Multi-Interaction Space Network-Based Recommendation and Recall Method, Device, Medium, and Product

Abstract

The present application relates to the field of computer technologies, and discloses a recommendation and recall method, device, medium, and product based on a multi-interaction space network. In this method, when a first user operation of a first user is detected at a first moment, an electronic device may, in response to the first user operation, obtain first interaction data within a first duration and second interaction data within a second duration. Then, the electronic device may determine a plurality of interaction spaces, that is, interaction spaces that the first user is interested in, according to the first interaction data and the second interaction data. In this way, by obtaining different interaction data within different durations, it is beneficial to improve the accuracy of recommended interaction spaces.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular, to a recommendation and recall method, device, medium, and computer program product based on a multi-interaction space network. Background Art

[0002] An interaction space application can be installed in an electronic device. A user can join an interaction space created by a house owner in the interaction space application and socialize with other users in the interaction space in a chat manner. In some specific implementations, the user can search for keywords of the interaction space they want to join, and the electronic device displays multiple interaction spaces corresponding to the keywords. Then, the user can select one of the multiple interaction spaces to join.

[0003] However, due to the highly diverse characteristics of user interests themselves and the relatively fast change speed of interests, keywords are only a superficial expression of user interests. The interaction space application only recommends interaction spaces to users based on keywords, which not only exposes the limitation of insufficient generalization ability, but also is difficult to deeply explore the complex and changeable real interests of users, unable to effectively capture the potential dimensions of user interests, and also difficult to adapt to the dynamic changes of interests. Moreover, just recommending around the keywords already input by the user easily makes the recommendation results often limited to the user's known interest fields, and it is very easy to cause the user to fall into an interest cocoon. Thus, the problem of the user interest cocoon caused by the insufficient generalization ability of the recommendation results not only affects the user experience in the interaction space application, but also may lead to a decrease in the user retention rate, which is not conducive to the long-term development of the interaction space application and the stability of the user group. Summary of the Invention

[0004] This application provides a recommendation and recall method, device, medium, and computer program product based on a multi-interaction space network. In this method, the electronic device can more comprehensively determine the interaction spaces that the first user is interested in by obtaining the first interaction data within the first time period corresponding to the long term and the second interaction data within the second time period corresponding to the short term, which is beneficial to improving the accuracy of the electronic device in recommending interaction spaces to users.

[0005] In a first aspect, this application provides a recommendation and recall method based on a multi-interaction space network. The method includes: detecting a first user operation of a first user at a first moment, and in response to the first user operation, obtaining the first interaction data of the first user within a first time period and the second interaction data within a second time period; wherein, the end moment of the first time period is earlier than the start moment of the second time period, and the end moment of the second time period is earlier than the first moment; generating a first graph network based on the first interaction data and a second graph network based on the second interaction data; determining a plurality of interaction spaces based on the first graph network and the second graph network; and performing a recall operation on the plurality of interaction spaces to determine at least one interaction space to be displayed.

[0006] In this way, by obtaining different interaction data of the first user within different time periods, the electronic device can combine the interaction spaces within different time periods and recommend the interaction space to the first user more comprehensively, thereby facilitating the improvement of the accuracy of the recommended interaction space.

[0007] In a possible implementation of the first aspect described above, both the first graph network and the second graph network include a plurality of nodes. The plurality of nodes include a first node representing the first user and second nodes representing each second user. Moreover, there is an edge weight between the first node and each second node, and the edge weight represents the degree of association between the first user and each second user.

[0008] In a possible implementation of the first aspect described above, the first interaction data includes social relationship data, first interaction data, and guild data, and the second interaction data includes second interaction data; the social relationship data includes at least one of the following: follow data, fan data, friend data, blacklist data; the first interaction data and the second interaction data include: interaction type, interaction time, and interaction frequency. Among them, the interaction type includes at least one of the following: voice interaction type, message interaction type, gift-giving interaction type; the guild data includes at least one of the following: check-in data, message-sending data. Among them, the check-in data is the number of times the first user checks in within the first time period, and the message-sending data is the number of times the first user sends messages in the guild within the first time period.

[0009] In a possible implementation of the first aspect described above, the edge weight between the first node and each second node in the first graph network is determined by the following method: obtaining the preset weights corresponding to each data in the social relationship data between the first node and each second node; determining the preset weights corresponding to each data in the social relationship data between the first node and each second node as the edge weight between the first node and each second node.

[0010] In a possible implementation of the first aspect described above, the preset weights corresponding to the follow data, fan data, and friend data in the social relationship data are all 1, and the preset weight corresponding to the blacklist data in the social relationship data is -1.

[0011] In a possible implementation of the first aspect described above, the edge weight between the first node and each second node in the first graph network is determined by the following method: obtaining the preset weights corresponding to each data in the social relationship data between the first node and each second node; updating the preset weights corresponding to each data in the social relationship data between the first node and each second node based on the first interaction data and guild data between the first node and each second node to obtain the edge weight between the first node and each second node.

[0012] In a possible implementation of the above first aspect, the method includes: using the preset weights corresponding to the data in the social relationship data between the first node and each second node as the first edge weights; determining the second edge weights corresponding to the first interaction data between the first node and each second node based on the first interaction data between the first node and each second node; determining the third edge weights corresponding to the guild data between the first node and each second node based on the guild data between the first node and each second node; and using the sum of the first edge weights, the second edge weights, and the third edge weights as the edge weights between the first node and each second node.

[0013] In a possible implementation of the above first aspect, determining the second edge weights corresponding to the first interaction data between the first node and each second node based on the first interaction data between the first node and each second node includes:

[0014] E a = E a，1 +E a，2 +E a，3 ;

[0015] E a，1 =W p *exp (-0.2*DayDiff p )*freq p ;

[0016] E a，2 =W a *exp (-0.2*DayDiff a )*freq a ;

[0017] E a，3 =W m *exp (-0.2*DayDiff m )*freq m ;

[0018] Wherein, E a represents the second edge weight corresponding to the first interaction data between the first node and the first interaction data, E a，1 represents the first sub-edge weight corresponding to the gift-giving interaction type, E a，2 represents the second sub-edge weight corresponding to the voice interaction type, E a，3 represents the third sub-edge weight corresponding to the message interaction type, W p is the weight corresponding to the gift-giving interaction type, W a is the weight corresponding to the voice interaction type, W m is the weight corresponding to the message interaction type, -0.2 is the time decay coefficient, DayDiff p represents the number of days from the generation of the gift-giving interaction type to the first moment, DayDiffa Indicates the number of days from the first moment when the voice interaction type is generated, DayDiff m Indicates the number of days from the first moment when the message interaction type is generated, freq p Indicates the interaction frequency corresponding to the gift-giving interaction type, freq a Indicates the interaction frequency corresponding to the voice interaction type, freq m Indicates the interaction frequency corresponding to the message interaction type.

[0019] In a possible implementation of the above first aspect, based on the guild data of the first node and each second node, determining the third edge weight corresponding to the guild data of the first node and each second node includes:

[0020] E c =W s +max(W n *freq n , 0.8);

[0021] Among them, E c Indicates the third edge weight corresponding to the first node and the guild data; W s Indicates the weight corresponding to the check-in data; W n Indicates the weight corresponding to the sent message data; freq n Indicates the sent message data.

[0022] In a possible implementation of the above first aspect, determining multiple interaction spaces based on the first graph network and the second graph network includes: determining a first eigenvector representing the first graph network and a second eigenvector representing the second graph network based on the first graph network and the second graph network; performing a vector indexing operation according to the first eigenvector and the second eigenvector, and retrieving multiple interaction spaces from the stored interaction space database, and taking the multiple interaction spaces as the first candidate interaction space set.

[0023] In a possible implementation of the above first aspect, at least one displayed interaction space meets the recommendation conditions, where the recommendation conditions include: the interaction space is in an online state, and the matching degree between the first eigenvector and the third eigenvector corresponding to the owner of the interaction space is greater than the matching degree threshold.

[0024] In a possible implementation of the first aspect above, the method for determining the matching degree includes: regarding the interaction spaces with the online status in the first candidate interaction space set as the second candidate interaction space set; obtaining the third feature vectors of the house owners of each interaction space in the second candidate interaction space set; based on the first feature vector and the third feature vector, determining the cosine similarity between the first feature vector and the third feature vector, and taking the cosine similarity as the matching degree, or using the first feature vector and the third feature vector as the input of a recommendation algorithm, outputting the recommendation scores of each interaction space in the second candidate interaction space set, and taking the recommendation scores as the matching degree.

[0025] In a possible implementation of the first aspect above, the method for determining the matching degree further includes: obtaining the values of the cosine similarity and the recommendation score at the same order of magnitude; adding the first value corresponding to the cosine similarity and the second value corresponding to the recommendation score to obtain a third value; and taking the third value as the matching degree.

[0026] In a possible implementation of the first aspect above, obtaining the third feature vectors of the house owners of each interaction space in the second candidate interaction space set includes: according to the third interaction data of the house owners of each interaction space in the second candidate interaction space set within the first time period and the fourth interaction data within the second time period, determining the third graph network corresponding to the third interaction data and the fourth graph network corresponding to the fourth interaction data; and based on the third graph network and the fourth graph network, determining the third feature vector representing the third graph network and the fourth feature vector representing the fourth graph network.

[0027] In a possible implementation of the first aspect above, taking the cosine similarity as the matching degree, the cosine similarity is determined by the following method:

[0028] ;

[0029] where represents the first feature vector, represents the third feature vector, represents the cosine similarity between the first feature vector and the third feature vector; represents the first set corresponding to at least one user who interacts with the first user within the second time period, represents the second set corresponding to at least one user who interacts with the house owner of each interaction space in the second candidate interaction space set within the second time period; represents the weight between each user in the first set and the first user, the weight between each user in the second set and the house owner; and represents the weight parameter, represents the second feature vector, Represents the fourth eigenvector.

[0030] In a second aspect, embodiments of the present application provide an electronic device, which includes: a memory for storing instructions executed by one or more processors of the electronic device; and a processor, which is one of the processors of the electronic device, for executing the instructions stored in the memory to implement the above-mentioned first aspect and the recommendation and recall method based on a multi-interaction space network mentioned in the first aspect.

[0031] In a third aspect, the present application provides a computer-readable medium, on which instructions are stored, and when the instructions are executed on a computer, the computer is caused to execute the recommendation and recall method based on a multi-interaction space network mentioned in the first aspect and any one of the first aspect of the present application.

[0032] In a fourth aspect, embodiments of the present application provide a computer program product, including: a non-volatile computer-readable storage medium, and the non-volatile computer-readable storage medium contains computer program code for executing the recommendation and recall method based on a multi-interaction space network mentioned in the first aspect and any one of the first aspect of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 According to some embodiments of the present application, a flowchart of a recommendation and recall method based on a multi-interaction space network is shown;

[0034] Figure 2 According to some embodiments of the present application, a display interface of an interaction space application is shown;

[0035] Figure 3 According to some embodiments of the present application, a schematic diagram of a first sub-graph network is shown;

[0036] Figure 4 According to some embodiments of the present application, a schematic diagram of a second sub-graph network is shown;

[0037] Figure 5 According to some embodiments of the present application, a schematic diagram of a third sub-graph network is shown;

[0038] Figure 6 According to some embodiments of the present application, a schematic diagram of a first graph network is shown;

[0039] Figure 7 According to some embodiments of the present application, a schematic diagram of a second graph network is shown;

[0040] Figure 8 According to some embodiments of the present application, an interaction space list display interface is shown;

[0041] Figure 9According to some embodiments of the present application, a schematic diagram of a software architecture is shown;

[0042] Figure 10 According to some embodiments of the present application, a schematic diagram of the hardware structure of an electronic device 1000 is shown. Detailed implementation manners

[0043] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application.

[0044] It can be understood that the recommendation and recall method based on a multi-interaction space network mentioned in the embodiments of the present application can be applied to an electronic device. This electronic device can also be referred to as a mobile station (MS), a mobile terminal (MT), etc. In some specific implementation manners, the electronic device can be a mobile phone, a smart TV, a wearable device, a tablet computer (Pad), a desktop computer, a laptop computer, a virtual reality (VR) device, a server, etc. The embodiments of the present application do not limit the specific form of the electronic device.

[0045] In some specific implementations, the servers applicable to the recommendation and recall method based on a multi-interaction space network provided in the embodiments of the present application include, but are not limited to, application servers, server clusters, etc. The embodiments of the present application do not limit the specific form of the server.

[0046] In some optional embodiments, the interaction space may refer to a voice room or a video room created or joined by a user in software such as a social media platform or a game community for real-time interaction with other users. For example, the user can create or join a chat interaction space to have voice communication and / or video communication with other users in the interaction space; or the user can create or join a blind date interaction space to make friends with other users in the interaction space by means of voice chat.

[0047] Therefore, to solve the above problems, the present application provides a recommendation and recall method based on a multi-interaction space network. In this method, when a first user operation of a first user is detected at a first moment, the electronic device can, in response to the first user operation, obtain first interaction data of the first user within a first time period and second interaction data within a second time period. Wherein, the end moment of the first time period is earlier than the start moment of the second time period, and the end moment of the second time period is earlier than the first moment. Then, the electronic device can determine a plurality of interaction spaces based on the first interaction data and the second interaction data, and use the plurality of interaction spaces as a first candidate interaction space set. The plurality of interaction spaces are related to the interaction space corresponding to the first interaction data and are related to the interaction space corresponding to the second interaction data. In this way, by obtaining different interaction data within different time periods, the electronic device can more comprehensively determine the interaction spaces that the first user is interested in, which is beneficial to improving the accuracy of recommending interaction spaces to the user.

[0048] It can be understood that the first candidate interaction space set is the interaction space that the first user is interested in determined by the first interaction data and the second interaction data.

[0049] In some optional embodiments, the first time period can be represented as a long term, and the second time period can be represented as a short term, that is, the first time period is greater than the second time period. For example, the first time period is 30 days and the second time period is 1 day; or, the first time period is 21 days and the second time period is 1 day. Therefore, by obtaining the first interaction data within the first time period, the electronic device can determine the interaction spaces that the first user is interested in in the long term (such as 30 days); and by obtaining the second interaction data within the second time period, the electronic device can determine the interaction spaces that the first user is interested in in the short term. Furthermore, the electronic device comprehensively considers the interaction spaces that the first user is interested in in the long term and the interaction spaces that the first user is interested in in the short term, and determines a plurality of interaction spaces. In this way, taking 30 days as the first time period can more clearly reflect the interaction spaces that the first user is interested in in the long term, and can avoid the interaction data within too long a time period (such as 100 days) from affecting the electronic device's determination of the interaction spaces that the first user is interested in in the long term due to excessive data volume and too long a time span. And taking 1 day as the second time period can specifically determine the interaction spaces that the first user is interested in in the short term, which is beneficial for the electronic device to further combine the interaction spaces that the first user is interested in in the short term on the basis of the interaction spaces that the first user is interested in in the long term, accurately determine the interaction spaces that the first user is interested in, and is beneficial to improving the accuracy of recommending interaction spaces.

[0050] In some optional embodiments, the first interaction data may include social relationship data, first interaction data, and guild data, and the second interaction data may include second interaction data.

[0051] Among them, the social relationship data may include: follow data, fan data, friend data, blacklist data; the first interaction data and the second interaction data may include: interaction type, interaction time, and interaction frequency, and the interaction type may include: voice interaction type, message interaction type, gift-giving interaction type. The guild data may include: check-in data, message-sending data, where the check-in data refers to the number of times the first user checks in within the first time period, and the message-sending data refers to the number of times the first user sends messages in the guild within the first time period.

[0052] In this way, corresponding to the first time period representing the long term, by using the social relationship data, interaction data, and guild data generated through the interaction between the first user and multiple second users to determine the first interaction data of the first user, the type of interaction space that the first user is interested in can be determined more comprehensively. And corresponding to the second time period representing the short term, considering that in the short term, the changes in the social relationship data and guild data of the first user are relatively small, while the changes in the second interaction data are relatively large. For example, the first user has multiple voice interactions with the second user within the second time period, but the social relationship data of the first user may not change. Therefore, the second interaction data is emphasized within the second time period.

[0053] In this way, in the case of obtaining the interaction data within the first time period and the second time period, according to the different time periods, the focus of the obtained interaction data is also different. On the basis of the interaction space that the first user is interested in determined in multiple aspects, and then according to the second interaction data, the interaction space that the first user is interested in is supplemented. Furthermore, the interaction space that the first user is interested in can be determined more precisely, which is beneficial to improving the accuracy of the interaction space determined by the electronic device based on the first interaction data and the second interaction data.

[0054] In some alternative embodiments, the electronic device may generate a first graph network based on the first interaction data and a second graph network based on the second interaction data. The first graph network and the second graph network include multiple nodes, and the multiple nodes include a first node representing the first user and a second node representing each second user. And there is an edge weight between the first node and each second node. It can be understood that the second user is the user who interacts with the first user. For example, the second user is the user followed by the first user, etc. Therefore, the electronic device can determine multiple interaction spaces according to the first graph network and the second graph network.

[0055] In some alternative embodiments, the edge weight represents the degree of association between the first user and each second user, and the greater the edge weight, the higher the degree of association between the first user and the corresponding second user.

[0056] In some alternative embodiments, the preset weights corresponding to the respective data in the social relationship data between the first node and each second node may be determined as the first edge weights between the first node and each second node. Among them, the preset weights corresponding to the respective data in the social relationship data between the first node and each second node may include: the preset weights corresponding to the follow data, fan data, and friend data are all 1, and the preset weight corresponding to the blacklist data in the social relationship data is -1.

[0057] In some alternative embodiments, the first graph network is composed of a first sub-graph network corresponding to social data, a second sub-graph network corresponding to first interaction data, and a third sub-graph network corresponding to guild data. Therefore, the edge weights between the first node and each second node in the first graph network may be determined by updating the preset weights corresponding to the respective data in the social relationship data between the first node and each second node based on the first interaction data and guild data between the first node and each second node.

[0058] In some alternative embodiments, the update of the preset weights corresponding to the respective data in the social relationship data between the first node and each second node may be reflected in: determining the second edge weights corresponding to the first interaction data between the first node and each second node based on the first interaction data between the first node and each second node; determining the third edge weights corresponding to the guild data between the first node and each second node based on the guild data between the first node and each second node; and using the sum of the first edge weights, second edge weights, and third edge weights as the edge weights between the first node and each second node, thereby realizing the update of the preset weights.

[0059] Exemplarily, taking the determination of the edge weight between user A and user B within the first duration as an example, corresponding to the interaction type between user A and user B being the message interaction type, the edge weight E a，3 =W m *exp (-0.2*DayDiff m )*freq m . Among them, W m is the weight corresponding to the message interaction type, W m = 0.1. -0.2 is the time decay coefficient. DayDiff m represents the number of days since the message interaction type occurred from the first moment. freq m represents the interaction frequency corresponding to the message interaction type. For example, if the interaction time between user A and user B is 7 days before the first moment and the interaction frequency is 20 times, then the edge weight E a，3 = 0.1*exp (-0.2*7)*20. Another example is that user B is a user followed by user A, and the preset weight E f，AB= 1. Further, in the first graph network, based on the first interaction data between the first node and each second node, the preset weights corresponding to the data in the social relationship data between the first node and each second node can be updated, and the edge weight between user A and user B can be determined as E a，3 + E f，AB .

[0060] In some alternative embodiments, in the second graph network, based on the second interaction data between the first node and each second node, the fourth edge weight corresponding to the second interaction data between the first node and each second node can be determined. Taking the determination of the fourth edge weight between user A and user B within the second time period as an example, corresponding to the interaction type between user A and user B being the gift-giving interaction type, the edge weight E a，4 = W p * exp(-0.2 * DayDiff p ) * freq p . Wherein, W p is the weight corresponding to the gift-giving interaction type, W p = 0.6. freq p represents the interaction frequency corresponding to the gift-giving interaction type. The time decay coefficient is -0.2. DayDiff p represents the number of days since the gift-giving interaction type occurred until the first moment. It can be understood that DayDiff p = 0. Therefore, in the second graph network, when the interaction type between user A and user B is the gift-giving interaction type, the corresponding fourth edge weight E a，4 = W p * freq p .

[0061] In some other alternative embodiments, in the second graph network, determining the fourth edge weight corresponding to the second interaction data between the first node and each second node may further include: determining the fourth edge weight based on the interaction type and the partial interaction frequency corresponding to the interaction type in the second interaction data between the first node and each second node.

[0062] Among them, the partial interaction frequency corresponding to the interaction type can be understood correspondingly as: when the data volume corresponding to the second interaction data within the second time period is greater than the data volume threshold, to improve the operation speed of the electronic device, the interaction types can be sorted in reverse order of interaction time, and a first interaction list can be generated. Further, determine the interaction types that meet the preset quantity and the corresponding interaction frequencies in the first interaction list, and then determine the fourth edge weight between the first node and each second node in the second graph network according to the preset quantity of interaction types and the corresponding interaction frequencies.

[0063] Thus, different edge weight calculation methods are set according to the different generation methods of social relationship data, interaction data (including first interaction data and second interaction data), and guild data, and the edge weights between the first user and the second user are reflected in the first graph network and the second graph network. By more deeply integrating the interaction data of the second user and determining the closeness between the second user and the first user through the values of the edge weights, multiple interaction spaces can be determined, which can improve the accuracy of recommended interaction spaces.

[0064] The following introduces the recommendation and recall method based on a multi-interaction space network provided by the embodiments of the present application with reference to the accompanying drawings.

[0065] Figure 1 FIG. shows a schematic flowchart of a recommendation and recall method based on a multi-interaction space network. This method can be applied to an electronic device, such as the server mentioned above. Specifically, the recommendation and recall method based on a multi-interaction space network may include:

[0066] S101: Detect a user operation of the first user.

[0067] In some optional embodiments, the electronic device may have an interaction space application. The user recommended an interaction space by the interaction space application may be referred to as the first user, and the user who interacts with the first user in the interaction space application may be referred to as the second user.

[0068] As Figure 2 shown, FIG. shows a display interface of an interaction space application. The first user information display sub-interface 201 in the display interface 200 may be used to display the account (identity document, ID), i.e., the ID number, level, and other user information of the first user. Among them, the ID number may represent the unique identification number of the first user in the interaction space application, and the level may represent the familiarity of the first user with using the interaction space application.

[0069] In some specific implementations, the display interface 200 may further include a novice area component 202, an entertainment area component 203, a master area component 204, a leisure area component 205, a home page component 206, a message component 207, and a my component 208. Among them, the novice area component 202, the entertainment area component 203, and the master area component 204 are used to provide an entrance for the first user to participate in games, and the first user can interact by participating in games; the leisure area component 205 is used to provide an interaction room for the first user to interact. The message component 207 is used to provide an entrance for the first user to perform message interactions.

[0070] Exemplarily, the electronic device can jump to the corresponding game interface in response to the first user clicking on the novice area component 202, the entertainment area component 203, or the expert area component 204. The electronic device can display at least one interaction space to the first user in response to the first user clicking on the casual area component 205. And corresponding to the first user clicking on the home page component 206, the electronic device presents the display interface 200 to the first user. The electronic device can jump to the message display interface in response to the first user clicking on the message component 207, and the first user can interact with multiple second users through the message display interface. The electronic device can jump to the "My" display interface in response to the first user clicking on the "My" component 208, and the first user can fill in personal information such as hobbies and personality through the "My" display interface.

[0071] S102: Detect a first user operation of the first user at a first moment, and in response to the first user operation, obtain first interaction data and second interaction data.

[0072] In some specific implementations, the electronic device can detect a first user operation at a first moment, that is, the electronic device can detect that the first user clicks on the casual area component 205 at the first moment, and this first user operation can be used to obtain a recommended interaction space. It can be understood that the first user operation can also include voice control operations, etc., and the present application does not make specific limitations. It can be understood that one illustration of the first moment can refer to the moment when it is detected that the first user clicks on the casual area component 205.

[0073] In some specific implementations, after detecting the first user operation of the first user, the electronic device, in response to the first user operation, obtains first interaction data within a first duration and second interaction data within a second duration.

[0074] In some optional instances, the start moment of the first duration is earlier than the first moment, the end moment of the first duration is earlier than the start moment of the second duration, and the duration corresponding to the first duration is greater than the duration corresponding to the second duration.

[0075] In some optional embodiments, the first duration can be represented as long - term, and the second duration can be represented as short - term, that is, the first duration is greater than the second duration. For example, the first duration is 30 days and the second duration is 1 day; or, the first duration is 21 days and the second duration is 1 day.

[0076] In some optional embodiments, the first interaction data can include social relationship data, first interaction data, and guild data, and the second interaction data can include second interaction data.

[0077] Among them, the social relationship data may include: follow data, fan data, friend data, and blacklist data. The first interaction data and the second interaction data may include: interaction type, interaction time, and interaction frequency. The interaction type may include: voice interaction type, message interaction type, and gift-giving interaction type. The guild data may include: check-in data and message-sending data. Among them, the check-in data refers to the number of times the first user checks in within the first time period, and the message-sending data refers to the number of times the first user sends messages in the guild within the first time period.

[0078] In this way, corresponding to the first time period representing the long term, the first interaction data of the first user is determined based on the social relationship data, interaction data, and guild data generated by the interaction between the first user and multiple second users. And, corresponding to the second time period representing the short term, considering that in the short term, the changes in the social relationship data and guild data of the first user are relatively small, while the changes in the second interaction data are relatively large. For example, the first user has multiple voice interactions with the second user within the second time period, but the social relationship data of the first user may not change. Therefore, the second interaction data is emphasized within the second time period.

[0079] In this way, when obtaining the interaction data within the first time period and the second time period, the focus of the obtained interaction data is different according to the different time periods. On the basis of the interaction space that the first user is interested in determined in multiple aspects, and then according to the second interaction data, the interaction space that the first user is interested in is supplemented. Furthermore, the interaction space that the first user is interested in can be determined more precisely, which is beneficial to improving the accuracy of the interaction space determined by the electronic device according to the first interaction data and the second interaction data.

[0080] S103: Generate a first graph network and a second graph network based on the first interaction data and the second interaction data.

[0081] In some alternative embodiments, the electronic device may generate a first graph network based on the first interaction data and a second graph network based on the second interaction data. The first graph network and the second graph network include multiple nodes, and the multiple nodes include a first node representing the first user and a second node representing each second user. And, there is an edge weight between the first node and each second node. It can be understood that the second user is the user who interacts with the first user. For example, the second user is the user followed by the first user, etc.

[0082] In some alternative embodiments, the edge weight represents the degree of association between the first user and each second user. The larger the edge weight, the higher the degree of association between the first user and the corresponding second user.

[0083] In some alternative embodiments, the preset weights corresponding to the respective data in the social relationship data between the first node and each second node may be determined as the first edge weights between the first node and each second node. Among them, the preset weights corresponding to the respective data in the social relationship data between the first node and each second node may include: the preset weights corresponding to the follow data, fan data, and friend data are all 1, and the preset weight corresponding to the blacklist data in the social relationship data is -1.

[0084] In some alternative embodiments, since the electronic device determines the first interaction data by obtaining social relationship data, first interaction data, and guild data, the electronic device further determines the first graph network by determining the first sub-graph network corresponding to the social relationship data, the second sub-graph network corresponding to the first interaction data, and the third sub-graph network corresponding to the guild data. Moreover, the edge weights between the first node and each second node in the first graph network may be determined by updating the preset weights corresponding to the respective data in the social relationship data between the first node and each second node based on the first interaction data and guild data between the first node and each second node.

[0085] In some alternative embodiments, the update of the preset weights corresponding to the respective data in the social relationship data between the first node and each second node may be reflected in: determining the second edge weights corresponding to the first interaction data between the first node and each second node based on the first interaction data between the first node and each second node; determining the third edge weights corresponding to the guild data between the first node and each second node based on the guild data between the first node and each second node; using the sum of the first edge weights, second edge weights, and third edge weights as the edge weights between the first node and each second node, thereby realizing the update of the preset weights.

[0086] In some specific implementations, Figure 3 shows a schematic diagram of a first sub-graph network, the first sub-graph network G f，U can display the interaction between the first user and multiple second users in terms of social relationship data. The first edge weights between the first node and each second node in the first sub-graph network are the preset weights corresponding to the social relationship data.

[0087] Exemplarily, the first user is user A, and the second users include user B, user C, user D, and user E. User B is a friend of user A, user C is a follow of user A, user D is a fan of user A, and user E is in the blacklist of user A. Then the preset weight between user A and user B is 1, the preset weight between user A and user C is 1, the preset weight between user A and user D is 1, and the preset weight between user A and user E is -1. Furthermore, the edge weight E f，AB between user A and user B = 1, the edge weight E f，AC= 1. Edge weight E between user A and user D f，AD = 1. Edge weight E between user A and user E f，AE = -1. That is, for the first edge weight corresponding to the social relationship data, the edge weights generated by the follow data, fan data, and friend data in the social relationship data are 1, and the edge weight generated by the blacklist data in the social relationship data is -1.

[0088] In some specific implementations, the second edge weight corresponding to the first interaction data between the first node and the second node is determined by adding the first sub-edge weight corresponding to the gift-giving interaction type, the second sub-edge weight of the voice interaction type, and the third sub-edge weight of the message interaction type.

[0089] Among them, the first sub-edge weight is determined by the weight corresponding to the gift-giving interaction type, the number of days from the first moment when the first user and the second user generate the gift-giving interaction type, and the interaction frequency corresponding to the gift-giving interaction type between the first user and the second user. The second sub-edge weight is determined by the weight corresponding to the voice interaction type, the number of days from the first moment when the first user and the second user generate the voice interaction type, and the interaction frequency corresponding to the voice interaction type between the first user and the second user. The third sub-edge weight is determined by the weight corresponding to the message interaction type, the number of days from the first moment when the first user and the second user generate the message interaction type, and the interaction frequency corresponding to the message interaction type between the first user and the second user.

[0090] Figure 4 Shows a schematic diagram of a second subgraph network, the second subgraph network G a，U Can display the interaction between the first user and multiple second users in terms of interaction data. In the second subgraph network G a，U Among them, the first user is user A, and the multiple second users are user B and user F respectively. Furthermore, the electronic device can determine the edge weight between user A and user B, and the edge weight between user A and user F according to the interaction type, interaction time, interaction frequency, and the weight corresponding to the interaction type.

[0091] Specifically, the second edge weight in the second subgraph network G a，U Can be determined by the following formula.

[0092] E a，1 = W p * exp(-0.2 * DayDiff p ) * freq p (1);

[0093] E a，2 = W a * exp(-0.2 * DayDiff a ) * freq a (2);

[0094] E a，3 =W m *exp(-0.2*DayDiff m )*freq m (3);

[0095] E a = E a，1 +E a，2 +E a，3 (4);

[0096] Among them, E a represents the second edge weight corresponding to the first node and the first interaction data. E a，1 represents the first sub-edge weight corresponding to the gift-giving interaction type, E a，2 represents the second sub-edge weight corresponding to the voice interaction type, E a，3 represents the third sub-edge weight corresponding to the message interaction type. W p is the weight corresponding to the gift-giving interaction type, W p =0.6. W a is the weight corresponding to the voice interaction type, W a =0.3. W m is the weight corresponding to the message interaction type, W m =0.1. -0.2 is the time decay coefficient. DayDiff p represents the number of days from the first moment when the first user and the second user have a gift-giving interaction, that is, the interaction time. DayDiff p represents the number of days from the first moment when the gift-giving interaction type occurs. DayDiff a represents the number of days from the first moment when the voice interaction type occurs. DayDiff m represents the number of days from the first moment when the message interaction type occurs. freq p represents the interaction frequency corresponding to the gift-giving interaction type. freq a represents the interaction frequency corresponding to the voice interaction type. freq m represents the interaction frequency corresponding to the message interaction type.

[0097] Exemplarily, user A gave a gift to user B once one day before the first moment, user A had 10 voice interactions with user F three days before the first moment, and user A had 20 message interactions with user F seven days before the first moment.

[0098] Thus, based on formulas (1) to (4), determine the edge weight E between user A and user B a，AB = 0.6*exp(-0.2*1)*1. The edge weight E between user A and user Fa，AF = 0.3*exp (-0.2*3)*10+0.1*exp (-0.2*7)*20。

[0099] Furthermore, in the first graph network, based on the first interaction data between the first node and each second node, the preset weights corresponding to the data in the social relationship data between the first node and each second node are updated, and the edge weight between user A and user B can be determined as E a，AB +E f，AB 。

[0100] In some specific implementations, Figure 5 a schematic diagram of a third sub-graph network is shown. The third sub-graph network G c，U can display the interactions between the first user and multiple second users in terms of the guild. In the third sub-graph network G c，U the first user is user A, and the multiple second users are user B and user D respectively. Moreover, user A, user B, and user D are users in the same guild. Furthermore, the electronic device can determine the edge weight between user A and user B, and the edge weight between user A and user D based on the check-in data and message-sending data of user A in the guild.

[0101] Specifically, the third edge weight in the third sub-graph network G c，U can be determined by the following formula.

[0102] E c =W s +max(W n *freq n ,0.8) (5);

[0103] wherein, E c represents the third edge weight corresponding to the guild data between the first node and the second node. W s represents the weight corresponding to the check-in data. If the number of check-ins of the first user in the guild is less than or equal to the check-in threshold, W s =0, if the number of check-ins of the first user in the guild is greater than the check-in threshold, W s =0.4. W n represents the weight corresponding to the message-sending data in the guild, W n =0.1. freq n represents the message-sending data, that is, the number of messages sent by the first user in the guild within the first time period.

[0104] Exemplarily, when the number of check-ins of user A in the guild within the first time period is 10 times, which is greater than the check-in threshold, and the number of messages sent in the guild is 10 times. Then the edge weight E between user A and user B c，ABEqual to the edge weight E between user A and user D c，AD , that is, E c，AB = E c，AD = 0.4 + max(0.1 * 10, 0.8).

[0105] In some specific implementations, based on Figure 3 , Figure 4 and Figure 5 the sub-network schematic diagrams shown, the first graph network schematic diagram can be determined as shown in Figure 6 . The first graph network adds up multiple edge weights generated by the first user and the same second user in the first sub-network, the second sub-network, and the third sub-network. For example, the edge weights between user A and user B include E f，AB , E a，AB , E c，AB . By adding E f，AB , E a，AB , E c，AB , all the edge weights between user A and user B can be determined.

[0106] In this way, the electronic device obtains the first interaction data of the first user within the first time period, generates the first graph network according to the first interaction data, and through the first graph network and the edge weights in the first graph network, can more clearly show the degree of association between the first user and multiple second users. Thus, combined with the interaction spaces of interest to the second users, multiple interaction spaces can be more accurately determined.

[0107] In some specific implementations, generating the second graph network according to the second interaction data may include: generating the second graph network according to multiple second users who interact with the first user within the second time period, as well as the interaction types and interaction times of the interaction. Among them, the interaction types include gift-giving interaction type, message interaction type, and voice interaction type. The weight corresponding to each interaction type in the second interaction data can be the fourth weight, and the fourth edge weight between the first node and each second node in the second graph network can be determined according to the fourth weight and all the interaction frequencies corresponding to the interaction type.

[0108] In some alternative embodiments, in the second graph network, based on the second interaction data of the first node and each second node, the fourth edge weight corresponding to the second interaction data of the first node and each second node can be determined. Taking the determination of the fourth edge weight between user A and user B within the second time period as an example, corresponding to the interaction type between user A and user B being the gift-giving interaction type, the edge weight E a，4 = W p * exp(-0.2 * DayDiff p ) * freq p . Among them, Wp is the weight corresponding to the gift - giving interaction type, W p = 0.6. freq p represents the interaction frequency corresponding to the gift - giving interaction type. - 0.2 is the time decay coefficient. DayDiff p represents the number of days from the generation of the gift - giving interaction type to the first moment. It can be understood that DayDiff p = 0. Therefore, in the second graph network, when the interaction type between user A and user B is the gift - giving interaction type, the weight E of the fourth edge a，4 = W p * freq p .

[0109] For example Figure 7 shows a schematic diagram of a second graph network. In the second graph network, the first user is user A, and the multiple second users are user B, user C, and user D respectively. Exemplarily, user B has 20 voice interactions with user A, user C has 30 message interactions with user A, and user D has 40 gift - giving interactions with user A. Furthermore, the edge weight E between user A and user B a，AB = 0.3 * 20, the edge weight E between user A and user C a，AC = 0.3 * 30, and the edge weight E between user A and user D a，AD = 0.3 * 40.

[0110] In this way, for the different generation methods of social relationship data, interaction data (including first - type interaction data and second - type interaction data), and guild data, different edge - weight calculation methods are set, and the edge weights between the first user and the second user are reflected by the first graph network and the second graph network. More deeply combining the interaction space that the second user is interested in, and determining the closeness between the second user and the first user through the numerical value of the edge weight, multiple interaction spaces can be determined, which can improve the accuracy of the recommended interaction space.

[0111] S104: Determine multiple interaction spaces based on the first graph network and the second graph network, and display at least one interaction space.

[0112] In some alternative embodiments, based on the first graph network and the second graph network, a first eigenvector representing the first graph network and a second eigenvector representing the second graph network can be determined. Then, perform a vector indexing operation on the first eigenvector and the second eigenvector, retrieve multiple interaction spaces from the stored interaction space database, and use the multiple interaction spaces as the first candidate interaction space set.

[0113] Among them, the interactive space database is used to store interactive spaces that meet the storage conditions, and the storage conditions may include: the creation time of the interactive space is within a preset time period, the online duration of the interactive space is within a preset duration range, or the number of users joining the interactive space is greater than a preset number, etc.

[0114] It can be understood that the first feature vector and the second feature vector respectively represent the first interaction data and the second interaction data of the first user. Therefore, an operation of vector indexing is performed on the first feature vector and the second feature vector, and multiple interactive spaces are initially retrieved from the stored interactive space database. Then, further screening is performed on the first candidate interactive space set where the multiple interactive spaces are located, and at least one interactive space that can be displayed to the first user can be determined.

[0115] Specifically, a recall operation is performed on the first candidate interactive space set to determine at least one interactive space to be displayed, and the at least one interactive space to be displayed needs to meet the recommendation conditions. The recommendation conditions include: the interactive space is in an online state, and the matching degree between the first feature vector and the third feature vector of the owner of the interactive space is greater than the matching degree threshold.

[0116] In some optional embodiments, first, according to information such as whether the owner of the interactive space corresponding to the interactive space in the first candidate interactive space set is online, a second candidate interactive space set can be initially determined from the first candidate interactive space set (for example, the second candidate interactive space set includes 100 interactive spaces).

[0117] Furthermore, the electronic device also needs to determine the matching degree between the first feature vector and the third feature vector of the owner of each interactive space in the second candidate interactive space set.

[0118] In some optional embodiments, the matching degree can be determined in the following manner:

[0119] First, obtain the third feature vector of the owner of each interactive space in the second candidate interactive space set. In some instances, the electronic device can determine the third graph network corresponding to the third interaction data and the fourth graph network corresponding to the fourth interaction data according to the third interaction data of the owner of each interactive space in the second candidate interactive space set within the first time period and the fourth interaction data within the second time period. Furthermore, based on the third graph network and the fourth graph network, the third feature vector representing the third graph network and the fourth feature vector representing the fourth graph network can be determined.

[0120] Then, based on the first feature vector of the first user and the third feature vectors of the homeowners of each interaction space in the second candidate interaction space set, the cosine similarity between the first feature vector and the third feature vectors is determined, and the cosine similarity is used as the matching degree. The higher the cosine similarity, the higher the matching degree, and this interaction space should be recommended to the first user. Alternatively, the first feature vector and the third feature vectors are used as the input of a recommendation algorithm (e.g., a fine-ranking model), and the recommendation scores of each interaction space in the second candidate interaction space set are output, and the recommendation scores are used as the matching degree. The higher the recommendation score, the higher the matching degree, and this interaction space should be recommended to the first user.

[0121] In some embodiments, the way to determine the matching degree further includes: obtaining the values of the cosine similarity and the recommendation score at the same order of magnitude. For example, the value ranges of both the cosine similarity and the recommendation score are [0, 1]. The first value corresponding to the cosine similarity is added to the second value corresponding to the recommendation score to obtain a third value. The third value is used as the matching degree.

[0122] In some alternative embodiments, the cosine similarity can be determined in the following way:

[0123] Formula (6);

[0124] Wherein, represents the first feature vector, represents the third feature vector, represents the cosine similarity between the first feature vector and the third feature vector; represents the first set corresponding to at least one user who interacts with the first user within the second time period, represents the second set corresponding to at least one user who interacts with the homeowners of each interaction space in the second candidate interaction space set within the second time period; represents the weight between each user in the first set and the first user, the weight between each user in the second set and the homeowner; and represents the weight parameter, represents the second feature vector, represents the fourth feature vector.

[0125] In some alternative embodiments, represents the first set corresponding to at least one user who interacts with the first user within the second time period, that is, the second user who interacts with the first user within the second time period. Correspondingly, can be represented as the weight between the second user and the first user, such as the edge weight between the first user and the second user mentioned above.

[0126] In some alternative embodiments, corresponding to the electronic device determining the cosine similarity between the first feature vector and the third feature vector, as well as the recommendation scores of each interaction space in the second candidate interaction space set. The electronic device may use, as the recommended interaction spaces, the interaction spaces in the second candidate interaction space set whose cosine similarity is greater than the similarity threshold (corresponding to the matching degree threshold mentioned above). Alternatively, the electronic device may use, as the recommended interaction spaces, the interaction spaces in the second candidate interaction space set whose recommendation scores are greater than the score threshold (corresponding to the matching degree threshold mentioned above). Or, the electronic device uses, as the recommended interaction spaces, the interaction spaces in the second candidate interaction space set whose third value is greater than the matching degree threshold.

[0127] In this way, based on the first interaction data and the second interaction data of the user, and further using the cosine similarity and the recommendation algorithm (such as the refined ranking model), the electronic device can specifically recommend interaction spaces to the first user according to the interaction spaces that the first user is interested in, which is beneficial to improving the user experience and can increase the user retention rate.

[0128] In some alternative examples, at least one interaction space recommended by the electronic device to the first user may be presented in the form of a list, such as Figure 8 illustrates an interaction space list display interface. The interaction space list display interface 801 may include multiple interaction space components, such as the interaction space component one 802 corresponding to homeowner one, the interaction space component two 803 corresponding to homeowner two, the interaction space component three 804 corresponding to homeowner three, the interaction space component four 805 corresponding to homeowner four, and the interaction space component five 806 corresponding to homeowner five.

[0129] Taking the interaction space component one 802 corresponding to homeowner one as an example, in the interface corresponding to the interaction space component one 802 corresponding to homeowner one, interface one 811 may display the homeowner name, homeowner one; interface two 812 may display the type of the interaction room, such as type 1; interface three 813 may be used to display the use of the interaction room, such as making friends; interface four 814 may be used to display the number of the interaction room, such as room 4676. The first user can find the interaction room by searching for the number of the interaction room; interface five 815 may be used to display the number of people in the interaction room and the maximum number of people who can join. For example, 8 / 11 may indicate that there are 8 users in the current interaction room and the current interaction room can accommodate a maximum of 11 users.

[0130] In some alternative embodiments, the first user may also click on the random component 807 in the interaction space list display interface 801 to randomly join an interaction room.

[0131] The following introduces the software architecture mentioned in the embodiments of the present application with reference to the accompanying drawings.

[0132] like Figure 9 As shown, the software architecture mentioned in the embodiment of the present application includes a data processing layer, a model training layer, a recall layer, a sorting layer and a re-arrangement layer.

[0133] The data processing layer includes a social relationship data network manager, an interactive data network manager, a guild data network manager, and a real-time interactive sorting manager. The social relationship data network manager is used to obtain the social relationship data in the first interactive data, and generate a first subgraph network based on the social relationship data. The interactive data network manager is used to obtain the first interactive data in the first interactive data, and the second interactive data in the second interactive data, and generate a second subgraph network based on the first interactive data, and generate a second graph network based on the second interactive data. The guild data network manager is used to obtain the guild data in the first interactive data, and generate a third subgraph network based on the guild data. The real-time interactive sorting manager is used to sort the first interactive data and the second interactive data of the first user in reverse chronological order.

[0134] The model training layer is used to use the first graph network and the second graph network as inputs of the graph convolutional network model to determine the first eigenvector and the second eigenvector.

[0135] The recall layer is used to perform multi-interaction space network recommendation recall, that is, to perform the above-mentioned S104 based on the first graph network and the second graph network to determine multiple interaction spaces based on the user vector database.

[0136] In some embodiments, an interaction space list may be determined based on the interaction spaces in the second candidate interaction space set whose cosine similarity is greater than a similarity threshold. The interaction space list may be called a user recall list.

[0137] The ranking layer is used to execute the above S104 based on the refined ranking model: taking the first feature vector and the third feature vector as inputs of the refined ranking model, and outputting a recommendation score for each interaction space in the second candidate interaction space set.

[0138] The reordering layer can be used to break up the types of interactive spaces, support cold start traffic for new hosts, and recommend interactive spaces created by new hosts to first users. Figure 8 In the example, the first user clicks on the random component 807 in the interactive space list display interface 801, and the electronic device can also recommend the interactive space created by the new homeowner to the first user, or recommend the type of interactive space that the first user is not interested in to the first user.

[0139] The hardware structure of the electronic device is introduced below. For example, the electronic device may be the electronic device 1000 mentioned below.

[0140] Figure 10 The figure shows a schematic diagram of the hardware structure of the electronic device 1000. It can be understood that the structure illustrated in the embodiments of the present invention does not constitute a specific limitation on the electronic device 1000. In other embodiments of the present application, the electronic device 1000 may include more or fewer components than shown in the figure, or combine certain components, or split certain components, or have different component arrangements. The illustrated components may be implemented in hardware, software, or a combination of software and hardware.

[0141] The electronic device 1000 may include a processor 1001, an external memory interface 1002, an internal memory 1003, a universal serial bus (USB) interface 1004, a charging management module 1005, a power management module 1006, a battery 1007, an antenna one, an antenna two, a mobile communication module 1008, a wireless communication module 1009, an audio module 1010, a sensor module 1011, a key 1012, a motor 1013, an indicator 1014, a camera 1015, a display screen 1016, and a subscriber identification module (SIM) card interface 1017, etc. Among them, the sensor module 1011 may include a pressure sensor 1011A, a gyroscope sensor 1011B, a barometric pressure sensor 1011C, a magnetic sensor 1011D, an acceleration sensor 1011E, a distance sensor 1011F, a proximity light sensor 1011G, a fingerprint sensor 1011H, a temperature sensor 1011J, a touch sensor 1011K, an ambient light sensor 1011L, a bone conduction sensor 1011M, etc.

[0142] The processor 1001 may include one or more processing units. For example, the processor 1001 may include an application processor (AP), a modem processor, a graphics processing unit (GPU), an image signal processor (ISP), a controller, a video codec, a digital signal processor (DSP), a baseband processor, and / or a neural-network processing unit (NPU), etc. Among them, different processing units may be independent devices or integrated in one or more processors.

[0143] The controller may generate operation control signals according to the instruction operation code and timing signals to complete the control of fetching and executing instructions.

[0144] A memory may also be provided in the processor 1001 for storing instructions and data. In some embodiments, the memory in the processor 1001 is a cache memory. This memory can hold the instructions or data that the processor 1001 has just used or recycled. If the processor 1001 needs to use the instruction or data again, it can directly call it from the memory. This avoids repeated accesses, reduces the waiting time of the processor 1001, and thus improves the efficiency of the system.

[0145] In some embodiments, the processor 1001 may include one or more interfaces. The interfaces may include an inter-integrated circuit (I2C) interface, an inter-integrated circuit sound (I2S) interface, a pulse code modulation (PCM) interface, a universal asynchronous receiver / transmitter (UART) interface, a mobile industry processor interface (MIPI), a general-purpose input / output (GPIO) interface, a subscriber identity module (SIM) interface, and / or a universal serial bus (USB) interface, etc.

[0146] The charging management module 1005 is configured to receive a charging input from a charger. The charger may be a wireless charger or a wired charger. While charging the battery 1007, the charging management module 1005 can also supply power to the electronic device 1000 through the power management module 1006.

[0147] The power management module 1006 is used to connect the battery 1007, the charging management module 1005, and the processor 1001. The power management module 1006 receives the inputs from the battery 1007 and / or the charging management module 1005 and supplies power to the processor 1001, the internal memory 1003, the display screen 1016, the camera 1015, the wireless communication module 1009, etc.

[0148] The wireless communication function of the electronic device 1000 can be implemented through Antenna 1, Antenna 2, the mobile communication module 1008, the wireless communication module 1009, the modulation and demodulation processor, and the baseband processor, etc.

[0149] Antenna 1 and Antenna 2 are used for transmitting and receiving electromagnetic wave signals. Each antenna in the electronic device 1000 can be used to cover a single or multiple communication frequency bands. Different antennas can also be multiplexed to improve the utilization rate of the antennas.

[0150] The mobile communication module 1008 can provide solutions for wireless communications including 2G / 3G / 4G / 5G, etc. applied to the electronic device 1000.

[0151] The modulation and demodulation processor can include a modulator and a demodulator. Among them, the modulator is used to modulate the low-frequency baseband signal to be transmitted into a medium-high frequency signal. The demodulator is used to demodulate the received electromagnetic wave signal into a low-frequency baseband signal.

[0152] The wireless communication module 1009 can provide solutions for wireless communications including wireless local area networks (WLAN) (such as wireless fidelity (Wi-Fi) networks), Bluetooth (BT), global navigation satellite system (GNSS), frequency modulation (FM), near field communication (NFC), infrared technology (IR), etc. applied to the electronic device 1000. The wireless communication module 1009 can be one or more devices integrating at least one communication processing module. The wireless communication module 1009 receives electromagnetic waves via Antenna 2, performs frequency modulation and filtering processing on the electromagnetic wave signals, and sends the processed signals to the processor 1001. The wireless communication module 1009 can also receive the signals to be transmitted from the processor 1001, perform frequency modulation and amplification on them, and convert them into electromagnetic waves through Antenna 2 for radiation.

[0153] The electronic device 1000 realizes the display function through the GPU, the display screen 1016, and the application processor, etc. The GPU is a microprocessor for image processing, connected to the display screen 1016 and the application processor. The GPU is used to perform mathematical and geometric calculations for graphics rendering. The processor 1001 can include one or more GPUs, which execute program instructions to generate or change the display information.

[0154] The display screen 1016 is used to display images, videos, etc. The display screen 1016 includes a display panel. The display panel can be a liquid crystal display (LCD), an organic light-emitting diode (OLED), an active-matrix organic light-emitting diode (AMOLED), a flexible light-emitting diode (FLED), a Miniled, a MicroLed, a Micro-oLed, a quantum dot light-emitting diode (QLED), etc. In some embodiments, the electronic device 1000 may include one or N display screens 1016, where N is a positive integer greater than 1.

[0155] In some alternative examples, the display screen 1016 can display such as Figure 2 and Figure 8 the interface shown.

[0156] The electronic device 1000 can implement the shooting function through the ISP, the camera 1015, the video codec, the GPU, the display screen 1016, the application processor, etc.

[0157] The external memory interface 1002 can be used to connect an external memory card, such as a Micro SD card, to expand the storage capacity of the electronic device 1000. The external memory card communicates with the processor 1001 through the external memory interface 1002 to implement the data storage function.

[0158] The internal memory 1003 can be used to store computer-executable program code, and the executable program code includes instructions. The internal memory 1003 can include a program storage area and a data storage area. Among them, the program storage area can store the operating system, application programs required for at least one function, etc. The data storage area can store data created during the use of the electronic device 1000, etc. In addition, the internal memory 1003 can include a high-speed random access memory, and can also include a non-volatile memory, such as at least one disk storage device, a flash memory device, a universal flash storage (UFS), etc. The processor 1001 executes various functional applications and data processing of the electronic device 1000 by running the instructions stored in the internal memory 1003 and / or the instructions stored in the memory provided in the processor.

[0159] The button 1012 includes a power-on button, volume buttons, etc. The button 1012 can be a mechanical button or a touch button. The electronic device 1000 can receive button inputs and generate key signal inputs related to the user settings and function controls of the electronic device 1000.

[0160] The motor 1013 can generate vibration prompts. The motor 1013 can be used for incoming call vibration prompts or touch vibration feedback.

[0161] The indicator 1014 can be an indicator light and can be used to indicate the charging state, power change, or to indicate messages, missed calls, notifications, etc.

[0162] The SIM card interface 1017 is used to connect the SIM card. The SIM card can be inserted into or removed from the SIM card interface 1017 to achieve contact and separation from the electronic device 1000. The electronic device 1000 can support 1 or N SIM card interfaces, where N is a positive integer greater than 1. The SIM card interface 1017 can support Nano SIM cards, Micro SIM cards, SIM cards, etc. Multiple cards can be inserted into the same SIM card interface 1017 simultaneously. The types of multiple cards can be the same or different. The SIM card interface 1017 can also be compatible with different types of SIM cards. The SIM card interface 1017 can also be compatible with external memory cards. The electronic device 1000 interacts with the network through the SIM card to implement functions such as calls and data communication. In some embodiments, the electronic device 1000 uses an eSIM, i.e., an embedded SIM card. The eSIM card can be embedded in the electronic device 1000 and cannot be separated from the electronic device 1000.

[0163] The embodiments of the present application provide a computer-readable medium with instructions stored thereon. When the instructions are executed on an electronic device, the electronic device executes the recommendation and recall method based on a multi-interaction space network provided by the embodiments of the present application.

[0164] The embodiments of the present application provide a computer program product, including: a non-volatile computer-readable storage medium containing computer program code for executing the recommendation and recall method based on a multi-interaction space network mentioned in the present application.

[0165] It should be noted that each unit / module mentioned in the device embodiments of the present application is a logical unit / module. Physically, a logical unit / module can be a physical unit / module, a part of a physical unit / module, or can be implemented by a combination of multiple physical units / module. The physical implementation manner of these logical units / modules themselves is not the most important. The combination of the functions implemented by these logical units / modules is the key to solving the technical problems proposed in the present application. In addition, in order to highlight the innovative part of the present application, the above device embodiments of the present application do not introduce units / modules that are not closely related to solving the technical problems proposed in the present application. This does not mean that there are no other units / modules in the above device embodiments.

[0166] It should be noted that in the examples and descriptions of the present application, the term "comprising", "including" or any other variation thereof is intended to cover non-exclusive inclusion, such that a process, method, article or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or device. Without further limitation, an element defined by the phrase "comprising one" does not exclude the presence of additional identical elements in the process, method, article or device comprising the element.

[0167] Although the present application has been illustrated and described by reference to certain embodiments thereof, those of ordinary skill in the art should understand that various changes may be made in form and detail without departing from the spirit and scope of the present application.

Claims

1. A recommendation recall method based on a multi-interaction space network, characterized in that Applied to an electronic device, the method includes: Detecting a first user operation of a first user at a first moment, In response to the first user operation, obtaining first interaction data of the first user within a first duration and second interaction data within a second duration; wherein, an end moment of the first duration is earlier than a start moment of the second duration, and an end moment of the second duration is earlier than the first moment; Generating a first graph network based on the first interaction data, and generating a second graph network based on the second interaction data; Determining a plurality of interaction spaces based on the first graph network and the second graph network; Performing a recall operation on the plurality of interaction spaces to determine at least one interaction space to be displayed; The generating a first graph network based on the first interaction data includes: Determining a first sub-graph network corresponding to social relationship data in the first interaction data, a second sub-graph network corresponding to first interaction data in the first interaction data, and a third sub-graph network corresponding to guild data in the first interaction data; Generating the first graph network according to the first sub-graph network, the second sub-graph network, and the third sub-graph network; The generating a second graph network based on the second interaction data includes: Generating the second graph network according to interaction types and interaction times of a plurality of second users interacting with the first user within the second duration; The determining a plurality of interaction spaces based on the first graph network and the second graph network includes: Determining a first feature vector representing the first graph network and a second feature vector representing the second graph network based on the first graph network and the second graph network; Performing a vector indexing operation according to the first feature vector and the second feature vector, retrieving a plurality of interaction spaces from a stored interaction space database, and taking the plurality of interaction spaces as a first candidate interaction space set.

2. The method according to claim 1, wherein Both the first graph network and the second graph network include a plurality of nodes, the plurality of nodes include a first node representing the first user and second nodes representing each second user, and there is an edge weight between the first node and each second node, and the edge weight represents the degree of association between the first user and each second user.

3. The method according to claim 2, wherein The first interaction data includes social relationship data, first interaction data, and guild data, and the second interaction data includes second interaction data; The social relationship data includes at least one of the following: follow data, fan data, friend data, blacklist data; The first interaction data and the second interaction data include: interaction type, interaction time, and interaction frequency, wherein the interaction type includes at least one of the following: voice interaction type, message interaction type, gift-giving interaction type; The guild data includes at least one of the following: check-in data, message sending data, wherein the check-in data is the number of times the first user checks in within the first duration, and the message sending data is the number of times the first user sends messages in the guild within the first duration.

4. The method according to claim 3, characterized in that, Determine the edge weights between the first node and each second node in the first graph network in the following manner: Obtain the preset weights corresponding to each data in the social relationship data between the first node and each second node; Determine the preset weights corresponding to each data in the social relationship data between the first node and each second node as the edge weights between the first node and each second node.

5. The method according to claim 4, wherein The preset weights corresponding to the follow data, the fan data, and the friend data in the social relationship data are all 1, and the preset weight corresponding to the blacklist data in the social relationship data is -1.

6. The method according to claim 3, characterized in that, Determine the edge weights between the first node and each second node in the first graph network in the following manner: Obtain the preset weights corresponding to each data in the social relationship data between the first node and each second node; Based on the first interaction data and the guild data between the first node and each second node, update the preset weights corresponding to each data in the social relationship data between the first node and each second node to obtain the edge weights between the first node and each second node.

7. The method according to claim 6, wherein The method includes: Take the preset weights corresponding to each data in the social relationship data between the first node and each second node as the first edge weights; Based on the first interaction data between the first node and each second node, determine the second edge weights corresponding to the first interaction data between the first node and each second node; Based on the guild data between the first node and each second node, determine the third edge weights corresponding to the guild data between the first node and each second node; Take the sum of the first edge weights, the second edge weights, and the third edge weights as the edge weights between the first node and each second node.

8. The method according to claim 7, wherein The determining the second edge weights corresponding to the first interaction data between the first node and each second node based on the first interaction data between the first node and each second node includes: E a = E a，1 + E a，2 + E a，3 ; E a，1 =W p *exp(-0.2*DayDiff p )*freq p ; E a，2 =W a *exp(-0.2*DayDiff a )*freq a ; E a，3 =W m *exp(-0.2*DayDiff m )*freq m ; Among them, E a represents the second edge weight corresponding to the first node and the first interaction data, E a，1 represents the first sub-edge weight corresponding to the gift-giving interaction type, E a，2 represents the second sub-edge weight corresponding to the voice interaction type, E a，3 represents the third sub-edge weight corresponding to the message interaction type, W p is the weight corresponding to the gift-giving interaction type, W a is the weight corresponding to the voice interaction type, W m is the weight corresponding to the message interaction type, -0.2 is the time decay coefficient, DayDiff p represents the number of days from the generation of the gift-giving interaction type to the first moment, DayDiff a represents the number of days from the generation of the voice interaction type to the first moment, DayDiff m represents the number of days from the generation of the message interaction type to the first moment, freq p represents the interaction frequency corresponding to the gift-giving interaction type, freq a represents the interaction frequency corresponding to the voice interaction type, freq m represents the interaction frequency corresponding to the message interaction type.

9. The method according to claim 7, characterized in that, The determining the third edge weights corresponding to the guild data between the first node and each second node based on the guild data between the first node and each second node includes: E c =W s + max(W n * freq n , 0.8); Among them, E c represents the weight of the third edge corresponding to the first node and the guild data; W s represents the weight corresponding to the check-in data; W n represents the weight corresponding to the send message data; freq n represents the send message data.

10. The method according to claim 1, wherein At least one of the displayed interaction spaces meets the recommendation conditions, wherein the recommendation conditions include: the interaction space is in an online state, and the matching degree between the first feature vector and the third feature vector corresponding to the owner of the interaction space is greater than the matching degree threshold.

11. The method according to claim 10, wherein The determining manner of the matching degree includes: Take the interaction spaces in the first candidate interaction space set with the status of being in an online state as the second candidate interaction space set; Obtain the third feature vectors of the owners of each interaction space in the second candidate interaction space set; Based on the first feature vector and the third feature vector, determine the cosine similarity between the first feature vector and the third feature vector, and use the cosine similarity as the matching degree. Alternatively, use the first feature vector and the third feature vector as the input of a recommendation algorithm, output the recommendation scores of each interaction space in the second candidate interaction space set, and use the recommendation scores as the matching degree.

12. The method according to claim 11, wherein The determination method of the matching degree further includes: Obtain the values of the cosine similarity and the recommendation score at the same order of magnitude; Add the first value corresponding to the cosine similarity and the second value corresponding to the recommendation score to obtain a third value; Use the third value as the matching degree.

13. The method according to claim 11, wherein The obtaining of the third feature vector of the house owner of each interaction space in the second candidate interaction space set includes: According to the third interaction data of the house owner of each interaction space in the second candidate interaction space set within the first time period and the fourth interaction data within the second time period, determine the third graph network corresponding to the third interaction data and the fourth graph network corresponding to the fourth interaction data; Based on the third graph network and the fourth graph network, determine the third feature vector representing the third graph network and the fourth feature vector representing the fourth graph network.

14. The method according to claim 13, characterized in that, The cosine similarity is determined by the following method: ； Wherein, represents the first eigenvector, represents the third eigenvector, represents the cosine similarity between the first eigenvector and the third eigenvector; represents a first set corresponding to at least one user who interacts with the first user within the second time period, represents a second set corresponding to at least one user who interacts with the homeowner of each interaction space in the second candidate interaction space set within the second time period; represents the weight between each user in the first set and the first user, the weight between each user in the second set and the homeowner; and represents a weight parameter, represents the second eigenvector, represents the fourth eigenvector.

15. An electronic device, characterized in that, Including: A memory for storing instructions executed by one or more processors of the electronic device, and the processor, which is one of the one or more processors of the electronic device, for executing the multi-interaction space network-based recommendation and recall method according to any one of claims 1 to 14.

16. A computer-readable medium, characterized in that, Instructions are stored on the readable medium, and when the instructions are executed on a computer, the computer executes the multi-interaction space network-based recommendation and recall method according to any one of claims 1 to 14.

17. A computer program product, characterized in that, Including a computer program / instructions, and when the computer program / instructions are executed by a processor, the multi-interaction space network-based recommendation and recall method according to any one of claims 1 to 14 is implemented.

Images