describes the sharing action between the candidate shared user as a shared user and its corresponding sharing user.

[0051] As an optional implementation method, the third CTDG of the target object is obtained based on historical sharing behavior data, including: determining the sharing user, shared user and sharing object involved in each historical sharing behavior from the historical sharing behavior data; for each historical sharing behavior, abstracting the sharing user and shared user into user nodes, abstracting the sharing object into object nodes, creating a directed edge from the user node corresponding to the sharing user to the object node, and a directed edge from the object node to the user node corresponding to the shared user, so as to obtain the third CTDG of the target object.

[0052] Specifically, historical sharing behavior data is analyzed to determine at least one sharing user and at least one shared user associated with each sharing user, as well as the shared object shared by each sharing user to the corresponding shared user; at least one sharing user and at least one shared user associated with each sharing user are abstracted as a user node, and the shared object shared by each sharing user to the corresponding shared user is abstracted as an object node; for each sharing user and its associated shared user and object, a directed edge is created from the user node corresponding to the sharing user to the object node corresponding to its associated object, as well as a directed edge from the corresponding object node to the user node corresponding to the associated shared user, thereby constructing the third CTDG of the target object.

[0053] In one application scenario, the candidate shared user is a user in the target sharing user's friend list. After determining the target sharing user, the candidate shared user can be directly determined from the target sharing user's friend list. In another application scenario, historical users who have a sharing relationship with the target sharing user can be identified as candidate shared users based on the target sharing user's historical behavior data. For the second application scenario, the candidate shared user can be determined during the analysis of the historical sharing behavior data before creating the first and second CTDGs of the candidate shared users. Specifically, in the process of analyzing historical sharing behavior data, at least one sharing user and at least one shared user associated with each sharing user can be determined; then, based on the information of the target sharing user, on the one hand, at least one shared user associated with the target sharing user when it was a sharing user (that is, the shared user who received the sharing behavior initiated by the target sharing user in the historical sharing behavior) is determined, and on the other hand, at least one sharing user associated with the target sharing user during the period when it was a shared user is determined (the sharing user who initiated the sharing behavior to the target sharing user in the historical sharing behavior). For the sake of ease of description and distinction, the sharing users who initiated the sharing behavior to the target sharing user in the historical sharing behavior and the shared users who received the sharing behavior initiated by the target sharing user are referred to as the associated user cluster of the target sharing user. Next, all users in the associated user cluster of the target sharing user can be used as candidate shared users for this sharing behavior, or some users can be selected from the associated user cluster of the target sharing user as candidate shared users for this sharing behavior. If some users are selected from the target sharing user's associated user cluster as candidate shared users for this sharing behavior, some users can be randomly selected according to a preset number of users, or some users who had sharing behaviors with the target sharing user closer in time can be given priority based on the order of sharing behaviors with the target sharing user, or some users who had sharing behaviors with the target sharing user more frequently can be given priority based on the frequency of sharing behaviors with the target sharing user, and so on.

[0054] In this embodiment, in the process of performing the aggregation operation of historical neighbor information, after obtaining the first CTDG of the target sharing user and the candidate shared user in the sharing dimension and the second CTDG in the receiving dimension, as well as the third CTDG of the target object based on the historical sharing behavior data, the first CTDG of the target sharing user and the candidate shared user in the sharing dimension and the second CTDG in the receiving dimension can be respectively input into the temporal attention network for feature learning to obtain the first historical neighbor information of the target sharing user and the candidate shared user in the sharing dimension and the second historical neighbor information in the receiving dimension; the third CTDG of the target object is input into the temporal attention network for feature learning to obtain the historical neighbor information of the target object.

[0055] Specifically, the target sharing user's first CTDG in the sharing dimension is input into the temporal attention network for feature learning to obtain the target sharing user's first historical neighbor information in the sharing dimension. This first historical neighbor information includes, but is not limited to: the user ID of the shared user connected to the target sharing user by a directed edge, the sharing time information when the target sharing user initiates the sharing behavior with the shared user, and the object ID of the sharing object, etc. This first historical neighbor information indicates which users the target sharing user is more inclined to share with in the sharing dimension, and is simply a representation of the target sharing user's sharing tendency or willingness in the sharing dimension.

[0056] The second CTDG of the target sharing user in the receiving dimension is input into the temporal attention network for feature learning to obtain the second historical neighbor information of the target sharing user in the receiving dimension. This second historical neighbor information includes, but is not limited to: the user ID of the sharing user connected to the target sharing user with a directed edge, the sharing time information of the sharing behavior initiated to the target sharing user, and the object ID of the sharing object, etc. This second historical neighbor information indicates which users are more inclined to initiate sharing behavior to the target sharing user in the receiving dimension, that is, the target sharing user is more inclined to receive information shared by which users. In short, it is a representation of the target sharing user's receiving tendency or willingness in the receiving dimension.

[0057] The first CTDG of the candidate shared user in the sharing dimension is input into the temporal attention network for feature learning to obtain the first historical neighbor information of the candidate shared user in the sharing dimension. This first historical neighbor information includes, but is not limited to: the user IDs of the shared users connected to the candidate shared user by a directed edge, the sharing time information when the candidate shared user initiates the sharing behavior with the shared user, and the object ID of the sharing object, etc. This first historical neighbor information indicates which users the candidate shared user is more inclined to share with in the sharing dimension, and is simply a representation of the candidate shared user's sharing tendency or willingness in the sharing dimension.

[0058] The second CTDG of the candidate shared user in the receiving dimension is input into the temporal attention network for feature learning to obtain the second historical neighbor information of the candidate shared user in the receiving dimension. This second historical neighbor information includes, but is not limited to: the user ID of the sharing user connected to the candidate shared user by a directed edge, the sharing time information of the sharing behavior initiated with the candidate shared user, and the object ID of the sharing object, etc. This second historical neighbor information indicates which users are more inclined to initiate sharing behavior with the candidate shared user in the receiving dimension, that is, which users' shared information the candidate shared user is more inclined to receive. In short, it represents the candidate shared user's receiving tendency or willingness in the receiving dimension.

[0059] The target object's third CTDG is input into the temporal attention network for feature learning to obtain the target object's historical neighbor information. This target object's historical neighbor information includes, but is not limited to, the user IDs of the sharing or shared users connected to the target object with directed edges, as well as the sharing time information and the target object's object ID. The target object's historical neighbor information indicates which users are more likely to share the target object in terms of sharing, and which users are more likely to receive the target object in terms of receiving. This represents the target object's tendency to be shared.

[0060] In this embodiment, after obtaining the first historical neighbor information of the target sharing user in the sharing dimension and the second historical neighbor information in the receiving dimension, the sharing representation information of the target sharing user at time t can be generated based on the first historical neighbor information of the target sharing user in the sharing dimension and the second historical neighbor information in the receiving dimension. Similarly, after obtaining the first historical neighbor information of the candidate shared user in the sharing dimension and the second historical neighbor information in the receiving dimension, the sharing representation information of the candidate shared user at time t can be generated based on the first historical neighbor information of the candidate shared user in the sharing dimension and the second historical neighbor information in the receiving dimension. After obtaining the historical neighbor information of the target object, the sharing representation information of the target object at time t can be generated based on the historical neighbor information of the target object.

[0061] In this embodiment, the sharing representation information of each user is divided into two parts, namely, the first sharing representation information on the sharing dimension (representing their sharing tendency or sharing willingness) and the second sharing representation information on the receiving dimension (representing their receiving tendency or tendency willingness). For example, for the target sharing user, their sharing representation information at a certain sharing moment includes the first sharing representation information on the sharing dimension at the sharing moment and the second sharing representation information on the receiving dimension at the sharing moment; similarly, for the candidate shared user, their sharing representation information at a certain sharing moment includes the first sharing representation information on the sharing dimension at the sharing moment and the second sharing representation information on the receiving dimension at the sharing moment. Among them, no matter which user, their first sharing representation information on the sharing dimension at time t can be determined based on their first historical neighbor information on the sharing dimension before time t, and their second sharing representation information on the receiving dimension at time t can be determined based on their second historical neighbor information on the receiving dimension before time t.

[0062] Based on the above analysis, in an optional embodiment, for the target sharing user's sharing representation information at time t, the target sharing user's first historical neighbor information in the sharing dimension and the second historical neighbor information in the receiving dimension can be directly combined to obtain the sharing representation information. This method is relatively simple and easy to implement. In another optional embodiment, in addition to the target sharing user's first historical neighbor information in the sharing dimension and the second historical neighbor information in the receiving dimension, the target sharing user's historical sharing representation information can be combined to generate the target sharing user's sharing representation information at time t. This method of simultaneously integrating historical sharing representation information is conducive to improving the accuracy of sharing representation information, and thus is conducive to improving the accuracy of subsequent sharing probabilities, thereby achieving accurate sharing recommendations. The historical sharing representation information of the target sharing user used in the embodiment of the present application can include sharing representation information obtained after one or more historical sharing operations. Preferably, the historical sharing representation information obtained after the last sharing operation can be used. By using the historical sharing representation information after the last sharing operation, only the historical sharing representation information after the historical sharing operation needs to be saved, which can save memory resources, be applicable to large-scale sharing scenarios, and at the same time ensure the accuracy of sharing recommendations.

[0063] In an optional implementation, the above-mentioned method of generating the sharing representation information of the target sharing user at time t based on the target sharing user's first historical neighbor information in the sharing dimension and the second historical neighbor information in the receiving dimension, as well as its historical sharing representation information, includes: extracting the target sharing user's first historical sharing representation information in the sharing dimension and the second historical sharing representation information in the receiving dimension from the historical sharing representation information of the target sharing user; performing feature fusion of the target sharing user's first historical neighbor information in the sharing dimension and the first historical sharing representation information to obtain the first sharing representation information of the target sharing user in the sharing dimension at time t; performing feature fusion of the target sharing user's second historical neighbor information in the receiving dimension and the second historical sharing representation information to obtain the second sharing representation information of the target sharing user in the receiving dimension at time t; performing feature splicing of the target sharing user's first sharing representation information in the sharing dimension and the second sharing representation information in the receiving dimension at time t to obtain the sharing representation information of the target sharing user at time t.

[0064] Furthermore, in an optional embodiment, for the first sharing representation information of the target sharing user at time t in the sharing dimension, the first historical neighbor information of the target sharing user at time t can be directly fused with the first historical sharing representation information to obtain the first sharing representation information of the target sharing user at time t in the sharing dimension. Correspondingly, for the second sharing representation information of the target sharing user at time t in the receiving dimension, the second historical neighbor information of the target sharing user at time t can be directly fused with the second historical sharing representation information to obtain the second sharing representation information of the target sharing user at time t in the receiving dimension.

[0065] Further optionally, considering that a user may not participate in sharing behavior for a long time, or may participate in sharing behavior frequently, and because different users have different perception abilities of different time intervals, the user's historical sharing representation information can be personalized projected through a time point process model, and the latest sharing representation information can be generated based on the projection result; wherein, in the case where a user does not participate in sharing behavior for a long time, the projection operation can have a certain penalty effect on the user's long-term non-participation in sharing behavior, avoid the lag of the user's historical sharing representation information, and improve the accuracy of the latest sharing representation information obtained based on the historical sharing representation information; for users who frequently participate in sharing behavior, the projection operation can have a certain reward effect on the user's long-term non-participation in sharing behavior, give play to the positive effect of historical sharing representation information, and improve the accuracy of the latest sharing representation information obtained based on historical sharing representation information. Based on this, in another optional embodiment, for the first sharing representation information of the target sharing user at time t on the sharing dimension, the first historical sharing representation information of the target sharing user at time t on the sharing dimension can also be projected according to the time point process model of the target sharing user to obtain the first projected sharing representation information of the target sharing user at the sharing dimension; the first historical neighbor information, the first historical sharing representation information and the first projected sharing representation information of the target sharing user at the sharing dimension are sent to the multi-layer perceptron (MLP) network for feature fusion to obtain the first sharing representation information of the target sharing user at time t on the sharing dimension. Correspondingly, for the second sharing representation information of the target sharing user at time t on the receiving dimension, the second sharing representation information of the target sharing user at time t on the receiving dimension can also be projected according to the time point process model of the target sharing user to obtain the second sharing representation information of the target sharing user at the receiving dimension; the second historical neighbor information, the second historical sharing representation information and the second projected sharing representation information of the target sharing user at the receiving dimension are sent to the multi-layer perceptron MLP network for feature fusion to obtain the second sharing representation information of the target sharing user at time t on the receiving dimension. Further optionally, as an example, based on the time point process model of the target sharing user, the first historical sharing representation information of the target sharing user on the sharing dimension is projected to obtain the first projected sharing representation information of the target sharing user on the sharing dimension. The next expected sharing moment of the target sharing user can be predicted based on the time point process model of the target sharing user; the projection factor is determined based on the difference between the next expected sharing moment and the moment t; based on the projection factor, the first historical sharing representation information of the target sharing user on the sharing dimension is projected to obtain the first projected sharing representation information of the target sharing user on the sharing dimension.Accordingly, based on the target sharing user's time point process model, the target sharing user's second sharing representation information on the receiving dimension is projected. When the target sharing user's second sharing representation information on the receiving dimension is obtained, the target sharing user's next expected sharing moment can be predicted based on the target sharing user's time point process model; the projection factor is determined based on the difference between the next expected sharing moment and the moment t; and based on the projection factor, the target sharing user's second historical sharing representation information on the receiving dimension is projected to obtain the target sharing user's second projected sharing representation information on the receiving dimension. It should be noted that the "predicting the next expected sharing moment" and "determining the projection factor" in the process of obtaining the first projected sharing representation information can be the same process as the "predicting the next expected sharing moment" and "determining the projection factor" in the process of obtaining the second projected sharing representation information, and are executed once.

[0066] In this embodiment, the expected time of the next interaction is calculated based on the timing point process, and a personalized projection factor is designed according to the difference between the expected interaction time and the actual interaction time. The projection factor can reflect the perception ability of different users for different time intervals, thereby realizing personalized projection user representation.

[0067] See also Figure 3 In the example, user u is taken as the target sharing user, and the historical sharing representation information h of user u is u (t-) is divided into the first historical sharing representation information h of user u in the sharing dimension u a (t-) (also denoted as Ua(t-)) and the second history sharing representation information h in the receiving dimension u b (t-)((also denoted as Ub(t-))), similarly, user u’s shared representation information h at time t u (t) is divided into the first sharing representation information h of user u in the sharing dimension at time t u a (t) (also denoted as Ua(t)) and the second shared representation information h of user u on the receiving dimension at time t u b (t)((also denoted as Ub(t))).

[0068] Based on the time point process model, the first historical sharing representation information h of user u in the sharing dimension is u a(t-) (i.e., Ua(t-)) is projected to obtain the first projection sharing representation information Ua(t) of user u at time t on the sharing dimension. Assume that the projection factor is denoted by σ; the difference between Ua(t-) and Ua(t) is denoted by Ua(t-)-Ua(t); the difference between the next expected sharing time and time t is denoted by t + -t, based on the time difference, the projection factor can be calculated, for example, σ=(t + -t) / t, accordingly, Ua(t-)-Ua(t) / Ua(t-)=σ, based on Ua(t-)-Ua(t) / Ua(t-)=σ, the projection Ua(t) of Ua(t-) can be calculated. Then, the first historical neighbor information of user u in the sharing dimension and the first historical sharing representation information h u a (t-) and the first projection sharing representation information Ua(t) are sent to the MLP network for feature fusion to obtain the first sharing representation information h of user u at time t in the sharing dimension u a (t).

[0069] Similarly, based on the time point process model, the second historical sharing representation information h of user u in the sharing dimension is u b (t-) (i.e., Ub(t-)) is projected to obtain the second projection sharing representation information Ub(t) of user u at time t on the sharing dimension. Assume that the projection factor is denoted as σ; the difference between Ub(t-) and Ub(t) is denoted as Ub(t-)-Ub(t); the difference between the next expected sharing time and time t is denoted as t + -t, based on the time difference, the projection factor can be calculated, for example, σ=(t + -t) / t, accordingly, Ub(t-)-Ub(t) / Ub(t-)=σ, based on Ub(t-)-Ub(t) / Ub(t-)=σ, the projection Ub(t) of Ub(t-) can be calculated. Then, the second historical neighbor information of user u in the sharing dimension and the second historical sharing representation information h u b (t-) and the second projection sharing representation information Ub(t) are sent to the MLP network for feature fusion to obtain the second sharing representation information h of user u at time t in the sharing dimension u b (t).

[0070] Similarly, for the sharing representation information of the candidate shared user at time t, the first historical neighbor information of the candidate shared user in the sharing dimension and the second historical neighbor information in the receiving dimension can be directly combined to obtain the information. This method is relatively simple and easy to implement. In another optional embodiment, in addition to the first historical neighbor information of the candidate shared user in the sharing dimension and the second historical neighbor information in the receiving dimension, the historical sharing representation information of the candidate shared user can be combined to generate the sharing representation information of the candidate shared user at time t. This method of simultaneously integrating historical sharing representation information is conducive to improving the accuracy of sharing representation information, and thus is conducive to improving the accuracy of subsequent sharing probabilities, thereby achieving accurate sharing recommendations. The historical sharing representation information of the candidate shared user adopted in the embodiment of the present application can include sharing representation information obtained after one or more historical sharing operations. Preferably, the historical sharing representation information obtained after the last sharing operation can be used. By using the historical sharing representation information after the last sharing operation, only the historical sharing representation information after the historical sharing operation needs to be saved once, which can save memory resources, be applicable to large-scale sharing scenarios, and at the same time ensure the accuracy of sharing recommendations.

[0071] In an optional implementation, the above-mentioned method of generating the sharing representation information of the candidate shared user at time t based on the candidate shared user's first historical neighbor information in the sharing dimension and the second historical neighbor information in the receiving dimension, as well as its historical sharing representation information, includes: extracting the candidate shared user's first historical sharing representation information in the sharing dimension and the second historical sharing representation information in the receiving dimension from the candidate shared user's historical sharing representation information; performing feature fusion of the candidate shared user's first historical neighbor information in the sharing dimension and the first historical sharing representation information to obtain the first sharing representation information of the candidate shared user in the sharing dimension at time t; performing feature fusion of the candidate shared user's second historical neighbor information in the receiving dimension and the second historical sharing representation information to obtain the second sharing representation information of the candidate shared user in the receiving dimension at time t; performing feature splicing of the candidate shared user's first sharing representation information in the sharing dimension and the second sharing representation information in the receiving dimension at time t to obtain the sharing representation information of the candidate shared user at time t.

[0072] Furthermore, in an optional embodiment, for the first sharing representation information of the candidate shared user at time t in the sharing dimension, the first historical neighbor information of the candidate shared user at the sharing dimension can be directly fused with the first historical sharing representation information to obtain the first sharing representation information of the candidate shared user at time t in the sharing dimension. Correspondingly, for the second sharing representation information of the candidate shared user at time t in the receiving dimension, the second historical neighbor information of the candidate shared user at the receiving dimension can be directly fused with the second historical sharing representation information to obtain the second sharing representation information of the candidate shared user at time t in the receiving dimension.

[0073] Further optionally, considering that a user may not participate in sharing behavior for a long time, or may participate in sharing behavior frequently, and because different users have different perception abilities of different time intervals, the user's historical sharing representation information can be personalized projected through a time point process model, and the latest sharing representation information can be generated based on the projection result. Based on this, in another optional embodiment, for the first sharing representation information of the candidate shared user on the sharing dimension at time t, the first historical sharing representation information of the candidate shared user on the sharing dimension can also be projected according to the time point process model of the candidate shared user to obtain the first projected sharing representation information of the candidate shared user on the sharing dimension; the first historical neighbor information, the first historical sharing representation information, and the first projected sharing representation information of the candidate shared user on the sharing dimension are sent to the MLP network for feature fusion to obtain the first sharing representation information of the candidate shared user on the sharing dimension at time t. Correspondingly, for the second sharing representation information of the candidate shared user on the receiving dimension at time t, the second sharing representation information of the candidate shared user on the receiving dimension can also be projected according to the time point process model of the candidate shared user to obtain the second sharing representation information of the candidate shared user on the receiving dimension; the second historical neighbor information, the second historical sharing representation information and the second projected sharing representation information of the candidate shared user on the receiving dimension are sent to the MLP network for feature fusion to obtain the second sharing representation information of the candidate shared user on the receiving dimension at time t.

[0074] Further optionally, as an example, based on the time point process model of the candidate shared user, the first historical sharing representation information of the candidate shared user on the sharing dimension is projected to obtain the first projected sharing representation information of the candidate shared user on the sharing dimension. The next expected sharing moment of the candidate shared user can be predicted based on the time point process model of the candidate shared user; the projection factor is determined based on the difference between the next expected sharing moment and the moment t; based on the projection factor, the first historical sharing representation information of the candidate shared user on the sharing dimension is projected to obtain the first projected sharing representation information of the candidate shared user on the sharing dimension. Accordingly, when the second sharing representation information of the candidate shared user on the receiving dimension is projected based on the time point process model of the candidate shared user, and the second sharing representation information of the candidate shared user on the receiving dimension is obtained, the next expected sharing moment of the candidate shared user can be predicted based on the time point process model of the candidate shared user; the projection factor is determined based on the difference between the next expected sharing moment and the moment t; and the second historical sharing representation information of the candidate shared user on the receiving dimension is projected based on the projection factor to obtain the second projected sharing representation information of the candidate shared user on the receiving dimension. It should be noted that the "predicting the next expected sharing moment" and "determining the projection factor" in the process of obtaining the first projected sharing representation information can be the same process as the "predicting the next expected sharing moment" and "determining the projection factor" in the process of obtaining the second projected sharing representation information, and are executed once.

[0075] Continued from Figure 3 In the embodiment shown, user v is taken as a candidate shared user, and the historical sharing representation information h of user v is v (t-) is divided into the first historical sharing representation information h of user v in the sharing dimension v a (t-) (also denoted as Va(t-)) and the second history sharing representation information h in the receiving dimension v b (t-)((also denoted as Vb(t-))), similarly, user v’s shared representation information h at time t v (t) is divided into the first sharing representation information h of user v in the sharing dimension at time t v a (t) (also denoted as Va(t)) and the second shared representation information h of user v on the receiving dimension at time t v b (t)((also denoted as Vb(t))).

[0076] Based on the time point process model, the first historical sharing representation information h of user v in the sharing dimension is v a(t-) (i.e. Va(t-)) is projected to obtain the first projection sharing representation information Va(t) of user v at time t on the sharing dimension. Assume that the projection factor is denoted by σ; the difference between Va(t-) and Va(t) is denoted by Va(t-)-Va(t); the difference between the next expected sharing time and time t is denoted by t + -t, based on the time difference, the projection factor can be calculated, for example, σ=(t + -t) / t, accordingly, Va(t-)-Va(t) / Va(t-)=σ, based on Va(t-)-Va(t) / Va(t-)=σ, the projection Va(t) of Va(t-) can be calculated. Then, the first historical neighbor information of user v in the sharing dimension and the first historical sharing representation information h v a (t-) and the first projection sharing representation information Va(t) are sent to the MLP network for feature fusion to obtain the first sharing representation information h of user vt at the sharing dimension v a It should be noted that, in addition to being calculated according to the above formula, the projection factor can also be obtained by other methods, for example, in the following embodiments, it can also be obtained by model training.

[0077] Similarly, based on the time point process model, the second historical sharing representation information h of user v in the sharing dimension is v b (t-) (i.e. Vb(t-)) is projected to obtain the second projection sharing representation information Vb(t) of user v at time t on the sharing dimension. Assume that the projection factor is denoted as σ; the difference between Vb(t-) and Vb(t) is denoted as Vb(t-)-Vb(t); the difference between the next expected sharing time and time t is denoted as t + -t, based on the time difference, the projection factor can be calculated, for example, σ=(t + -t) / t, accordingly, Vb(t-)-Vb(t) / Vb(t-)=σ, based on Vb(t-)-Vb(t) / Vb(t-)=σ, the projection Vb(t) of Vb(t-) can be calculated. Then, the second historical neighbor information of user v in the sharing dimension and the second historical sharing representation information h v b (t-) and the second projection sharing representation information Vb(t) are sent to the MLP network for feature fusion to obtain the second sharing representation information h of user vt at the sharing dimension v b (t).

[0078] In the embodiment of the present application, the time point process model is recorded as N(t), which records how many sharing events occurred before time t. The event of receiving sharing behavior corresponds to one time point process, and the event of initiating sharing behavior corresponds to another time point process. For N(t), a conditional intensity function is used to characterize it:

[0079]

[0080] Among them, λ(t): represents the conditional intensity function; H(t): represents the historical information before time t; dt: represents the differential of time t; dN(t): represents the differential of N(t); P{} represents the probability of a sharing event occurring under the corresponding conditions, and E[] represents the expected value of a sharing event under the corresponding conditions.

[0081] At a given time t, the probability S(t) that no event occurs between [0, t) is:

[0082]

[0083] And the conditional probability of the event occurring at time t is:

[0084]

[0085] Given a time t, the probability p(t|t0) of an event occurring at time t within [t0,t] is:

[0086]

[0087] Assume that λ(t) = αt, where α is a given known parameter. For the target sharing user or candidate shared user, assume that the event of the user initiating the sharing behavior and the event of receiving the sharing behavior are two different time-series processes. The time-series process can be designed based on the user's sharing representation information at time t (the sharing representation information can be an embedding vector) to achieve user personalization and user personalization at different times.

[0088] The event of initiating sharing behavior is expressed using the following formula (5):

[0089]

[0090] Where t′: the time of last sharing (the time when the last event occurred);

[0091] The sharing representation information of user u in the sharing direction at time t′;

[0092] W invite: It is the model parameter of the time series process model in the sharing direction. The model parameter can be learned during the model training process and used directly during the inference process.

[0093] The following formula (6) is used to express the event of receiving sharing behavior:

[0094]

[0095] Where t′: the time of last reception (the time when the last event occurred);

[0096] The shared representation information of user u in the receiving direction at time t′;

[0097] W vote : It is the model parameter of the time point process model in the receiving direction. The model parameter can be learned during the model training process and used directly in the inference process.

[0098] According to formula (7), the expected time of user u's next sharing behavior is calculated based on the time sequence process. And according to formula (8), the expected time of user u's next receiving behavior is calculated based on the time sequence process

[0099]

[0100]

[0101] Then, based on the expected time calculated by formula (7) or (8), the first historical sharing representation information of the target sharing user or the candidate shared user in the sharing dimension is projected; specifically, if the actual interaction time (i.e., time t) is greater than the expected time (i.e., the case of delayed sharing), the first projected sharing representation information corresponding to the first historical sharing representation information is calculated according to formula (9); if the actual time is less than the expected time (i.e., the case of early sharing), the first projected sharing representation information corresponding to the first historical sharing representation information is calculated according to formula (10).

[0102]

[0103]

[0104] in, is the sharing representation information of user u in the sharing direction at time t;

[0105] It refers to the projection factor in the sharing direction under the condition of sharing delay. This projection factor can be learned during model training and used directly during inference.

[0106] It refers to the projection shadow in the sharing direction when sharing is advanced. This projection factor can be learned during model training and used directly during inference.

[0107] It refers to the sharing representation information of user u in the sharing direction at time t′.

[0108] Specifically, if the actual interaction time (i.e., time t) is greater than the expected time (i.e., the case of delayed sharing), the second projected sharing representation information corresponding to the second historical sharing representation information is calculated according to formula (11); if the actual time is less than the expected time (i.e., the case of early sharing), the second projected sharing representation information corresponding to the second historical sharing representation information is calculated according to formula (12).

[0109]

[0110]

[0111] is the shared representation information of user u in the receiving direction at time t;

[0112] It refers to the projection factor in the receiving direction under the condition of shared delay. This projection factor can be learned during model training and used directly during inference.

[0113] It refers to the projection shadow in the receiving direction under the sharing advance condition. This projection factor can be learned during model training and used directly during inference.

[0114] It refers to the shared representation information of user u in the receiving direction at time t′. It is noted that, preferably, each of the projection factors mentioned above may conform to a zero-mean Gaussian distribution, but is not limited thereto.

[0115] Furthermore, in an optional embodiment, the sharing representation information of each shared object is divided into two parts, namely static sharing representation information and dynamic sharing representation information. Among them, static sharing representation information mainly refers to some static attributes (i.e., unchanging attributes) of the shared object. Taking the shared object as a commodity as an example, it can be the name, category, price, weight, manufacturer, origin, etc. of the commodity; dynamic sharing representation information refers to some information of the shared object that will change with sharing, which can reflect the popularity of the shared object in the sharing behavior and other information. For example, it can reflect which users the shared object tends to be shared with and which users tend to receive it. Based on this, for the sharing representation information of the target object at time t, static sharing representation information and historical dynamic sharing representation information can be extracted from the historical sharing representation information of the target object; the historical neighbor information and the historical dynamic sharing representation information of the target object are feature fused to obtain the dynamic representation information of the target object at time t; the static sharing representation information and the dynamic sharing representation information of the target object at time t are feature spliced ​​to obtain the sharing representation information of the target object at time t. Continued from Figure 3 In the embodiment shown, the target object is object i, and the historical sharing representation information h of object i is i (t-) is divided into static shared representation information h of object i i ～ Share representation information with historical dynamics h i - (t-); the historical neighbor information and historical dynamic sharing representation information h of object i i - (t-) is sent to the MLP network for feature fusion to obtain the dynamic representation information h of object i at time t i - (t); Then, the static shared representation information h of object i i ～ and the dynamic representation information h at time t i - (t) Perform feature splicing to obtain the shared representation information h of the target object at time t i (t).

[0116] After obtaining the sharing representation information of the target sharing user, the candidate shared user, and the target object at time t, the sharing representation information of the target sharing user, the candidate shared user, and the target object at time t can be directly spliced ​​together, and the MLP network can be used to predict the spliced ​​sharing representation information to obtain the sharing probability of the target sharing user sharing the target object with the candidate shared user at time t. Figure 3As shown in the figure, when predicting the sharing probability, the triplet translation method can also be used to reflect the asymmetry of sharing behavior. Specifically, the sharing representation information of the target sharing user at time t and the sharing representation information of the candidate shared user at time t can be feature-added to obtain the user-object binary representation information; then, the user-object binary representation information and the sharing representation information of the candidate shared user at time t are feature-concatenated to obtain the user-object-user ternary representation information; the MLP network is used to perform probability prediction processing on the user-object-user ternary representation information to obtain the sharing probability P(u,i,v,t) that the target sharing user shares the target object with the candidate shared user at time t. By considering the asymmetry of sharing behavior, it is beneficial to improve the prediction accuracy of the sharing probability.

[0117] In this embodiment, the number of candidate shared users is one or more, preferably multiple. In the case of multiple candidate shared users, for each candidate shared user, after obtaining the sharing probability that the target sharing user will share the target object with the candidate shared user at time t, several candidate shared users with the highest sharing probability can be selected from at least one candidate shared user as target shared users, or several candidate shared users with a sharing probability greater than a preset sharing probability threshold can be selected as target shared users, and the information of the target shared users can be pushed to the target sharing user, for example, the information of the target shared user can be pushed to the target sharing user in the form of a list. The target sharing user selects one or more target shared users from the list and shares the target object with the selected one or more target shared users.

[0118] The technical solution provided by the embodiment of the present application innovatively models the asymmetric sharing interaction relationship between users, and can simultaneously couple time information and asymmetric sharing relationships, respectively aggregating the historical neighbor user information of the sharing user and the candidate shared user in both the sharing and receiving directions, to generate the sharing representation information of the corresponding user at that time, as well as the sharing representation information of the shared object. Then, based on these sharing representation information, the sharing probability of the sharing user sharing the target object to the candidate shared user at the sharing moment is predicted, and finally, based on the sharing probability, the target shared user is locked and information is shared. In this way, the characteristics of the user's sharing preference changing over time are fully considered, and the user's current sharing willingness and receiving willingness are taken into account, and the shared user who is more interested in the target object to be shared is automatically recommended to the sharing user, which is conducive to shortening the time of searching for the shared user, reducing the tediousness of information sharing, and improving the efficiency of information sharing and user experience.

[0119] Furthermore, a temporal graph attention network based on the CTDG of the sharing dimension and the CTDG of the receiving dimension aggregates the user's bidirectional (i.e., receiving and sharing) historical neighbor information, making it possible to express the user's sharing representation information as a sharing tendency representation and a receiving tendency representation. Furthermore, the sharing representation information of the target object is added to the sharing representation information of the sharing user, making the sharing representation information of the sharing user and the sharing representation information of the candidate shared users asymmetric, reflecting the asymmetry of sharing behavior. Furthermore, the asymmetry of sharing behavior is combined with the continuous time dynamic graph to provide instant and effective sharing recommendations.

[0120] In order to better understand the technical solutions provided by the embodiments of this application, Figure 3 The application scenario diagram shown is used to illustrate. Assume that the target sharing user is recorded as user u, any candidate sharing user of the target sharing user is recorded as user v, and the target object that the target sharing user needs to share at time t is recorded as object i. Before time t, the historical sharing representation information of user u is recorded as h u (t-), where h u (t-) included in the sharing dimension h u a (t-) and h in the receiving dimension u b (t-). If at time t, user u initiates the sharing behavior of sharing object i, obtain user u's first CTDG in the sharing dimension and the second CTDG in the receiving dimension; node Ua in user u's first CTDG is the node corresponding to user u as the sharing user, and nodes v1, v2, and v3 are the nodes corresponding to the shared users of user u. The directed edges of the first CTDG point from node Ua to nodes v1, v2, and v3, indicating that the sharing behavior is initiated by user u and received by the shared users corresponding to nodes v1, v2, and v3. Node Ub in user u's second CTDG is the node corresponding to user u as the shared user, and nodes v4, v5, and v6 are the nodes corresponding to the shared users of user u. The directed edges of the second CTDG point to node Ub, indicating that the sharing behavior is initiated by the users corresponding to nodes v4, v5, and v6 and received by user u.

[0121] For the first shared representation information of user u at time t in the sharing dimension, the first CTDG of user u is input into the temporal attention network for feature learning to obtain the first historical neighbor information of user u in the sharing dimension; and the h in the historical shared representation information of user u is converted into u a(t-) (also denoted as Ua(t-)) is projected to obtain the first projected shared representation information Ua(t) of user u in the sharing dimension at time t; Ua(t-), Ua(t) and the first historical neighbor information of user u in the sharing dimension are input into MLP (Multilayer Perceptron) for feature fusion processing to obtain the first shared representation information of user u in the sharing dimension at time t.

[0122] For the second shared representation information of user u at time t on the receiving dimension, the second CTDG of user u is input into the temporal attention network for feature learning to obtain the second historical neighbor information of user u on the receiving dimension; and the historical shared representation information h of user u is converted into u h in (t-) u b (t-) (also denoted as Ub(t-)) is projected to obtain the second projection shared representation information Ub(t) of user u at time t on the receiving dimension; Ub(t-), Ub(t) and the second historical neighbor information of user u on the receiving dimension are input into MLP (Multilayer Perceptron) for feature fusion processing to obtain the second shared representation information of user u at time t on the receiving dimension.

[0123] After obtaining the first sharing representation information of user u at time t in the sharing dimension and the second sharing representation information of user u at time t in the receiving dimension, feature splicing is performed on the first sharing representation information of user u at time t in the sharing dimension and the second sharing representation information of user u at time t in the receiving dimension to obtain the sharing representation information h of user u at time t u (t).

[0124] Before time t, the historical sharing representation information of user v is recorded as h v (t-), where h v (t-) included in the sharing dimension h v a (t-) and h in the receiving dimension v b(t-). If at time t, user u initiates the sharing behavior of sharing object i, obtain user v's first CTDG in the sharing dimension and the second CTDG in the receiving dimension; node va in user v's first CTDG is the node corresponding to user v as the sharing user, and nodes u1, u2, and u3 are the nodes corresponding to the users shared with as user v. The directed edges in the first CTDG point from node va to nodes u1, u2, and u3, indicating that the sharing behavior was initiated by user v and received by the users shared with corresponding to nodes u1, u2, and u3. Node vb in user v's second CTDG is the node corresponding to user v as the shared user, and nodes u4, u5, and u6 are the nodes corresponding to the users shared with as user v. The directed edges in the second CTDG point to node vb, indicating that the sharing behavior was initiated by the users corresponding to nodes u4, u5, and u6 and received by user v.

[0125] For the first shared representation information of user v at time t in the sharing dimension, the first CTDG of user v is input into the temporal attention network for feature learning to obtain the first historical neighbor information of user v in the sharing dimension; and h in the historical shared representation information of user v is converted into v a (t-) (also denoted as va(t-)) is projected to obtain the first projected sharing representation information va(t) of user v in the sharing dimension at time t; va(t-), va(t) and the first historical neighbor information of user v in the sharing dimension are input into MLP for feature fusion processing to obtain the first sharing representation information of user v in the sharing dimension at time t.

[0126] For the second shared representation information of user v at time t on the receiving dimension, the second CTDG of user v is input into the temporal attention network for feature learning to obtain the second historical neighbor information of user v on the receiving dimension; and the historical shared representation information h of user v is converted into v h in (t-) v b (t-) (also denoted as Vb(t-)) is projected to obtain the second projected shared representation information Vb(t) of user v at time t on the receiving dimension; Vb(t-), Vb(t) and the second historical neighbor information of user v on the receiving dimension are input into MLP for feature fusion processing to obtain the second shared representation information of user v at time t on the receiving dimension.

[0127] After obtaining the first sharing representation information of user v at time t in the sharing dimension and the second sharing representation information of user v at time t in the receiving dimension, feature splicing is performed on the first sharing representation information of user v at time t in the sharing dimension and the second sharing representation information of user v at time t in the receiving dimension to obtain the sharing representation information h of user v at time t v (t).

[0128] Before time t, the historical shared representation information of object i is recorded as h i (t-), h i (t-) includes static shared representation information h i ～ Share representation information with historical dynamics h i - (t-). If at time t, user u initiates the sharing of object i, obtain the third CTDG of object i. The node id in the third CTDG of object i is the node corresponding to object i. The user at node u1 shared object i with the user at node v1 before time t, so there is a directed edge from node u1 to node id, and a directed edge from node id to node v1. The user at node u2 shared object i with the user at node v2 before time t, so there is a directed edge from node u2 to node id, and a directed edge from node id to node v2.

[0129] Input the third CTDG of object i into the temporal attention network for feature learning to obtain the historical neighbor information of object i; and the historical shared representation information h of object i i The historical dynamic sharing representation information h in (t-) i - The historical neighbor information of object i is input into MLP for feature fusion processing to obtain the dynamic shared representation information of object i at time t; the historical shared representation information h of object i is converted into i The static shared representation information h in (t-) i ～ The shared representation information of object i at time t is concatenated with the dynamic shared representation information of object i at time t to obtain the shared representation information h of object i at time t i (t).

[0130] Continue to see Figure 3 , after obtaining the shared representation information h of user u at time t u (t), user v’s shared representation information h at time t v (t) and the shared representation information h of object i at time t i After (t), the sharing probability prediction operation is performed. First, the sharing representation information h of user u at time t is u(t) and the shared representation information h of object i at time t i (t) Perform feature addition operation to obtain the binary representation information of user u-object i. It is important to only use the shared representation information h of user u at time t u (t) and the shared representation information h of object i at time t i (t) performs feature addition operation without adding the shared representation information h of user v at time t v (t) and the shared representation information h of object i at time t i (t) Perform feature addition operation to make the binary representation information of user u-object i and the shared representation information h of user v at time t v (t) is asymmetric, thus reflecting the asymmetry of sharing behavior; then the user u-object i binary representation information and the user v's sharing representation information h at time t are combined v (t) Perform feature concatenation to obtain the ternary representation information of user u, object i, and user v. This ternary representation information of user u, object i, and user v is input into the MLP for prediction processing to obtain the sharing probability P(u, i, v, t) of user u sharing object i with user v at time t. Similarly, the sharing probability of user u sharing object i with any user v at time t can be obtained. After obtaining the sharing probability of user u sharing object i with each of multiple users v at time t, the user v with the highest sharing probability is selected as the recommended shared user, and the selected recommended shared users are pushed to user u, allowing user u to select several recommended shared users. The access entry link of object i is also pushed to the selected recommended shared users, and so on.

[0131] In this embodiment, a continuous-time dynamic graph is used to model dynamic asymmetric sharing interaction relationships, thereby simultaneously coupling fine-grained temporal information and asymmetric topological structure information. A time-series graph attention network based on a bidirectional continuous-time dynamic graph is used to aggregate neighboring user information in both the sharing and receiving directions, respectively, to characterize the user's sharing tendency and receiving tendency. A personalized projection factor is designed based on the time-series point process to project the user representation, which can reflect the perception of different users for different time intervals. Therefore, a new user representation can be obtained at any time using the projection, and the problem of delayed update of user representation after a long period of non-interaction is solved in a personalized way. Thus, a sharing recommendation method based on a bidirectional continuous-time dynamic graph with time interval perception is provided, which can support the real-time dynamic recommendation and ranking optimization of the user's (sharer) friends (shared with) when sharing products, thereby improving the user's real-time product sharing experience. On this basis, a set of dynamic user representations is provided, which can simultaneously reflect the user's immediate sharing willingness and receiving willingness, and can be widely used in downstream tasks such as user recommendation and marketing decision-making related to social interaction scenarios.

[0132] Figure 4 This is a flow chart of another information sharing method provided in an embodiment of the present application. Figure 4 As shown, the method may include the following steps:

[0133] 401. Respond to the information sharing request initiated by the target sharing user for the target object at time t, and aggregate the first historical neighbor information of the target sharing user and the candidate shared users in the sharing dimension and the second historical neighbor information in the receiving dimension based on the historical sharing behavior data, as well as the historical neighbor information of the target object.

[0134] 402. Generate sharing representation information of the target sharing user and the candidate shared user at time t based on their respective first historical neighbor information in the sharing dimension and second historical neighbor information in the receiving dimension, and generate sharing representation information of the target object at time t based on the historical neighbor information of the target object.

[0135] 403. Predict the sharing probability of the target sharing user sharing the target object with the candidate sharing users at time t based on the sharing representation information of the target sharing user, the candidate sharing users, and the target object at time t.

[0136] 404. Provide information of a target user among candidate users to the target sharing user based on the sharing probability, so that the target sharing user shares the target object with the target user.

[0137] 405. After the target sharing user shares the target object with the target shared user, the sharing representation information of the target sharing user, the candidate shared user, and the target object at time t is updated based on the recurrent neural network to make the next sharing prediction based on the updated sharing representation information.

[0138] For steps 401 to 404 in this embodiment, reference may be made to the relevant descriptions of steps 201 to 204 in the aforementioned embodiment, and will not be repeated here.

[0139] In this embodiment, the recurrent neural network (RNN) includes, but is not limited to, GRU.

[0140] In this embodiment, a first GRU, a second GRU, and a third GRU may be provided. The first GRU is used to update the explicit representation, the second GRU is used to update the implicit representation, and the third GRU is used to update the dynamic shared representation information of the object.

[0141] The sharing representation information of the target sharing user and the candidate shared user at time t respectively includes the first sharing representation information in the sharing dimension at time t and the second sharing representation information in the receiving dimension at time t. The first sharing representation information of the target sharing user in the sharing dimension at time t and the second sharing representation information of the candidate shared user in the receiving dimension at time t are explicit representations; the second sharing representation information of the target sharing user in the receiving dimension at time t and the first sharing representation information of the candidate shared user in the sharing dimension at time t are implicit representations. Therefore, the first sharing representation information of the target sharing user in the sharing dimension at time t and the second sharing representation information of the candidate shared user in the receiving dimension at time t are respectively input into the first GRU for updating the explicit representation; the second sharing representation information of the target sharing user in the receiving dimension at time t and the first sharing representation information of the candidate shared user in the sharing dimension at time t are respectively input into the second GRU for updating the implicit representation; the updated first sharing representation information in the sharing dimension and the updated second sharing representation information in the receiving dimension of the target sharing user and the candidate shared user are concatenated to obtain the updated sharing representation information of the target sharing user and the candidate shared user.

[0142] The target object's shared representation information at time t includes static shared representation information and dynamic shared representation information at time t. Therefore, the target object's dynamic shared representation information at time t is fed into the third GRU for update. The static shared representation information and the updated dynamic shared representation information of the target object are concatenated to obtain the updated shared representation information of the target object.

[0143] Continue to see Figure 3 , the first GRU in Figure 3 The dominant GRU is shown in Figure 3 The third GRU is shown in Figure 3 The object GRU is shown in .

[0144] Assume that the first sharing representation information of user u at time t in the sharing dimension is recorded as h u a (t), the second shared representation information h of user u at time t on the receiving dimension u b (t); h u a (t) Input to the explicit GRU for update, and get h u a (t+); and h u b (t) Input to the implicit GRU for update to get h u b (t+), h u a (t+) and h u b (t+) performs feature splicing to obtain the updated h u (t+).

[0145] Assume that the first shared representation information of user v at time t in the sharing dimension is recorded as h v a (t), the second shared representation information h of user v at time t on the receiving dimension v b (t); h v b (t) Input to the explicit GRU for update, and get h v b (t+); and h v a (t) Input to the implicit GRU for update to get h v a (t+), h v a (t+) and h v b (t+) performs feature splicing to obtain the updated h v (t+).

[0146] Assume that the shared representation information h of object i at time t is i The static shared representation information in (t) is denoted as h i ～ , the shared representation information h of object i at time ti The dynamic sharing representation information in (t) is denoted as h i - (t); h i - (t) Input to the object GRU for update, and get h i - (t+); and h i ～ and h i - (t+) performs feature splicing to obtain the updated h i (t+).

[0147] The technical solution provided in the embodiment of the present application automatically recommends to the sharing user the users who are more interested in the target object to be shared, thereby reducing the tediousness of information sharing and improving the efficiency of information sharing. Furthermore, it also supports updating the sharing representation information of users and objects to help improve the accuracy of sharing recommendations. In order to reflect the effect of the embodiment of the present application, the technical solution of the embodiment of the present application was compared with the traditional solution through testing. The technical solution provided in the embodiment of the present application has a significant improvement in multiple performance indicators such as accuracy (ACC), precision (also known as P precision), recall (also known as R recall), F1-score, AUC (Area Under Curve, area under the ROC curve) value, ROC (receiver operating characteristic curve, receiver operating characteristic curve) and PR curve in sharing recommendations. Among them, P in the PR curve represents precision, and R represents recall, which represents the relationship between precision and recall. F1 Score, also known as F1 score, is an indicator used in statistics to measure the accuracy of a binary classification model.

[0148] Figure 5 This is a structural diagram of an information sharing processing device provided in an embodiment of the present application. Figure 5 , the apparatus may include:

[0149] Neighbor information acquisition module 51 is used to respond to an information sharing request initiated by a target sharing user for a target object at time t, and aggregate the first historical neighbor information of the target sharing user and candidate shared users in the sharing dimension and the second historical neighbor information in the receiving dimension based on historical sharing behavior data, as well as aggregate the historical neighbor information of the target object;

[0150] a representation information generation module 52 for generating sharing representation information of the target sharing user and the candidate shared user at time t based on their respective first historical neighbor information in the sharing dimension and second historical neighbor information in the receiving dimension, and for generating sharing representation information of the target object at time t based on the target object's historical neighbor information;

[0151] The sharing probability prediction module 53 is used to predict the sharing probability of the target sharing user sharing the target object with the candidate sharing users at time t based on the sharing representation information of the target sharing user, the candidate sharing users, and the target object at time t;

[0152] The user information recommendation module 54 is configured to provide the target user's information among the candidate users to the target sharing user according to the sharing probability, so that the target sharing user can share the target object with the target user.

[0153] Further optionally, the neighbor information acquisition module 51 aggregates the first historical neighbor information of the target sharing user and the candidate shared user in the sharing dimension and the second historical neighbor information in the receiving dimension, and aggregates the historical neighbor information of the target object based on the historical sharing behavior data, specifically for:

[0154] According to the historical sharing behavior data, a temporal attention network based on continuous time dynamic graph (CTDG) is used to aggregate the first historical neighbor information of the target sharing user and the candidate shared users in the sharing dimension and the second historical neighbor information in the receiving dimension, as well as the historical neighbor information of the target object.

[0155] Further optionally, the neighbor information acquisition module 51 uses a temporal attention network based on a continuous-time dynamic graph CTDG to aggregate the first historical neighbor information of the target sharing user and the candidate shared user in the sharing dimension and the second historical neighbor information in the receiving dimension, and aggregates the historical neighbor information of the target object according to the historical sharing behavior data. It is specifically used to: obtain the first CTDG of the target sharing user and the candidate shared user in the sharing dimension and the second CTDG in the receiving dimension, as well as the third CTDG of the target object according to the historical sharing behavior data; input the first CTDG of the target sharing user and the candidate shared user in the sharing dimension and the second CTDG in the receiving dimension into the temporal attention network for feature learning, and obtain the first historical neighbor information of the target sharing user and the candidate shared user in the sharing dimension and the second historical neighbor information in the receiving dimension; input the third CTDG of the target object into the temporal attention network for feature learning, and obtain the historical neighbor information of the target object.

[0156] Further optionally, when the neighbor information acquisition module 51 obtains the first CTDG of the target sharing user and the candidate shared user in the sharing dimension and the second CTDG in the receiving dimension based on the historical sharing behavior data, it is specifically used to: determine the sharing user, shared user, sharing object information and sharing time information involved in each historical sharing behavior from the historical sharing behavior data; for each historical sharing behavior, abstract the sharing user and the shared user into user nodes, create a directed edge from the sharing user to the shared user between the user nodes, and use the sharing object information and sharing time information as attribute information of the directed edge to obtain the global CTDG; take the target sharing user and the candidate shared user as query objects respectively, obtain the local CTDG when the query object is the sharing user from the global CTDG as its first CTDG in the sharing dimension, and obtain the local CTDG when the query object is the shared user as its second CTDG in the receiving dimension.

[0157] Further optionally, when the neighbor information acquisition module 51 obtains the third CTDG of the target object based on the historical sharing behavior data, it is specifically used to: determine the sharing user, shared user and shared object involved in each historical sharing behavior from the historical sharing behavior data; for each historical sharing behavior, abstract the sharing user and shared user into user nodes, abstract the sharing object into object nodes, create a directed edge from the user node corresponding to the sharing user to the object node, and a directed edge from the object node to the user node corresponding to the shared user, so as to obtain the third CTDG of the target object.

[0158] Further optionally, when the representation information generation module 52 generates the sharing representation information of the target sharing user and the candidate shared user at time t based on the first historical neighbor information of the target sharing user and the candidate shared user in the sharing dimension and the second historical neighbor information of the target sharing user and the candidate shared user in the receiving dimension, it is specifically used to: generate the sharing representation information of the target sharing user and the candidate shared user at time t based on the first historical neighbor information of the target sharing user and the candidate shared user in the sharing dimension and the second historical neighbor information of the target sharing user and the candidate shared user in the receiving dimension, as well as their respective historical sharing representation information.

[0159] Further optionally, when the representation information generation module 52 generates the sharing representation information of the target sharing user at time t based on the target sharing user's first historical neighbor information in the sharing dimension and the second historical neighbor information in the receiving dimension, as well as its historical sharing representation information, it is specifically used to: extract the target sharing user's first historical sharing representation information in the sharing dimension and the second historical sharing representation information in the receiving dimension from the historical sharing representation information of the target sharing user; perform feature fusion of the target sharing user's first historical neighbor information in the sharing dimension with the first historical sharing representation information to obtain the first sharing representation information of the target sharing user in the sharing dimension at time t; perform feature fusion of the target sharing user's second historical neighbor information in the receiving dimension with the second historical sharing representation information to obtain the second sharing representation information of the target sharing user in the receiving dimension at time t; perform feature splicing of the target sharing user's first sharing representation information in the sharing dimension and the second sharing representation information in the receiving dimension at time t to obtain the sharing representation information of the target sharing user at time t.

[0160] Further optionally, the representation information generation module 52 performs feature fusion on the first historical neighbor information and the first historical sharing representation information of the target sharing user in the sharing dimension to obtain the first sharing representation information of the target sharing user at time t in the sharing dimension, which is specifically used for: projecting the first historical sharing representation information of the target sharing user in the sharing dimension according to the time point process model of the target sharing user to obtain the first projected sharing representation information of the target sharing user in the sharing dimension; sending the first historical neighbor information, the first historical sharing representation information and the first projected sharing representation information of the target sharing user in the sharing dimension into the multi-layer perceptron MLP network for feature fusion to obtain the first sharing representation information of the target sharing user in the sharing dimension at time t.

[0161] Further optionally, the representation information generation module 52 projects the first historical sharing representation information of the target sharing user on the sharing dimension according to the time point process model of the target sharing user, and obtains the first projected sharing representation information of the target sharing user on the sharing dimension. It is specifically used to: predict the next expected sharing moment of the target sharing user according to the time point process model of the target sharing user; determine the projection factor according to the difference between the next expected sharing moment and the moment t; and project the first historical sharing representation information of the target sharing user on the sharing dimension according to the projection factor to obtain the first projected sharing representation information of the target sharing user on the sharing dimension.

[0162] Further optionally, when the representation information generation module 52 generates the sharing representation information of the target object at time t based on the historical neighbor information of the target object, it is specifically used to: extract static sharing representation information and historical dynamic sharing representation information from the historical sharing representation information of the target object; perform feature fusion of the historical neighbor information and the historical dynamic sharing representation information of the target object to obtain the dynamic representation information of the target object at time t; perform feature splicing of the static sharing representation information and the dynamic sharing representation information of the target object at time t to obtain the sharing representation information of the target object at time t.

[0163] Further optionally, the sharing probability prediction module 53 predicts the sharing probability that the target sharing user shares the target object with the candidate shared user at time t based on the sharing representation information of the target sharing user, the candidate shared user and the target object at time t. It is specifically used to: add the sharing representation information of the target sharing user at time t and the sharing representation information of the candidate shared user at time t to obtain user-object binary representation information; perform feature splicing on the user-object binary representation information and the sharing representation information of the candidate shared user at time t to obtain user-object-user ternary representation information; use a multi-layer perceptron MLP to perform probability prediction processing on the user-object-user ternary representation information to obtain the sharing probability that the target sharing user shares the target object with the candidate shared user at time t.

[0164] Further optionally, the above-mentioned device also includes: an update module, which is used to update the sharing representation information of the target sharing user, the candidate shared user and the target object at time t based on a recurrent neural network after the target sharing user shares the target object with the target shared user, so as to make the next sharing prediction based on the updated sharing representation information.

[0165] Further optionally, the sharing representation information of the target sharing user and the candidate shared user at time t respectively includes first sharing representation information in the sharing dimension at time t and second sharing representation information in the receiving dimension at time t;

[0166] The update module updates the sharing representation information of the target sharing user and the candidate shared user at time t based on the recurrent neural network, specifically for:

[0167] Input the first sharing representation information of the target sharing user at time t in the sharing dimension and the second sharing representation information of the candidate shared user at time t in the receiving dimension into the first gated neural network GRU to update the explicit representation;

[0168] Input the second sharing representation information of the target sharing user at time t in the receiving dimension and the first sharing representation information of the candidate shared user at time t in the sharing dimension into the second GRU respectively to update the implicit representation;

[0169] The first sharing representation information updated in the sharing dimension and the second sharing representation information updated in the receiving dimension of the target sharing user and the candidate shared user are concatenated to obtain the updated sharing representation information of the target sharing user and the candidate shared user.

[0170] Further optionally, the sharing representation information of each target object at time t includes static sharing representation information and dynamic sharing representation information at time t; when the update module updates the sharing representation information of the target object at time t based on the recurrent neural network, it is specifically used to: input the dynamic sharing representation information of the target object at time t into the third GRU for updating, and splice the static sharing representation information and the updated dynamic sharing representation information of the target object to obtain the updated sharing representation information of the target object.

[0171] Figure 5 The device shown can perform Figure 2 The method shown in the embodiment shown, its implementation principle and technical effect are not described in detail. Figure 5 The specific manner in which each module and unit performs operations in the device shown has been described in detail in the embodiment of the method and will not be elaborated here.

[0172] It should be noted that the execution entity of each step of the method provided in the above embodiment can be the same device, or the method can be executed by different devices. For example, the execution entity of steps 401 to 405 can be device A; for another example, the execution entity of steps 401 and 403 can be device A, and the execution entity of steps 404 and 405 can be device B; and so on.

[0173] In addition, in some of the processes described in the above embodiments and the accompanying drawings, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this article or may be executed in parallel. The sequence numbers of the operations, such as 401, 402, etc., are only used to distinguish between different operations, and the sequence numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to being different types.

[0174] Figure 6This is a schematic diagram of the structure of a computer device provided in an embodiment of the present application. Figure 6 As shown, the computer device includes a memory 61 and a processor 62. The memory 61 is used to store computer programs and can be configured to store various other data to support operations on the computing platform. Examples of such data include instructions for any application or method operating on the computing platform, contact data, phone book data, messages, pictures, videos, etc.

[0175] The memory 61 can be implemented by any type of volatile or non-volatile memory device or a combination thereof, such as static random-access memory (SRAM), electrically erasable programmable read only memory (EEPROM), erasable programmable read only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0176] The processor 62 is coupled to the memory 61 and is used to execute the computer program in the memory 61, so as to: respond to the information sharing request initiated by the target sharing user for the target object at time t, and aggregate the first historical neighbor information of the target sharing user and the candidate shared user in the sharing dimension and the second historical neighbor information in the receiving dimension, as well as the historical neighbor information of the target object according to the historical sharing behavior data; generate the sharing representation information of the target sharing user and the candidate shared user at time t according to the first historical neighbor information of the target sharing user and the candidate shared user in the sharing dimension and the second historical neighbor information of the candidate shared user, and generate the sharing representation information of the target object at time t according to the historical neighbor information of the target object; predict the sharing probability of the target sharing user sharing the target object with the candidate shared user at time t according to the sharing representation information of the target sharing user, the candidate shared user and the target object at time t; and provide the information of the target shared user among the candidate shared users to the target sharing user according to the sharing probability, so that the target sharing user shares the target object with the target shared user.

[0177] Further, if Figure 6 As shown, the computer device also includes: a communication component 63, a display 64, a power component 65, an audio component 66 and other components. Figure 6Only some components are shown schematically, which does not mean that the computer equipment only includes Figure 6 In addition, Figure 6 The components in the dotted box are optional components, not mandatory components, and the specific components may depend on the product form of the computer device. The computer device of this embodiment can be implemented as a terminal device such as a desktop computer, a laptop computer, a smart phone or an IOT (Internet of Things) device, or a server device such as a conventional server, a cloud server or a server array. If the computer device of this embodiment is implemented as a terminal device such as a desktop computer, a laptop computer, a smart phone, etc., it may include Figure 6 If the computer device of this embodiment is implemented as a server device such as a conventional server, a cloud server or a server array, it may not include the components in the dotted box; Figure 6 Components within the dotted box.

[0178] The detailed implementation process of the processor executing each action can be found in the relevant description in the aforementioned method embodiment or device embodiment, and will not be repeated here.

[0179] Accordingly, an embodiment of the present application further provides a computer-readable storage medium storing a computer program, which, when executed, can implement the steps that can be executed by a computer device in the above method embodiment.

[0180] Accordingly, an embodiment of the present application also provides a computer program product, including a computer program / instruction. When the computer program / instruction is executed by a processor, the processor is enabled to implement the steps that can be executed by a computer device in the above method embodiment.

[0181] The above-mentioned communication component is configured to facilitate wired or wireless communication between the device where the communication component is located and other devices. The device where the communication component is located can access a wireless network based on a communication standard, such as WiFi, 2G, 3G, 4G / LTE, 5G and other mobile communication networks, or a combination thereof. In an exemplary embodiment, the communication component receives a broadcast signal or broadcast-related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component also includes a near field communication (NFC) module to facilitate short-range communication. For example, the NFC module can be implemented based on radio frequency identification (RFID) technology, infrared data association (IrDA) technology, ultra wide band (UWB) technology, Bluetooth (BT) technology and other technologies.

[0182] The above-mentioned display includes a screen, which may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen can be implemented as a touch screen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touch, slide, and gestures on the touch panel. The touch sensor can not only sense the boundary of the touch or slide action, but also detect the duration and pressure associated with the touch or slide operation.

[0183] The power supply assembly provides power to various components of the device in which the power supply assembly is located. The power supply assembly may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to the device in which the power supply assembly is located.

[0184] The above-mentioned audio component can be configured to output and / or input audio signals. For example, the audio component includes a microphone (MIC), and when the device where the audio component is located is in an operating mode, such as call mode, recording mode, and voice recognition mode, the microphone is configured to receive external audio signals. The received audio signal can be further stored in a memory or sent via a communication component. In some embodiments, the audio component also includes a speaker for outputting audio signals.

[0185] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-readable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0186] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0187] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0188] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0189] In a typical configuration, a computing device includes one or more processors (central processing units, CPUs), input / output interfaces, network interfaces, and memory.

[0190] Memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0191] Computer-readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, Phase Change RAM (PRAM), Static Random-Access Memory (SRAM), Dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), Read Only Memory (ROM), Electrically-Erasable Programmable Read-Only Memory (EEPROM), flash memory or other memory technology, CD-ROM, Digital Versatile Disc (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission medium that can be used to store information that can be accessed by a computing device. According to the definition in this article, computer-readable media does not include transitory media such as modulated data signals and carrier waves.

[0192] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0193] The above are merely embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. An information sharing processing method, characterized in that: include: In response to an information sharing request initiated by a target sharing user for a target object at time t, based on historical sharing behavior data, the target sharing user and candidate recipient's first historical neighbor information in the sharing dimension and the second historical neighbor information in the receiving dimension are aggregated, along with the target object's historical neighbor information. The first historical neighbor information in the sharing dimension represents the relevant information regarding the sharing behavior initiated by the sharing user toward the recipient. The second historical neighbor information in the receiving dimension represents the relevant information about the receiving user receiving the sharing behavior of the sharing user; The target object's historical neighbor information represents relevant information about the sharing user initiating the sharing behavior for the target object and relevant information about the shared user receiving the sharing behavior for the target object; Generate sharing representation information of the target sharing user and the candidate shared user at time t based on their respective first historical neighbor information in the sharing dimension and second historical neighbor information in the receiving dimension, and generate sharing representation information of the target object at time t based on the historical neighbor information of the target object, wherein the sharing representation information represents the willingness to share or the willingness to receive; Predicting the probability of the target sharing user sharing the target object with the candidate sharing users at time t based on the sharing representation information of the target sharing user, the candidate sharing users, and the target object at time t; According to the sharing probability, information of a target user among the candidate users is provided to the target sharing user, so that the target sharing user shares the target object with the target user.

2. The method according to claim 1, characterized in that Based on the historical sharing behavior data, the first historical neighbor information of the target sharing user and the candidate shared user in the sharing dimension and the second historical neighbor information in the receiving dimension are aggregated, as well as the historical neighbor information of the target object, including: According to the historical sharing behavior data, a temporal attention network based on continuous time dynamic graph (CTDG) is used to aggregate the first historical neighbor information of the target sharing user and the candidate shared users in the sharing dimension and the second historical neighbor information in the receiving dimension, as well as the historical neighbor information of the target object.

3. The method according to claim 2, characterized in that According to the historical sharing behavior data, a temporal attention network based on continuous time dynamic graph (CTDG) is used to aggregate the first historical neighbor information of the target sharing user and the candidate shared user in the sharing dimension and the second historical neighbor information in the receiving dimension, as well as the historical neighbor information of the target object, including: According to the historical sharing behavior data, the first CTDG of the target sharing user and the candidate shared user in the sharing dimension and the second CTDG in the receiving dimension are obtained, as well as the third CTDG of the target object; Input the first CTDG of the target sharing user and the candidate shared user in the sharing dimension and the second CTDG in the receiving dimension into the temporal attention network for feature learning, and obtain the first historical neighbor information of the target sharing user and the candidate shared user in the sharing dimension and the second historical neighbor information of the target sharing user and the candidate shared user in the receiving dimension; The third CTDG of the target object is input into the temporal attention network for feature learning to obtain the historical neighbor information of the target object.

4. The method according to claim 3, characterized in that Based on historical sharing behavior data, the first CTDG of the target sharing user and the candidate shared user in the sharing dimension and the second CTDG in the receiving dimension are obtained, including: Determine the sharing user, shared user, sharing object information, and sharing time information involved in each historical sharing behavior from the historical sharing behavior data; For each historical sharing behavior, the sharing user and the shared user are abstracted into user nodes. A directed edge is created between the user nodes, from the sharing user to the shared user. The sharing object information and sharing time information are used as attribute information of the directed edge to obtain the global CTDG. The target sharing user and the candidate shared user are respectively used as query objects, and the local CTDG when the query object is used as the sharing user is obtained from the global CTDG as its first CTDG in the sharing dimension, and the local CTDG when the query object is used as the shared user is obtained as its second CTDG in the receiving dimension.

5. The method according to claim 3, characterized in that Based on historical sharing behavior data, obtain the target object's third CTDG, including: Determine the sharing user, shared user, and sharing object involved in each historical sharing behavior from the historical sharing behavior data; For each historical sharing behavior, the sharing user and the shared user are abstracted as user nodes, and the shared object is abstracted as object node. A directed edge is created from the user node corresponding to the sharing user to the object node, and a directed edge from the object node to the user node corresponding to the shared user, to obtain the third CTDG of the target object.

6. The method according to any one of claims 1 to 5, characterized in that Generate sharing representation information of the target sharing user and the candidate shared user at time t based on the first historical neighbor information of the target sharing user and the candidate shared user in the sharing dimension and the second historical neighbor information of the candidate shared user in the receiving dimension, including: Based on the first historical neighbor information of the target sharing user and the candidate shared user in the sharing dimension and the second historical neighbor information in the receiving dimension, as well as their respective historical sharing representation information, the sharing representation information of the target sharing user and the candidate shared user at time t is generated.

7. The method according to claim 6, characterized in that Generate the target sharing user's sharing representation information at time t based on the target sharing user's first historical neighbor information in the sharing dimension and the target sharing user's second historical neighbor information in the receiving dimension, as well as the target sharing user's historical sharing representation information, including: Extracting first historical sharing representation information in a sharing dimension and second historical sharing representation information in a receiving dimension from the historical sharing representation information of the target sharing user; Perform feature fusion on the first historical neighbor information of the target sharing user in the sharing dimension and the first historical sharing representation information to obtain the first sharing representation information of the target sharing user in the sharing dimension at time t; Perform feature fusion on the second historical neighbor information of the target sharing user in the receiving dimension and the second historical sharing representation information to obtain the second sharing representation information of the target sharing user in the receiving dimension at time t; The first sharing representation information of the target sharing user at time t in the sharing dimension and the second sharing representation information in the receiving dimension are feature spliced ​​to obtain the sharing representation information of the target sharing user at time t.

8. The method according to claim 7, characterized in that The first historical neighbor information of the target sharing user in the sharing dimension and the first historical sharing representation information are subjected to feature fusion to obtain the first sharing representation information of the target sharing user in the sharing dimension at time t, including: Projecting the first historical sharing representation information of the target sharing user on the sharing dimension according to the time point process model of the target sharing user to obtain the first projected sharing representation information of the target sharing user on the sharing dimension; The first historical neighbor information, the first historical sharing representation information and the first projected sharing representation information of the target sharing user in the sharing dimension are fed into the multi-layer perceptron MLP network for feature fusion to obtain the first sharing representation information of the target sharing user in the sharing dimension at time t.

9. The method according to claim 8, characterized in that Projecting the first historical sharing representation information of the target sharing user on the sharing dimension according to the time point process model of the target sharing user to obtain the first projected sharing representation information of the target sharing user on the sharing dimension includes: Predicting the next expected sharing moment of the target sharing user based on the time point process model of the target sharing user; Determining a projection factor according to a difference between the next expected sharing time and time t; The first historical sharing representation information of the target sharing user in the sharing dimension is projected according to the projection factor to obtain the first projected sharing representation information of the target sharing user in the sharing dimension.

10. The method according to claim 1, characterized in that Generating shared representation information of the target object at time t according to the historical neighbor information of the target object includes: Extracting static sharing representation information and historical dynamic sharing representation information from the historical sharing representation information of the target object; Perform feature fusion on the historical neighbor information of the target object and the historical dynamic sharing representation information to obtain the dynamic representation information of the target object at time t; The static sharing representation information and the dynamic sharing representation information of the target object at time t are feature-concatenated to obtain the sharing representation information of the target object at time t.

11. The method according to any one of claims 1 to 5 and 7 to 10, characterized in that Predicting the probability of the target sharing user sharing the target object with the candidate sharing users at time t based on the sharing representation information of the target sharing user, the candidate sharing users, and the target object at time t includes: Add the sharing representation information of the target sharing user at time t and the sharing representation information of the candidate shared user at time t to obtain user-object binary representation information; Perform feature concatenation on the binary representation information of the user-object and the sharing representation information of the candidate shared user at time t to obtain the user-object-user ternary representation information; A multi-layer perceptron (MLP) is used to perform probability prediction processing on the user-object-user ternary representation information to obtain the sharing probability that the target sharing user shares the target object with the candidate shared user at time t.

12. The method according to any one of claims 1 to 5 and 7 to 10, characterized in that Also includes: After the target sharing user shares the target object with the target shared user, the sharing representation information of the target sharing user, the candidate shared user and the target object at time t is updated based on the recurrent neural network to make the next sharing prediction based on the updated sharing representation information.

13. The method according to claim 12, characterized in that The sharing representation information of the target sharing user and the candidate shared user at time t respectively includes first sharing representation information in the sharing dimension at time t and second sharing representation information in the receiving dimension at time t; The sharing representation information of the target sharing user and the candidate shared user at time t is updated based on a recurrent neural network, including: Inputting the first sharing representation information of the target sharing user at time t in the sharing dimension and the second sharing representation information of the candidate shared user at time t in the receiving dimension into the first gated neural network GRU for updating the explicit representation; Input the second sharing representation information of the target sharing user in the receiving dimension at time t and the first sharing representation information of the candidate shared user in the sharing dimension at time t into the second GRU respectively to update the implicit representation; The first sharing representation information updated in the sharing dimension and the second sharing representation information updated in the receiving dimension of the target sharing user and the candidate shared user are spliced ​​to obtain the updated sharing representation information of the target sharing user and the candidate shared user.

14. The method according to claim 13, characterized in that The sharing representation information of each target object at time t includes static sharing representation information and dynamic sharing representation information at time t; Updating the shared representation information of the target object at time t based on a recurrent neural network includes: The dynamic shared representation information of the target object at time t is input into the third GRU for updating, and the static shared representation information and the updated dynamic shared representation information of the target object are concatenated to obtain the updated shared representation information of the target object.

15. An information sharing processing device, characterized in that: include: The neighbor information acquisition module is used to respond to the information sharing request initiated by the target sharing user for the target object at time t. Based on the historical sharing behavior data, it aggregates the first historical neighbor information of the target sharing user and the candidate shared user in the sharing dimension and the second historical neighbor information in the receiving dimension, as well as the historical neighbor information of the target object. The first historical neighbor information in the sharing dimension represents the relevant information of the sharing user initiating the sharing behavior with the shared user. The second historical neighbor information in the receiving dimension represents the relevant information about the receiving user receiving the sharing behavior of the sharing user; The target object's historical neighbor information represents relevant information about the sharing user initiating the sharing behavior for the target object and relevant information about the shared user receiving the sharing behavior for the target object; a representation information generation module, configured to generate sharing representation information of the target sharing user and the candidate shared user at time t based on their respective first historical neighbor information in the sharing dimension and second historical neighbor information in the receiving dimension, and to generate sharing representation information of the target object at time t based on the target object's historical neighbor information, wherein the sharing representation information represents the willingness to share or the willingness to receive; a sharing probability prediction module, configured to predict the probability of the target sharing user sharing the target object with the candidate sharing users at time t based on the sharing representation information of the target sharing user, the candidate sharing users, and the target object at time t; The user information recommendation module is configured to provide the target user's information among the candidate users to the target sharing user according to the sharing probability, so that the target sharing user shares the target object with the target user.

16. A computer device, characterized in that: include: memory and processor; The memory is used to store a computer program, and the processor is coupled to the memory and is used to execute the computer program to implement the steps in the method according to any one of claims 1 to 14.

17. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the processor is enabled to implement the steps of the method according to any one of claims 1 to 14.

Images