Information pushing method and device and electronic equipment
By constructing a social network graph, based on users' network access and social interaction data, the system identifies family and friends groups and pushes relevant information, solving the problem of information push not meeting user needs and achieving more accurate and efficient information push.
Patent Information
- Application Number
- CN202511697567.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-19
- Publication Date
- 2026-02-13
AI Technical Summary
The information pushed by existing technologies fails to meet the actual needs of users, resulting in a waste of resources.
By acquiring network access records and social interaction data from user devices, a social network graph can be constructed to identify the user's family and friends groups, and product information purchased in the past by these groups can be pushed to the user.
It improved the relevance of information delivery, reduced resource waste, and enhanced the reliability and accuracy of information delivery.
Smart Images
Figure CN121531025A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of computer technology, and in particular to an information push method, device and electronic device. Background Technology
[0002] With the rapid development of electronic devices, they have become indispensable necessities for people's travel and are widely integrated into various aspects of daily life. To attract users, businesses often use electronic devices to push various benefit information, such as coupons and discounts. However, because the pushed content often fails to meet users' actual needs, it not only inconveniences users but also wastes resources. Summary of the Invention
[0003] This application provides an information push method, apparatus, and electronic device to address the problem that the pushed content fails to meet the user's actual needs, resulting in resource waste.
[0004] To solve the above-mentioned technical problems, this application is implemented as follows:
[0005] In a first aspect, embodiments of this application provide an information push method, the method comprising:
[0006] When a first user equipment accesses a first network device, N user equipments that have historically accessed the first network device are identified. The first user equipment is the user equipment of the first user. The N user equipments correspond one-to-one with the N users, and N is an integer greater than 0.
[0007] Obtain the association relationships between the N users and the first user, including interaction relationships;
[0008] A social network graph is constructed based on the aforementioned relationships, and the family and friends group of the first user is determined based on the social network graph. The family and friends group includes at least some of the N users.
[0009] Obtain the first product purchased in the user's history in the family and friends group, and push the information of the first product to the first user device.
[0010] Secondly, embodiments of this application provide an information push device, the device comprising:
[0011] The first acquisition module is used to determine N user devices that have historically accessed the first network device when the first user device accesses the first network device. The first user device is the user device of the first user, and the N user devices correspond one-to-one with the N users, where N is an integer greater than 0.
[0012] The second acquisition module is used to acquire the association relationship between the N users and the first user, the association relationship including the interaction relationship;
[0013] The determination module is used to construct a social network graph based on the association relationship, and determine the family and friends group of the first user based on the social network graph, wherein the family and friends group includes at least some of the N users;
[0014] The third acquisition module is used to acquire the first product that the user in the family and friends group has purchased in the past, and push the information of the first product to the first user device.
[0015] Thirdly, embodiments of this application provide an electronic device, including: a processor, a memory, and a program stored in the memory and executable on the processor, wherein when the program is executed by the processor, it implements the steps of the information push method described in the first aspect.
[0016] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the information push method described in the first aspect.
[0017] Fifthly, a computer program product is provided, including computer instructions that, when executed by a processor, implement the steps of the information push method as described in the first aspect.
[0018] In this embodiment, by obtaining the network access records of the first user's device and the social interactions of the first user, a social network graph of the first user is constructed to determine the first user's relatives and friends groups, and messages about the first products purchased in the history of the relatives and friends groups are pushed to the first user. This can push information when the user arrives at a specific area, which is more in line with the user's needs and reduces resource waste. Attached Figure Description
[0019] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0020] Figure 1 This is a flowchart of an information push method provided in an embodiment of this application;
[0021] Figure 2 This is a schematic diagram of the structure of an information push device provided in an embodiment of this application;
[0022] Figure 3This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation
[0023] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0024] This application provides an information push method, apparatus, and electronic device to address the problem that the pushed content fails to meet the user's actual needs, resulting in resource waste.
[0025] See Figure 1 , Figure 1 This is a flowchart of an information push method provided in an embodiment of this application, such as... Figure 1 As shown, the method includes the following steps:
[0026] Step 101: When the first user equipment is connected to the first network device, determine the N user equipments that have historically connected to the first network device. The first user equipment is the user equipment of the first user. The N user equipments correspond one-to-one with the N users, and N is an integer greater than 0.
[0027] Step 102: Obtain the association relationship between the N users and the first user, the association relationship including the interaction relationship;
[0028] Step 103: Construct a social network graph based on the aforementioned relationships, and determine the first user's family and friends group based on the social network graph. The family and friends group includes at least some of the N users.
[0029] Step 104: Obtain the first product purchased in the user's history in the family and friends group, and push the information of the first product to the first user device.
[0030] The first network device can be an indoor distribution system or a WiFi device.
[0031] In some implementations, when a first user equipment is detected switching from a second network device (such as a macro base station) to the first network device, it indicates that the user has moved from outdoors to indoors, and the system retrieves the user equipment's historical access records to the first network device. By acquiring the user's access records to the first network device, the accuracy of in-store detection can be improved.
[0032] Obtain the association relationship between the first user corresponding to the first user device and N users, such as whether there is an interaction relationship or a preference relationship (whether they have the same preferences).
[0033] With user authorization, interactive relationships can be determined through phone calls, messages, and shared locations. They can also be established through interactions on social media platforms, such as message interactions, comment interactions, Moments interactions, and likes.
[0034] Preference relationships can be determined through transaction records, comments, and messages on the platform. For example, if user device one buys a beverage on a food delivery platform and comments "This drink tastes great," and user device two also comments "I really like this drink," it indicates that user device one and user device share the same preference. Alternatively, it can be obtained through methods such as interest tag matching.
[0035] Based on the above relationships, a social network graph can be constructed. Nodes in the graph represent each user, edges represent the relationships between two users, and the weight of an edge represents the strength of that relationship. Users with weights exceeding a preset threshold can be merged into friend / family groups.
[0036] In some implementations, connection weights between graph nodes are constructed based on shared access records and social closeness among users.
[0037] In some implementations, a graph is constructed based on network access data, and another graph is constructed based on interaction data. The two graphs are then fused according to their weights to obtain the final social network graph.
[0038] Social network graphs can be used to identify groups of friends and family members with strong connections to the first user. For example, community detection algorithms or other algorithms can be used to detect these groups.
[0039] The system retrieves the first product purchased in the user's family and friends group and pushes information about the first product to the first user's device. This information may include resources associated with the first product, such as coupons and discount information.
[0040] In some implementations, the system obtains the first product that users in a group of friends and family members have purchased most frequently in the past, and pushes a message about the first product to the first user.
[0041] In some implementations, the system obtains the first product purchased in the history of users in a group of friends and family. If the user's preferences are similar to those of the first user, a message about the first product is pushed to the first user.
[0042] By analyzing the network access records of the first user device and the social interactions of the first user, a social network graph of the first user is constructed, and the first user's family and friends groups are identified. When the first user device is detected to be connected to the first network device, a message about the first product historically purchased by the family and friends groups is pushed to the first user, which can meet the user's needs and reduce resource waste. In some implementations, when the first user device is detected to be connected to the first network device for a duration longer than a preset duration, a message about the first product is pushed to the first user device.
[0043] Optionally, the N user devices include the second user device corresponding to the second user; obtaining the association relationship between the N users and the first user includes:
[0044] The system acquires access data of the first user equipment and the second user equipment to the first network device, as well as interaction data between the first user and the second user.
[0045] The step of constructing a social network graph based on the aforementioned relationships and determining the first user's family and friends groups based on the social network graph includes:
[0046] A social network graph is constructed based on the access data and the interaction data, and the association strength between the first user and the second user is determined based on the social network graph.
[0047] If the association strength is greater than a preset strength value, the second user will be added to the first user's family and friends group.
[0048] The access data may include user equipment information, location identification information, access timestamps, etc. Based on the acquired access data, the access data for the first user equipment accessing the first network and the access data for the second user equipment accessing the first network can be determined.
[0049] The fact that both the first user equipment and the second user equipment have accessed the first network device indicates that both user equipment have accessed the first network device.
[0050] In addition, interaction data between the first and second users can be obtained.
[0051] A social network graph can be constructed based on access data and interaction data from a shared network. In this graph, nodes represent users, edges represent interactions between users, and edge weights indicate the strength of the interaction. If the weight of the edge between a first user and a second user is greater than a preset strength value, it indicates a strong interaction between them, and the second user is added to the first user's friend / family group.
[0052] By employing the methods described above, the accuracy of information obtained from friend and family groups can be improved, thereby enhancing the reliability of information recommendations for the primary product.
[0053] Optionally, determining the association strength between the first user and the second user based on the social network graph includes:
[0054] Based on the social network graph, the association strength between the first user and the second user is determined using the following formula:
[0055]
[0056] in, This indicates the strength of the association between the first user and the second user;
[0057] α and β are adjustment parameters, α+β=1;
[0058] "Joint Access Count" refers to the number of times the first user and the second user jointly access the first network.
[0059] max(Z) represents the maximum number of times that the first user and the second user access the first network device;
[0060] The expression is used to represent the interaction intensity between the first user and the second user, where max(s) represents the maximum interaction intensity between the first user and the second user.
[0061] Let the user set be U={ , ,…, Device access records contain the following characteristics:
[0062] Device ID: Access time: Access location: Social interaction data (first user (user)) With the second user (user) (interaction intensity) .
[0063] Construct a social network graph G=(V,E) using graph theory, where: nodes V represents user ,node V represents user ;side E, the weight of the edge. Indicates user and The strength of the association.
[0064] If the first user If there is no purchase history, the message will be pushed directly to the first user's device to the family and friends group. The first product purchased frequently in the medium to high frequency:
[0065] Recommend( )= ;
[0066] in, For indicator functions, This represents the number of recommendations.
[0067] By using the methods described above, the accuracy of obtaining information about the first user's family and friends groups can be improved, thereby increasing the reliability of the pushed information.
[0068] Optionally, adding the second user to the first user's family and friends group when the association strength is greater than a preset strength value includes:
[0069] The association strength between the N users and the first user is detected using a preset detection algorithm;
[0070] Target users whose association strength with the first user is greater than the preset strength value are identified as the first user's friends and family group, and the target users include the second user.
[0071] A community detection algorithm (e.g., Louvain's algorithm) is used to segment the social network graph, outputting results for family and friend groups. Specifically, by identifying high-density subgraphs in the social network, the first user and N users who share the same connection are assigned to corresponding communities based on their association strength. Target users whose association strength with the first user is greater than a preset strength value are assigned to the community corresponding to the user's family and friend group. A target user can include one or more users.
[0072] By using the above methods, the accuracy of segmenting the first user's family and friends groups can be improved, thereby increasing the reliability of push notifications for the first product.
[0073] Optionally, the family and friends group includes a third user; the step of obtaining the first product historically purchased by users in the family and friends group and pushing the information of the first product to the first user's device includes:
[0074] The system obtains a first rating from the first user for a second product purchased in the past and a second rating from the third user for the first product purchased in the past. The ratings are used to characterize the user's preference for the product.
[0075] Using the first rating and the second rating, the similarity between the first user and the third user is determined;
[0076] Based on the magnitude of the similarity, a first predicted rating for the first user on the first product is determined;
[0077] If the first predicted score is greater than the preset score, information about the first product is pushed to the first user device.
[0078] Obtain the historical purchase records of the first user's family and friends group (including at least some of the N users), and count the frequency of product purchases by users in this group. In some implementations, identify the product with the highest purchase frequency in the family and friends group as the first product.
[0079] In some implementations, each product in the purchase history is sequentially retrieved as the first product, along with a third user who previously purchased the first product.
[0080] Obtain a second rating from a third user for the first product they have purchased in the past. This rating is used to characterize the third user's preference for the first product.
[0081] Obtain the first user's first rating of the second product in the first user's history. This rating is used to characterize the first user's preference for the second product.
[0082] The rating can be statistically analyzed based on factors such as product type, purchase frequency, purchase preferences, and amount (with user authorization).
[0083] Let the product set be P={ , ,…, }, First User For the second product The product rating (or number of purchases) is Second user For the second product The product rating (or number of purchases) is Construct a sparse matrix R .
[0084] User-based collaborative filtering can calculate the first user in any of the following ways. Second User The similarity between them.
[0085] Cosine similarity: User and The similarity is:
[0086]
[0087] Pearson correlation coefficient (treatment rating bias):
[0088]
[0089] in For users The average rating.
[0090] For the first user Products not purchased The predicted score is:
[0091]
[0092] in, To and The K most similar user sets
[0093] Using the above method, based on the first user's rating of the second product, the third user's rating of the first product, and the similarity between the first and third users, a predicted rating for the first user on the first product can be determined. If this predicted rating exceeds a preset threshold, information about the first product is pushed to the first user's device when the first user's device triggers a push condition (such as entering a shopping mall's WiFi coverage area).
[0094] By calculating the similarity between users and products, recommending products similar to users' historical consumption behavior can improve the reliability of information push for the primary product, reduce interference with the primary user, and reduce resource waste.
[0095] Optionally, determining the first predicted rating of the first user for the first product based on the similarity score includes:
[0096] Based on the magnitude of the similarity, a second predicted rating for the first user on the first product is determined;
[0097] Using the second predicted score and the association between the first user and the third user, a first predicted score for the first user on the first product is determined.
[0098] The third user can be the same as or different from the second user. A second predicted rating for the first product is calculated based on the similarity between the first and third users, as well as the similarity between the products. Furthermore, considering social influence, a first predicted rating is calculated based on the second predicted rating.
[0099] First User For the product The final first prediction score for:
[0100]
[0101] in, For the first user Family and friends group; [0,1] is the balancing parameter (default value is 0.7).
[0102] For the first user and third-party users The intensity of interaction, where the third user and the second user are different users.
[0103] By incorporating the influence of social networks, the accuracy of the first user's predictive rating of the first product is improved.
[0104] Optionally, pushing the information of the first product to the first user equipment includes:
[0105] The first user's behavior data on the second product purchased in history is input into a pre-trained model to obtain information about the first product. The first user's third predicted score for the first product is higher than a preset score. The third predicted score is used to characterize the first user's preference for the first product.
[0106] The information of the first product is pushed to the first user equipment, wherein the first product is a single product or a combination of multiple products of different types.
[0107] Acquire user behavior data and input the data into the model to train the model.
[0108] For example, the system uses user purchase data such as product type, purchase frequency, number of purchases, and purchase amount as input data to train a model using a collaborative filtering algorithm. During training, the system continuously adjusts the model parameters to optimize the accuracy of the recommendation results.
[0109] The training process is as follows:
[0110] 1. Preprocess the data by performing min-max normalization on the rating matrix R:
[0111]
[0112] Missing value imputation: Use the average user rating or the average product rating to fill in the missing values.
[0113] 2. Model Training and Optimization
[0114] Objective function: Minimize the mean squared error (MSE) between the predicted and actual ratings.
[0115]
[0116] in , A potential factor matrix for users and products. is the regularization coefficient.
[0117] Bayesian optimization: Searching for hyperparameters (such as K, ...) using a Gaussian process. , This maximizes recommendation accuracy.
[0118] After training, the behavioral data of the first user regarding the second product is input into the model. This behavioral data can include information about the first user's purchasing behavior related to the second product, such as purchase frequency and preferences. The model can output the first user's purchase preferences, as well as the predicted ratings the first user gives to products in the product set. It will then output a list of products with predicted ratings greater than a preset value, including the first product.
[0119] In some implementations, the first product is a single product.
[0120] In some implementations, the first product includes multiple different types of products within the same store.
[0121] In some implementations, the first product includes a combination package of multiple products from different stores.
[0122] It is possible to push relevant information to the first user based on the first product acquired.
[0123] In some implementations, the predicted ratings of a first user for multiple products across stores can also be calculated in the following manner.
[0124] 1. Sharing of potential factors across stores
[0125] Let the set of shops be S = { , ,…, Each store's product collection is as follows: By sharing user potential factors Build cross-store rating prediction:
[0126]
[0127] in Let be the product potential factors for store s. User potential factors and product potential factors represent the frequency of user purchases and the frequency of product purchases, respectively.
[0128] 2. Joint Recommendation Strategy
[0129] Generate a joint recommendation list based on users' predicted ratings across stores:
[0130] Cross-Recommend )=
[0131] in For users In the store Historical activity.
[0132] The combined recommendation list can include a coefficient for recommending multiple products together.
[0133] Through the above methods, by sharing the user's latent factor matrix It constructs a cross-store rating model, combines users' historical activity to generate a cross-store joint recommendation list, breaks down data silos, supports multi-store collaboration, enhances user consumption stickiness, and improves the efficiency of merchant resource conversion and integration. It is suitable for commercial complexes or chain brand scenarios.
[0134] In some implementations, an initial cell is identified, and first communication network handover count data for that initial cell is obtained. Suspected cells are then identified using the first communication network handover count data. Communication network access data for the suspected cells is obtained. Signal strength distribution characteristic parameters and access count distribution characteristic parameters for the suspected cells are calculated based on the communication network access data. Coverage cells for the target location are then determined from the suspected cells based on the signal strength distribution characteristic parameters and the access count distribution characteristic parameters, thereby improving the accuracy and efficiency of coverage cell identification for the target location. However, this method, which relies on user access information from base stations, only reflects a portion of the in-store situation; some users may connect to the physical merchant's wireless network.
[0135] In other implementations, multiple target location data of the client are acquired; wherein, the target location data is data representing the location of the target client, and the multiple target location data includes target latitude and longitude data, scanned target beacon data, and target sensor data; the target location data is compared with the location fingerprints of multiple points of interest in a preset location fingerprint database to obtain points of interest that match any target location data, which are then used as candidate points of interest; based on a preset recommendation model, the points of interest visited by the client are determined according to the target location data and the location fingerprints of the candidate points of interest, which are then used as target points of interest; and recommendation information for the target points of interest is pushed to the client.
[0136] Some implementations also include a membership marketing platform that integrates business marketing solutions, comprising a membership management module, a merchant management module, a product management module, a marketing management module, a data analysis module, an order management module, and a database. The data analysis module includes a data collection module, a feature calculation module, a trust calculation module, an interest similarity calculation module, an emotion perception module, a time and location analysis module, and a comprehensive analysis module. This aims to address issues such as difficult customer management, rising labor costs, low customer traffic, and high promotion costs, as well as the problem of excessive fragmented information for merchants and the need to avoid manual billing. It achieves scenario-based intelligent operation, comprehensively integrating the merchant's customer flow, capital flow, product flow, and information flow to realize smart business management. However, analyzing user consumption patterns and factors influencing consumption makes it difficult to provide accurate commercial recommendations immediately upon user entry into the store, and it also cannot combine analysis from multiple physical stores.
[0137] In this embodiment, the system identifies a user's network access behavior through the network access records of a first user device. Specifically, the system analyzes information such as the user's device ID, access time, and access location to identify the user's network access mode and determine if the user has visited the store. Furthermore, the system constructs the user's social circle by analyzing shared access records with other devices and frequent interaction behaviors. Members in the social circle are linked through unique identifiers such as device IDs and social media accounts, forming the user's social network graph.
[0138] Then, the system records user consumption behavior, including purchase history, spending amount, purchase frequency, and consumption preferences. By analyzing user consumption behavior, the system can identify user consumption habits and preferences. Based on this data, the system uses a collaborative filtering algorithm, combining user consumption behavior with product characteristics, to generate a personalized product recommendation list. Specifically, the system recommends products similar to the user's historical consumption behavior by calculating the similarity between the user and the product.
[0139] In calculating similarity, a user-product matrix can be constructed. This matrix is built based on user consumption behavior, where each element represents a user's rating or purchase behavior for a product. A similarity matrix is generated by calculating the similarity between users or between products. Similarity calculations can employ methods such as cosine similarity and Pearson correlation coefficient. Based on the similarity matrix, the system generates a personalized product recommendation list. For products that a user has not purchased, a weighted average or weighted summation method can be used to predict the user's rating or purchase probability for that product.
[0140] By identifying the spending behavior of users within a friend or family group, the system infers products that the first user might be interested in. Specifically, the system uses a collaborative filtering algorithm to analyze the purchase records of users in the friend or family group and recommends products to the user that match the purchasing behavior of those users. This social network-based recommendation method can effectively improve the accuracy of recommendations, especially when the user has limited personal spending history.
[0141] User consumption behavior is recorded, and this data is used to train a recommendation model. Specifically, methods such as Bayesian optimization can be used to continuously adjust model parameters and improve the accuracy of recommendation results. After training, the model can output the user's consumption preferences and a list of recommended products. When outputting recommended products, a recommendation threshold is determined based on historical data. When the recommendation coefficient is greater than the threshold, the system considers the product suitable for recommendation to the user; when the recommendation coefficient is less than the threshold, the system considers the product unsuitable for recommendation. The determination of the recommendation threshold can be based on the relationship between historical recommendation success rate and recommendation coefficient, improving the accuracy and effectiveness of recommendations.
[0142] The system applies the trained recommendation model to collaborative activities among multiple stores. Specifically, by analyzing user spending behavior and recommendation results, the system identifies products suitable for joint recommendations across multiple stores. For example, the system can identify that if a user purchases a certain type of product from store A, they might be interested in a similar product from store B. Based on this, the system can generate cross-store joint recommendation strategies to help multiple stores jointly conduct cross-store bundled promotions and enhance the user's shopping experience.
[0143] In this embodiment, a user social graph is constructed using user device access records (such as device ID, access time, geographical location, and social interactions). Graph theory and community detection algorithms (such as the Louvain algorithm) are used to segment family and friend groups. Social influence is quantified and directly integrated into the recommendation system, solving the cold start problem and completing social network-driven user modeling.
[0144] In addition, based on traditional user-product collaborative filtering, a social influence-weighted score prediction is introduced. The system uses the average social circle rating to dynamically balance user preferences and the influence of social networks. By adjusting the parameter γ, a personalized recommendation strategy is achieved (e.g., γ→1 for users who prefer privacy, and γ→0 for users who rely on social networks), which improves the accuracy of recommendations, especially when user behavior is sparse or social circle consumption trends are significant.
[0145] The above methods solve the following problems: 1. How to analyze whether users are in-store customers based on their wireless device access status; 2. How to combine user purchasing behavior and interaction data to accurately push information to users; 3. Analyze the arrival of each user at each merchant based on big data sharing, output relevant consumption results, and thus carry out joint activities.
[0146] See Figure 2 , Figure 2 This is a schematic diagram of the structure of an information push device provided in an embodiment of this application, as shown below. Figure 2 As shown, the information push device 200 includes:
[0147] The first acquisition module 201 is used to determine N user devices that have historically accessed the first network device when the first user device accesses the first network device. The first user device is the user device of the first user, and the N user devices correspond one-to-one with the N users, where N is an integer greater than 0.
[0148] The second acquisition module 202 is used to acquire the association relationship between the N users and the first user, the association relationship including the interaction relationship;
[0149] The determining module 203 is used to construct a social network graph based on the association relationship, and determine the family and friends group of the first user based on the social network graph, wherein the family and friends group includes at least some of the N users;
[0150] The push module 204 is used to obtain the first product that the user in the family and friends group has purchased in the past, and push the information of the first product to the first user device.
[0151] Optionally, the N user equipments include the second user equipment corresponding to the second user; the second acquisition module is specifically used for:
[0152] The system acquires access data of the first user equipment and the second user equipment accessing the same network, as well as interaction data between the first user and the second user.
[0153] The determining module includes:
[0154] A construction submodule is used to construct a social network graph based on the access data and the interaction data, and to determine the association strength between the first user and the second user based on the social network graph;
[0155] A submodule is added to add the second user to the first user's family and friends group when the association strength is greater than a preset strength value.
[0156] Optionally, the construction submodule is specifically used for:
[0157] Based on the social network graph, the association strength between the first user and the second user is determined using the following formula:
[0158]
[0159] in, This indicates the strength of the association between the first user and the second user;
[0160] α and β are adjustment parameters, α+β=1;
[0161] This indicates the number of times the first user and the second user jointly access the first network;
[0162] max(Z) represents the maximum number of times that the first user and the second user access the first network device;
[0163] The expression is used to represent the interaction intensity between the first user and the second user, where max(s) represents the maximum interaction intensity between the first user and the second user.
[0164] Optionally, the addition of a submodule is specifically used for:
[0165] The association strength between the N users and the first user is detected using a preset detection algorithm;
[0166] Target users whose association strength with the first user is greater than the preset strength value are identified as the first user's friends and family group, and the target users include the second user.
[0167] Optionally, the family and friends group includes third-party users; the push module includes:
[0168] The acquisition submodule is used to acquire the first rating of the first user for the second product purchased in the past and the second rating of the third user for the first product purchased in the past, wherein the rating is used to characterize the user's preference for the product;
[0169] The first determining submodule is used to determine the similarity between the first user and the third user using the first score and the second score;
[0170] The second determining submodule is used to determine the first predicted rating of the first user for the first product based on the similarity score.
[0171] The push module is used to push information about the first product to the first user device when the first predicted score is greater than a preset score.
[0172] Optionally, the second determining submodule is specifically used for:
[0173] Based on the magnitude of the similarity, a second predicted rating for the first user on the first product is determined;
[0174] Using the second predicted score and the association between the first user and the third user, a first predicted score for the first user on the first product is determined.
[0175] Optionally, the push module is specifically used for:
[0176] The first user's behavior data on the second product purchased in history is input into a pre-trained model to obtain information about the first product. The first user's third predicted score for the first product is higher than a preset score. The third predicted score is used to characterize the first user's preference for the first product.
[0177] The information of the first product is pushed to the first user equipment, wherein the first product is a single product or a combination of multiple products of different types.
[0178] Information push devices can achieve Figure 1 The various processes implemented in the method embodiments can achieve the same technical effect, and will not be described again here to avoid repetition.
[0179] like Figure 3 As shown, this application embodiment also provides an electronic device 300, including: a processor 301, a memory 302, and a program stored in the memory 302 and executable on the processor 301. When the program is executed by the processor 301, it implements the various processes of the above-described information push method embodiment and can achieve the same technical effect. To avoid repetition, it will not be described again here.
[0180] This application also provides a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements the various processes of the above-described information push method embodiments and achieves the same technical effects. To avoid repetition, it will not be described again here. The computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, etc.
[0181] This application also provides a computer program product, including computer instructions, which, when executed by a processor, implement the above-described... Figure 1The various processes of the method embodiments shown can achieve the same technical effect, and will not be described again here to avoid repetition.
[0182] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.
[0183] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of this application.
[0184] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.
Claims
1. An information push method, characterized in that, include: When a first user equipment accesses a first network device, N user equipments that have historically accessed the first network device are identified. The first user equipment is the user equipment of the first user. The N user equipments correspond one-to-one with the N users, and N is an integer greater than 0. Obtain the association relationships between the N users and the first user, including interaction relationships; A social network graph is constructed based on the aforementioned relationships, and the family and friends group of the first user is determined based on the social network graph. The family and friends group includes at least some of the N users. Obtain the first product purchased in the user's history in the family and friends group, and push the information of the first product to the first user device.
2. The method according to claim 1, characterized in that, The N user devices include the second user device corresponding to the second user; obtaining the association relationship between the N users and the first user includes: The system acquires access data of the first user equipment and the second user equipment to the first network device, as well as interaction data between the first user and the second user. The step of constructing a social network graph based on the aforementioned relationships and determining the first user's family and friends groups based on the social network graph includes: A social network graph is constructed based on the access data and the interaction data, and the association strength between the first user and the second user is determined based on the social network graph. If the association strength is greater than a preset strength value, the second user will be added to the first user's family and friends group.
3. The method according to claim 2, characterized in that, Determining the association strength between the first user and the second user based on the social network graph includes: Based on the social network graph, the association strength between the first user and the second user is determined using the following formula: in, This indicates the strength of the association between the first user and the second user; α and β are adjustment parameters, α+β=1; This indicates the number of times the first user and the second user jointly access the first network; max(Z) represents the maximum number of times that the first user and the second user access the first network device; The expression is used to represent the interaction intensity between the first user and the second user, where max(s) represents the maximum interaction intensity between the first user and the second user.
4. The method according to claim 2, characterized in that, When the association strength is greater than a preset strength value, adding the second user to the first user's family and friends group includes: The association strength between the N users and the first user is detected using a preset detection algorithm; Target users whose association strength with the first user is greater than the preset strength value are identified as the first user's friends and family group, and the target users include the second user.
5. The method according to claim 1, characterized in that, The family and friends group includes third-party users; the step of obtaining the first product's historical purchase history from users in the family and friends group and pushing the information of the first product to the first user's device includes: The system obtains a first rating from the first user for a second product purchased in the past and a second rating from the third user for the first product purchased in the past. The ratings are used to characterize the user's preference for the product. Using the first rating and the second rating, the similarity between the first user and the third user is determined; Based on the magnitude of the similarity, a first predicted rating for the first user on the first product is determined; If the first predicted score is greater than the preset score, information about the first product is pushed to the first user device.
6. The method according to claim 5, characterized in that, Determining the first predicted rating of the first user for the first product based on the similarity score includes: Based on the magnitude of the similarity, a second predicted rating for the first user on the first product is determined; Using the second predicted score and the association between the first user and the third user, a first predicted score for the first user on the first product is determined.
7. The method according to claim 1, characterized in that, The step of pushing the information of the first product to the first user equipment includes: The first user's behavior data on the second product purchased in history is input into a pre-trained model to obtain information about the first product. The first user's third predicted score for the first product is higher than a preset score. The third predicted score is used to characterize the first user's preference for the first product. The information of the first product is pushed to the first user equipment, wherein the first product is a single product or a combination of multiple products of different types.
8. An information push device, characterized in that, include: The first acquisition module is used to determine N user devices that have historically accessed the first network device when the first user device accesses the first network device. The first user device is the user device of the first user, and the N user devices correspond one-to-one with the N users, where N is an integer greater than 0. The second acquisition module is used to acquire the association relationship between the N users and the first user, the association relationship including the interaction relationship; The determination module is used to construct a social network graph based on the association relationship, and determine the family and friends group of the first user based on the social network graph, wherein the family and friends group includes at least some of the N users; The push module is used to obtain the first product that the user in the family and friends group has purchased in the past, and push the information of the first product to the first user device.
9. An electronic device, characterized in that, include: A processor, a memory, and a program stored in the memory and executable on the processor, wherein the program, when executed by the processor, implements the steps of the information push method as described in any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the information push method as described in any one of claims 1 to 7.
11. A computer program product, characterized in that, It includes computer instructions that, when executed by a processor, implement the steps of the information push method as described in any one of claims 1 to 7.