Method, device, equipment and storage medium for recommending target users

Abstract

The present application relates to a method, device, equipment and storage medium for recommending target users, wherein the recommendation method includes: obtaining the rewarded goods configured for the currently initiated invitation task; obtaining the classification result of the user; wherein the classification result includes multiple user sets and the set category features corresponding to each of the user sets; wherein the set category features are used to characterize the category features of the set of goods associated with the historical orders of each user in the user set; from each of the set category features, determining the target set category features that are closest to the features of the rewarded goods; and using each member user in the user set corresponding to the target set category features as the recommended target user. The present application is used to solve the technical problem of low efficiency in completing invitation tasks in the prior art.

Description

Technical Field

[0001] The present application relates to the field of intelligent analysis technology, and in particular to a method, device, equipment and storage medium for recommending target users. Background Art

[0002] At present, in some applications, in order to increase the stickiness between users and applications, tasks, lotteries, points, props and other modules are set in the application. The operator can create tasks in the activity platform and put them in the application, and users can participate in the tasks. After completing the task, users can get corresponding rewards, such as: increase points, free lottery opportunities, and prizes. There are many ways to play the activity platform tasks, including movie watching tasks, ordinary tasks, sign-in tasks and invitation tasks. The rich tasks and prizes can attract more users to participate. Among them, the invitation task is a type of task that allows multiple users to participate at the same time, usually used in sharing fission scenarios.

[0003] Currently, invitation tasks can be shared through WeChat and other methods to invite other users to complete the task. Both the inviter and the invitee can obtain the rewards configured on the task. However, since different users have different interests in completing tasks, sharing with uninterested users may reduce the efficiency of completing the invitation task. Summary of the invention

[0004] The present application provides a method, device, equipment and storage medium for recommending target users, so as to solve the technical problem of low efficiency in completing invitation tasks in the prior art.

[0005] In a first aspect, the present application provides a method for recommending target users, comprising:

[0006] Get the reward products configured for the currently initiated invitation task;

[0007] Obtaining a classification result of a user; wherein the classification result includes multiple user sets and a set category feature corresponding to each of the user sets; wherein the set category feature is used to characterize the category feature of the commodity set associated with the historical orders of each user in the user set;

[0008] Determine, from among the set category features, a target set category feature that is closest to the feature of the rewarded product;

[0009] Each member user in the user set corresponding to the target set category feature is used as a recommended target user.

[0010] Optionally, obtaining the classification result of the user includes:

[0011] Obtain at least one commodity associated with each of the users;

[0012] For each of the users, obtain a label vector corresponding to each of the commodities; wherein the label vector is used to characterize the category characteristics of the commodities;

[0013] For each of the users, based on the label vectors of the commodities associated with the user, a user label vector of the user is calculated; wherein the user label vector is used to characterize the overall category characteristics of all commodities associated with the user;

[0014] Based on the user label vector corresponding to each of the users, cluster analysis is performed on each of the users, and the centroid label vector corresponding to the target centroid obtained by the cluster analysis is used as the set category feature; users classified into the same target centroid are classified as one category to obtain the classification result; wherein the centroid label vector is used to characterize the category characteristics of all commodities associated with all users belonging to the same target centroid.

[0015] Optionally, obtaining the label vector corresponding to each of the commodities includes:

[0016] Obtain a label vector library; wherein the label vector library stores a preset label vector corresponding to each of the commodities;

[0017] From the label vector library, a preset label vector corresponding to each commodity associated with the user is determined, and the preset label vector is used as the label vector.

[0018] Optionally, obtaining a label vector library includes:

[0019] For each of the commodities, obtain description information of the commodity; perform word segmentation on the description information to obtain a word segmentation result; wherein the description information is used to introduce the commodity;

[0020] Extracting a target keyword from the word segmentation result, determining a preset category corresponding to the target keyword, and using the preset category corresponding to the target keyword as the category of the product;

[0021] Based on the category of the commodity, generating a preset label vector of the commodity;

[0022] The label vector library is established according to the preset label vectors of the commodities.

[0023] Optionally, performing cluster analysis on each of the users based on the user label vectors corresponding to each of the users includes:

[0024] Selecting a plurality of the user label vectors from the user label vectors corresponding to the respective users as the centroid label vectors of the respective current centroids;

[0025] Cluster analysis process: for each of the users, calculate the first distance values ​​between the user label vector and the centroid label vectors of each of the current centroids; select the smallest first distance value from each of the first distance values; divide the users into a class of the current centroids corresponding to the smallest first distance value; for each of the current centroids, use each of the user label vectors of the same class as the current centroid to update the centroid label vector of the current centroid to obtain an updated centroid label vector; if it is determined that the updated centroid label vector is different from the centroid label vector of the current centroid, repeat the cluster analysis process until the centroid label vector no longer changes, obtain each of the target centroids, and complete the cluster analysis.

[0026] Optionally, the updating the centroid label vector of the current centroid by using each of the user label vectors of the same type as the current centroid to obtain an updated centroid label vector includes:

[0027] Calculate the vector sum of each of the user label vectors of the same type as the current centroid;

[0028] Calculate the modulus of the centroid label vector corresponding to the current centroid to obtain the modulus of the centroid;

[0029] The quotient between the vector sum and the modulus of the centroid is calculated to obtain the updated centroid label vector.

[0030] Optionally, determining, from among the set category features, a target set category feature that is closest to the feature of the rewarded product includes:

[0031] Obtaining a label vector of the rewarded product;

[0032] Calculating respective second distance values ​​between the label vector of the rewarded product and each of the set category features;

[0033] The set category feature with the smallest second distance value is selected as the target set category feature.

[0034] Optionally, obtaining the label vector of the rewarded product includes:

[0035] Acquire the description information of the rewarded product, perform word segmentation on the description information of the rewarded product, and obtain the word segmentation result of the rewarded product;

[0036] Extracting a target keyword from the word segmentation result of the rewarded product, determining a preset category corresponding to the target keyword, and using the preset category corresponding to the target keyword as the category of the rewarded product;

[0037] Based on the category of the bonus product, a label vector of the bonus product is generated.

[0038] In a second aspect, the present application provides a target user recommendation device, comprising:

[0039] The first acquisition module is used to acquire the reward product configured for the currently initiated invitation task;

[0040] A second acquisition module is used to obtain the classification results of the user; wherein the classification results include multiple user sets and set category features corresponding to each of the user sets; wherein the set category features are used to characterize the category features of the commodity sets associated with the historical orders of each user in the user set;

[0041] A determination module, used to determine, from each of the set category features, a target set category feature that is closest to the feature of the rewarded product;

[0042] The recommendation module is used to take each member user in the user set corresponding to the target set category feature as a recommended target user.

[0043] In a third aspect, the present application provides an electronic device comprising: a processor, a memory and a communication bus, wherein the processor and the memory communicate with each other via the communication bus; the memory is used to store computer programs; and the processor is used to execute the programs stored in the memory to implement the target user recommendation method described in the first aspect.

[0044] In a fourth aspect, the present application provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the method for recommending target users described in the first aspect.

[0045] The above technical solution provided by the embodiment of the present application has the following advantages over the prior art: the method provided by the embodiment of the present application selects the target set category features that are closest to the features of the rewarded goods according to the rewarded goods configured for the invitation task, and uses each member user in the user set corresponding to the target set category features as the recommended target user. In the embodiment of the present application, from the perspective of the feature similarity of the rewarded goods, the features of the rewarded goods are used to match interested users from each user set, which can improve the efficiency of completing the invitation task, attract more users to participate in the invitation task, and help improve the stickiness of users to the product (e.g., application program), promote user growth, and is of great significance for the promotion and application of the product. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0047] Figure 1 A flowchart of a method for recommending target users provided in an embodiment of the present application;

[0048] Figure 2 A schematic diagram of a user classification method provided in an embodiment of the present application;

[0049] Figure 3 A schematic diagram of a target user recommendation method provided in an embodiment of the present application;

[0050] Figure 4 A flowchart of a method for recommending target users based on commodities traded by users provided in an embodiment of the present application;

[0051] Figure 5 A system block diagram of a target user recommendation method provided in an embodiment of the present application;

[0052] Figure 6 A schematic diagram of the structure of a target user recommendation device provided in an embodiment of the present application;

[0053] Figure 7 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0054] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0055] In order to solve the technical problem of low efficiency in completing invitation tasks in the prior art, the present application embodiment provides a method for recommending target users, such as Figure 1 As shown, the target user recommendation method provided in the embodiment of the present application specifically includes the following steps:

[0056] Step 101, obtaining the reward product configured for the currently initiated invitation task;

[0057] In the gameplay of invitation tasks, in order to attract users to participate in the invitation tasks, each invitation task is configured with reward products as rewards, for example, the reward products can be physical products such as household goods and food, or virtual products such as electronic coupons and free memberships. Reward products can be allocated to invitation tasks from the pre-configured product library, and the allocation method can be random allocation or allocation in a preset order.

[0058] Step 102, obtaining the classification result of the user; wherein the classification result includes multiple user sets and set category features corresponding to each user set; wherein the set category features are used to characterize the category features of the commodity set associated with the historical orders of each user in the user set;

[0059] In specific implementation, the classification of users can be based on the categories of goods obtained by users in historical tasks or commodity redemption history. The collection category feature is mainly used to characterize the category features of the goods obtained or redeemed in the historical orders of all users in the user set, and is used to characterize the tendency of the types of goods obtained or redeemed by users in the entire user set.

[0060] For ease of understanding, an example is given here. For example: when completing the invitation task, the goods redeemed by user A, user B and user C belong to the household goods category; the goods redeemed by user D, user E and user F belong to the electronic voucher category; users are classified according to the goods they have traded, then user A, user B and user C are taken as one category to obtain user set S1, and the set category feature corresponding to user set S1 is: the category of goods traded by each user in the user set S1 is the household goods category; user D, user E and user F are taken as one category to obtain user set S2, and the set category feature corresponding to user set S2 is: the category of goods traded by each user in the user set S2 is the electronic voucher category.

[0061] In addition, it should be noted that in the embodiments of the present application, for ease of understanding, the commodities traded by users are used as an example for explanation.

[0062] Step 103, determining the target set category feature that is closest to the feature of the rewarded product from among the set category features;

[0063] Taking reward products as an example, according to the features of the prize products configured for the invitation task, the closest target set category features are determined from various set category features.

[0064] For ease of understanding, an example is given here: the collection category features included in the classification results include four collection category features: collection category feature E1 corresponding to furniture products, collection category feature E2 corresponding to food products, collection category feature E3 corresponding to e-voucher products and collection category feature E4 corresponding to membership benefits products.

[0065] If the reward product configured for the current invitation task is a 7-day membership, the target set category feature that is closest to the feature of the reward product is: the set category feature E4 corresponding to the membership rights and interests products.

[0066] In the specific implementation, the label vector can be used to represent the characteristics of the product; the set category characteristics can also be represented in the form of a vector. When determining the closest target set category characteristics, the second distance value between the label vector of the rewarded product and the set category characteristics in each vector form can be calculated, and the set category characteristics with the smallest second distance value are selected as the target set category characteristics.

[0067] The use of label vectors to represent the category characteristics of the rewarded products is described in detail in the subsequent step 202, and will not be described in detail here. The use of vectors to represent the category characteristics of the set is described in detail in the cluster analysis process described in the subsequent step 203, and will not be described in detail here.

[0068] Step 104: each member user in the user set corresponding to the target set category feature is used as a recommended target user.

[0069] Continuing with the example, when it is determined that the target set category feature that is closest to the feature of the reward product is set category feature E4 corresponding to the member benefits product, each member user in the user set corresponding to set category feature E4 will be used as the recommended user of the currently initiated invitation task, that is, the target user.

[0070] In the specific implementation, after determining the target user for recommendation, the recommendation result can be displayed to the user who initiated the invitation task. The user shares the link of the invitation task one by one according to the recommendation result. For example, in the specific implementation, the link of the invitation task can be shared to the target user through WeChat.

[0071] In addition, considering the actual situation, if there are many recommended target users, sharing the links one by one will be very cumbersome. In order to simplify the user's operation, after determining the recommended target user, the invitation task is automatically sent to the account associated with the target user (for example: WeChat or application), and the invitation task is sent directly to the target user without the need for the user to share manually.

[0072] In the embodiment of the present application, according to the reward product configured for the invitation task, the target set category feature closest to the feature of the reward product is selected, and each member user in the user set corresponding to the target set category feature is used as the recommended target user. In the embodiment of the present application, from the perspective of the feature similarity of the reward product, the features of the reward product are used to match interested users from each user set, which can improve the efficiency of completing the invitation task, attract more users to participate in the invitation task, and help improve the stickiness of users to the product (for example: application program), promote user growth, and have important significance for the promotion and application of the product.

[0073] In a specific embodiment, when implementing step 102, users may be classified by a clustering algorithm, such as Figure 2 As shown, the embodiment of the present application provides a method for classifying users, which specifically includes the following steps:

[0074] Step 201, obtaining at least one commodity associated with each user;

[0075] Specifically, taking the goods obtained or exchanged by users as an example, when users participate in tasks or activities, they can obtain corresponding reward goods after completion, and can also use user points to directly exchange goods in the points mall, where the exchange or issuance of goods is processed by the props background processing. After the user receives the reward goods or uses points to exchange goods, that is, after the commodity transaction is completed, the commodity order data is generated, where the commodity order data is stored in the database. The commodity order data includes: the user's unique identity (also called user ID) and the commodity name, identification information, such as the commodity number (also called commodity ID) and the description information of the commodity.

[0076] When obtaining the products associated with the user, the product order data can be counted from the user dimension, for example: according to the user's unique identity, the products corresponding to each unique identity are counted to obtain at least one product associated with each user.

[0077] For ease of understanding, here is an example. If the user ID is 0000001, and the user previously completed a task and received a reward of a mobile phone selfie stick and small snacks, then the corresponding relationship between user 0000001 and the mobile phone selfie stick, and the corresponding relationship between user 0000001 and small snacks are stored in the product order data in the database. According to the user's unique identity information (user ID 0000001), the products (products) associated with the user are mobile phone selfie sticks and small snacks.

[0078] Step 202: for each user, obtain a label vector corresponding to each product; wherein the label vector is used to represent the category characteristics of the product;

[0079] In the specific implementation, the label vector of the product is used to represent the category characteristics of the product. Specifically, a label vector library is obtained; wherein the label vector library stores the preset label vector corresponding to each product; from the label vector library, the preset label vector corresponding to each product associated with each user is determined, and the preset label vector is used as the label vector.

[0080] When establishing a label vector library, for each product, obtain the product description information; wherein the description information is used to introduce the product; perform word segmentation on the description information to obtain the word segmentation result; extract the target keyword from the word segmentation result, determine the preset category corresponding to the target keyword, and use the preset category corresponding to the target keyword as the category of the product; based on the category of the product, generate the preset label vector of the product. According to the preset label vectors of each product, establish a label vector library.

[0081] For example, the label vector library stores the label vectors corresponding to all commodities. When constructing the label vector library, the commodities configured in the props background are summarized and the description information of each commodity is extracted. For example, the commodity to be exchanged is a 7-day free membership. The description information of the virtual commodity 7-day free membership may include: the name of the virtual commodity, the exchange conditions, and the implementation method of the 7-day free membership. The description information is segmented. For example, it contains keywords such as membership and exchange. The virtual commodity 7-day free membership is divided into membership rights and interests commodities (i.e., the category of the commodity is determined). If the preset categories include furniture, food, electronic coupons, and membership rights and interests, the preset label vector of the virtual commodity 7-day free membership is [0 0 0 1]. In the same way, the preset label vectors are determined for all traded commodities, thereby establishing a label vector library.

[0082] In the actual application process, some commodities may be assigned to two categories. For example, when determining the category of a cake voucher, it may be assigned to the food category or the electronic voucher category. After determining the label vector of the commodity, the label vector of the commodity is normalized and the normalized label vector is used as the preset label vector. The normalization process refers to dividing each component of the label vector by the modulus of the label vector, and the new label vector obtained is the normalized label vector, that is, the preset label vector.

[0083] For ease of understanding, here is an example. If the preset categories include: furniture, food, e-vouchers, and membership benefits; the cake redemption voucher can be assigned to the food category or the e-voucher category, then the label vector of the product is determined to be [0 1 1 0], and the modulus of the vector is Then the normalized label vector is

[0084] After the label vector library is established, the label vector of the commodity traded by the user can be determined by querying the label vector library.

[0085] Step 203: For each user, a user label vector of the user is calculated based on the label vectors of each commodity associated with the user; wherein the user label vector is used to represent the overall category characteristics of all commodities associated with the user;

[0086] In actual applications, the number of items associated with a user may be one or more. If the number of items associated with a user is one, the label vector of the item is used as the user label vector of the user; if the number of items associated with a user is at least two, the average value of the label vectors of at least two items is calculated and used as the user label vector of the user.

[0087] For ease of understanding, here is an example. The products redeemed by user M are kitchen paper and cake. By querying the label vector library, the label vector of kitchen paper is [1 0 0 0], and the label vector of cake is [0 1 0 0]. The average value of the label vector [1 0 0 0] of kitchen paper and the label vector [0 1 0 0] of cake is calculated, that is, [0.5 0.5 0 0], and finally the user label vector of the user is obtained.

[0088] Step 204, based on the user label vector corresponding to each user, perform cluster analysis on each user, and use the centroid label vector corresponding to the target centroid obtained by the cluster analysis as the set category feature; the users classified into the same target centroid are classified as one category to obtain the classification result; wherein the centroid label vector is used to characterize the category characteristics of all commodities associated with all users belonging to the same target centroid.

[0089] Among them, in the specific implementation, the K-means clustering algorithm can be used for clustering analysis. The K-means clustering algorithm is an unsupervised clustering algorithm with relatively simple implementation and wide application range. The principle of the K-means algorithm is to divide the sample set into K clusters according to the size of the distance, so that the similarity of the samples in the cluster is high, and the distance between the clusters is large.

[0090] Specifically in the embodiment of the present application, after querying the commodity order data of each user, the distance between each user is calculated based on the user label vector of each user, and the users are divided into multiple user sets after iterative calculation by the K-means algorithm. Users in the same user set can be considered to be relatively similar at the commodity transaction level, which can be understood as the users in the user set have similar preferences for commodities. According to the reward commodities configured for the invitation task, the most matching user set can be selected by calculating the similarity, and the member users in the set can be recommended to the inviting user. Among them, in the specific implementation, the size of the distance can be used to characterize the similarity. The smaller the distance, the higher the similarity; the larger the distance, the lower the similarity.

[0091] Specifically, in the embodiment of the present application, the method of classification using the K-means clustering algorithm is as follows:

[0092] First, multiple user label vectors are selected from the user label vectors corresponding to each user as the centroid label vectors of each current centroid.

[0093] Specifically, K samples can be randomly selected from all user lists as current centroids, where each current centroid corresponds to a centroid label vector, and each current centroid represents a class.

[0094] Cluster analysis process: For each user, calculate the first distance value between the user label vector and the centroid label vector of each current centroid; select the smallest first distance value from each first distance value; divide the user into the current centroid class corresponding to the smallest first distance value; for each current centroid, use each user label vector of the same class as the current centroid to update the centroid label vector of the current centroid to obtain an updated centroid label vector. If it is determined that the updated centroid label vector is different from the centroid label vector of the current centroid, repeat the cluster analysis process until the centroid label vector no longer changes, and obtain each target centroid. At this time, the centroid label vector corresponding to each target centroid is the set category feature, and the cluster analysis is completed.

[0095] Among them, when updating the centroid label vector of the centroid, the vector sum of each user label vector of the same type as the current centroid is calculated; the modulus of the centroid label vector corresponding to the current centroid is calculated to obtain the modulus of the centroid; the quotient between the vector sum and the modulus of the centroid is calculated to obtain the updated centroid label vector.

[0096] The embodiment of the present application utilizes the K-means clustering algorithm to divide users and find the similarity between users from the perspective of commodities, thereby recommending target users who may be more interested in the invitation task to improve the completion efficiency of the invitation task.

[0097] To facilitate understanding, further explanation is given here with reference to specific examples.

[0098] Set the number of final partitioned user sets to K, and randomly select K samples V = {v1, v2, K, v k} as the centroid, the classification corresponding to these K samples is λ={λ1,λ2,K,λ k Then, we traverse the data in the user list and calculate the distance between each user’s user label vector and the centroid label vector of each centroid. i The corresponding user label vector can be expressed as p i , center of mass v j The centroid label vector is q j . Then user u i and the centroid v j The distance between them can be expressed as d ij =||p i -q j || 2 .

[0099] In the same way, we can calculate the distance between each user and each centroid, assign the user to the centroid closest to the user, and update the centroid. j The user label vector corresponding to the user of the class centroid is updated. Where p∈λ j Indicates partition to λ j The user label vector of the user of the class centroid, p represents the user label vector corresponding to the user. Then iterate continuously until the centroid label vector of the centroid no longer changes, and obtain each target centroid. The set λ of users belonging to each target centroid is the user set after clustering.

[0100] In an embodiment of the present application, users are divided according to the similarities between the commodities traded by each user, and users with similar types of commodities traded are grouped together, thereby recommending target users who may be interested in the prize commodities configured for the invitation task, which is beneficial to improving the completion efficiency of the invitation task and attracting more matching users to participate in the activity.

[0101] In order to facilitate a deeper understanding of the target user recommendation method provided in the embodiment of the present application, Figure 3 and Figure 4 The target user recommendation method provided in the embodiment of the present application is further described. Figure 3 : shows a schematic diagram of a target user recommendation method provided by an embodiment of the present application; Figure 4 A flow chart of the target user recommendation method provided in an embodiment of the present application is shown in FIG.

[0102] Assume that the system contains m users, and use U = {u1,u2,K,u m} to indicate that u i Represents a certain user. The props backend is configured with n products C = {c1,c2,K,c n}, use c i Represents a certain product. The main implementation process is as follows:

[0103] 1. Build a label vector library, also known as a product label system:

[0104] Summarize the products configured in the props background, extract the description information of each product and perform word segmentation, and determine the category of the product based on the word segmentation. Use vectors to describe the category of the product and perform normalization to obtain the product c i The label vector x i , label vector x i Used to identify the characteristics of the product for subsequent calculations and analysis.

[0105] 2. Collect the user's product order data and map it into a product label vector (that is, the label vector corresponding to the product):

[0106] In the product order database of the props backend, we query according to the user dimension to obtain the product order list of each user's transaction. Each user's product order list is associated with at least one product, thus obtaining a set of product label vectors for each user. The product label vector can be used to identify the characteristics of the user's transaction products, which serves as the basis for cluster analysis.

[0107] 3. Cluster analysis of users:

[0108] Use K-means algorithm to cluster users. Assume that there are K users after the division. Among all users, k samples are randomly selected as the centroid, and the corresponding classification is λ = {λ1,λ2,K,λ k}, then traverse the data in the user list and calculate the distance between each user's user label vector and the centroid label vector corresponding to the centroid. i The corresponding user label vector can be expressed as p i , center of mass v j The centroid label vector is q j . Then user u i and the centroid v j The distance between them can be expressed as d ij =||p i -q j || 2 .

[0109] In the same way, we can calculate the distance between each user and each centroid, assign the user to the centroid closest to the user, and update the centroid. j The user label vector corresponding to the user of the class centroid is updated. Where p∈λ j Indicates partition to λ j The user label vector of the user of the class centroid, p represents the user label vector corresponding to the user. Then iterate continuously until the centroid label vector of the centroid no longer changes, and obtain each target centroid. The set λ of users belonging to each target centroid is the user set after clustering.

[0110] 4. Recommend users for invitation tasks based on cluster analysis results:

[0111] Invite the user to initiate the invitation task, and query the label vector of the reward product configured for the invitation task as c l , calculate c l The distance to the centroid label vector of each target centroid is calculated, and the user set where the nearest target centroid is located is selected as the recommended user list. The recommended users are presented to the invited user, and the invited user can make a selection in the task.

[0112] In addition, in order to facilitate understanding of the technical solutions provided in the embodiments of the present application, Figure 5 A block diagram of an application system of a target user recommendation method provided in an embodiment of the present application is shown.

[0113] like Figure 5 As shown, the application system mainly includes three parts: prop background, analysis system and activity middle platform; among them, the prop background includes: product data storage module, product label processing module, user product order processing module and network communication module; the analysis system includes: cluster analysis module, user information storage module and network communication module; the activity middle platform includes: task creation module, task configuration information storage module and network communication module.

[0114] Among them, the commodity data storage module is used to store the commodity information configured in the props background;

[0115] The product label processing module is used to perform text analysis based on the product description information stored in the props background, build a product label system through word segmentation and classification, obtain the label vector of each product, and perform normalization processing.

[0116] User product order storage module: stores product orders generated by users receiving product rewards in various activities and tasks and directly redeeming them.

[0117] Cluster analysis module: Filter product orders according to user dimensions, obtain product information traded by each user, and map it to product label vectors. Use K-means clustering algorithm to cluster users according to the similarity of product label vectors to obtain the divided user set. Find the most matching user set based on the reward products configured in the invitation task, and return the recommended users to the APP.

[0118] The user information storage module is used to store user information; for example: user name, account number, password, gender, etc.

[0119] Task creation module: used to create various types of tasks and configure and deliver them.

[0120] Task configuration information storage module: stores configured task information and displays configured tasks to users.

[0121] Network communication module: used for network communication between various systems.

[0122] Based on the same concept, a target user recommendation device is provided in the embodiment of the present application. The specific implementation of the device can be found in the description of the method embodiment part, and the repeated parts will not be repeated. Figure 6 As shown, the device mainly includes:

[0123] The first acquisition module 601 is used to acquire the reward commodity configured for the currently initiated invitation task;

[0124] The second acquisition module 602 is used to obtain the classification result of the user; wherein the classification result includes multiple user sets and the set category feature corresponding to each of the user sets; wherein the set category feature is used to characterize the category feature of the commodity set associated with the historical order of each user in the user set;

[0125] A determination module 603 is used to determine, from each of the set category features, a target set category feature that is closest to the feature of the rewarded product;

[0126] The recommendation module 604 is configured to take each member user in the user set corresponding to the target set category feature as a recommended target user.

[0127] In a specific embodiment, the second acquisition module 602 is used to obtain at least one commodity associated with each of the users; for each of the users, obtain a label vector corresponding to each of the commodities; wherein the label vector is used to characterize the category characteristics of the commodities; for each of the users, calculate the user label vector of the user based on the label vectors of the commodities associated with the user; wherein the user label vector is used to characterize the overall category characteristics of all commodities associated with the user; based on the user label vectors corresponding to each of the users, perform cluster analysis on each of the users, and use the centroid label vector corresponding to the target centroid obtained by the cluster analysis as the set category characteristic; classify the users classified to the same target centroid as one class to obtain the classification result; wherein the centroid label vector is used to characterize the category characteristics of all commodities associated with all users belonging to the same target centroid.

[0128] In a specific embodiment, the second acquisition module 602 is used to obtain a label vector library; wherein the label vector library stores a preset label vector corresponding to each of the products; from the label vector library, determine the preset label vector corresponding to each of the products associated with each of the users, and use the preset label vector as the label vector.

[0129] In a specific embodiment, the second acquisition module 602 is used to obtain the description information of the product for each of the products; wherein the description information is used to introduce the product; the description information is segmented to obtain a segmentation result; from the segmentation result, a target keyword is extracted, and a preset category corresponding to the target keyword is determined, and the preset category corresponding to the target keyword is used as the category of the product; based on the category of the product, a preset label vector of the product is generated; and the label vector library is established according to the preset label vectors of each of the products.

[0130] In a specific embodiment, the second acquisition module 602 is used to select multiple user label vectors from the user label vectors corresponding to each of the users as the centroid label vectors of each current centroid; cluster analysis process: for each of the users, calculate the first distance values ​​between the user label vector and the centroid label vectors of each current centroid; select the minimum first distance value from each of the first distance values; divide the users into a class of the current centroids corresponding to the minimum first distance value; for each of the current centroids, use each of the user label vectors of the same class as the current centroid to update the centroid label vector of the current centroid to obtain an updated centroid label vector; if it is determined that the updated centroid label vector is different from the centroid label vector of the current centroid, repeat the cluster analysis process until the centroid label vector no longer changes, obtain each of the target centroids, and complete the cluster analysis.

[0131] In a specific embodiment, the second acquisition module 602 is used to calculate the vector sum of each of the user label vectors of the same type as the current centroid; calculate the modulus of the centroid label vector corresponding to the current centroid to obtain the modulus of the centroid; calculate the quotient between the vector sum and the modulus of the centroid to obtain the updated centroid label vector.

[0132] In a specific embodiment, the determination module 603 is used to obtain the label vector of the reward product; calculate the second distance value between the label vector of the reward product and each of the set category features; and select the set category feature with the smallest second distance value as the target set category feature.

[0133] In a specific embodiment, the determination module 603 is specifically used to obtain the description information of the reward product, segment the description information of the reward product, and obtain the segmentation result of the reward product; extract the target keyword from the segmentation result of the reward product, determine the preset category corresponding to the target keyword, and use the preset category corresponding to the target keyword as the category of the reward product; based on the category of the reward product, generate a label vector of the reward product.

[0134] In the embodiment of the present application, according to the reward product configured for the invitation task, the target set category feature closest to the feature of the reward product is selected, and each member user in the user set corresponding to the target set category feature is used as the recommended target user. In the embodiment of the present application, from the perspective of the feature similarity of the reward product, the features of the reward product are used to match interested users from each user set, which can improve the efficiency of completing the invitation task, attract more users to participate in the invitation task, and help improve the stickiness of users to the product (for example: application program), promote user growth, and have important significance for the promotion and application of the product.

[0135] Based on the same concept, an electronic device is also provided in the embodiment of the present application, such as Figure 7 As shown, the electronic device mainly includes: a processor 701, a memory 702 and a communication bus 703, wherein the processor 701 and the memory 702 communicate with each other through the communication bus 703. The memory 702 stores a program executable by the processor 701, and the processor 701 executes the program stored in the memory 702 to implement the following steps:

[0136] Get the reward products configured for the currently initiated invitation task;

[0137] Obtaining a classification result of a user; wherein the classification result includes multiple user sets and a set category feature corresponding to each of the user sets; wherein the set category feature is used to characterize the category feature of the commodity set associated with the historical orders of each user in the user set;

[0138] Determine, from among the set category features, a target set category feature that is closest to the feature of the rewarded product;

[0139] Each member user in the user set corresponding to the target set category feature is used as a recommended target user.

[0140] The communication bus 703 mentioned in the above electronic device can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. The communication bus 703 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 7 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0141] The memory 702 may include a random access memory (RAM) or a non-volatile memory, such as at least one disk memory. Alternatively, the memory may also be at least one storage device located away from the processor 701.

[0142] The above-mentioned processor 701 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc., and can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components.

[0143] In another embodiment of the present application, a computer-readable storage medium is provided, in which a computer program is stored. When the computer program is executed on a computer, the computer executes the method for recommending target users described in the above embodiment.

[0144] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer instruction is loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network or other programmable device. The computer instruction can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instruction is transmitted from a website site, a computer, a server or a data center by wired (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, microwave, etc.) mode to another website site, computer, server or data center. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server, a data center, etc. that contains one or more available media integration. The available medium can be a magnetic medium (such as a floppy disk, a hard disk, a magnetic tape, etc.), an optical medium (such as a DVD) or a semiconductor medium (such as a solid-state hard disk), etc.

[0145] It should be noted that, in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0146] The foregoing is merely a specific embodiment of the present invention, which enables those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features claimed herein.

Claims

1. A method for recommending target users, characterized in that: include: Get the reward products configured for the currently initiated invitation task; Obtaining a classification result of a user; wherein the classification result includes multiple user sets and a set category feature corresponding to each of the user sets; wherein the set category feature is used to characterize the category feature of the commodity set associated with the historical orders of each user in the user set; The obtaining of the classification result of the user comprises: performing cluster analysis on each of the users based on the user label vectors corresponding to each of the users, and taking the centroid label vector corresponding to the target centroid obtained by the cluster analysis as the set category feature; taking the users classified to the same target centroid as one category, and obtaining the classification result; wherein the centroid label vector is used to characterize the category features of all commodities associated with all users belonging to the same target centroid; The clustering analysis of each user based on the user label vector corresponding to each user includes: selecting multiple user label vectors from the user label vectors corresponding to each user as the centroid label vectors of each current centroid; clustering analysis process: for each user, calculating the first distance value between the user label vector and the centroid label vector of each current centroid; selecting the smallest first distance value from each of the first distance values; dividing the users into a class of the current centroids corresponding to the smallest first distance value; for each current centroid, using each of the user label vectors of the same class as the current centroid to update the centroid label vector of the current centroid to obtain an updated centroid label vector; if it is determined that the updated centroid label vector is different from the centroid label vector of the current centroid, repeating the clustering analysis process until the centroid label vector no longer changes, obtaining each of the target centroids, and completing the clustering analysis; Determine, from among the set category features, a target set category feature that is closest to the feature of the rewarded product; Each member user in the user set corresponding to the target set category feature is used as a recommended target user.

2. The method for recommending target users according to claim 1, characterized in that: The obtaining of the classification result of the user includes: Obtain at least one commodity associated with each of the users; For each of the users, obtain a label vector corresponding to each of the commodities; wherein the label vector is used to characterize the category characteristics of the commodities; For each of the users, based on the label vectors of the commodities associated with the user, a user label vector of the user is calculated; wherein the user label vector is used to characterize the overall category characteristics of all commodities associated with the user; Based on the user label vector corresponding to each of the users, cluster analysis is performed on each of the users, and the centroid label vector corresponding to the target centroid obtained by the cluster analysis is used as the set category feature; users classified into the same target centroid are classified as one category to obtain the classification result; wherein the centroid label vector is used to characterize the category characteristics of all commodities associated with all users belonging to the same target centroid.

3. The method for recommending target users according to claim 2, characterized in that: The obtaining of the label vector corresponding to each of the commodities includes: Obtain a label vector library; wherein the label vector library stores a preset label vector corresponding to each of the commodities; From the label vector library, a preset label vector corresponding to each commodity associated with the user is determined, and the preset label vector is used as the label vector.

4. The method for recommending target users according to claim 3, characterized in that: The step of obtaining a label vector library includes: For each of the commodities, obtain description information of the commodity; perform word segmentation on the description information to obtain a word segmentation result; wherein the description information is used to introduce the commodity; Extracting a target keyword from the word segmentation result, determining a preset category corresponding to the target keyword, and using the preset category corresponding to the target keyword as the category of the product; Based on the category of the commodity, generating a preset label vector of the commodity; The label vector library is established according to the preset label vectors of the commodities.

5. The method for recommending target users according to claim 1, characterized in that: The updating of the centroid label vector of the current centroid by using each of the user label vectors of the same type as the current centroid to obtain an updated centroid label vector includes: Calculate the vector sum of each of the user label vectors of the same type as the current centroid; Calculate the modulus of the centroid label vector corresponding to the current centroid to obtain the modulus of the centroid; The quotient between the vector sum and the modulus of the centroid is calculated to obtain the updated centroid label vector.

6. The method for recommending target users according to claim 3, characterized in that: The step of determining, from among the set category features, a target set category feature that is closest to the feature of the rewarded product, comprises: Obtaining a label vector of the rewarded product; Calculating respective second distance values ​​between the label vector of the rewarded product and each of the set category features; The set category feature with the smallest second distance value is selected as the target set category feature.

7. A target user recommendation device, characterized in that: include: The first acquisition module is used to acquire the reward product configured for the currently initiated invitation task; The second acquisition module is used to obtain the classification results of users; wherein the classification results include multiple user sets and set category features corresponding to each of the user sets; wherein the set category features are used to characterize the category features of the set of goods associated with the historical orders of each user in the user set; the acquisition of the classification results of users includes: based on the user label vectors corresponding to each of the users, clustering analysis is performed on each of the users, and the centroid label vector corresponding to the target centroid obtained by the clustering analysis is used as the set category feature; users classified into the same target centroid are classified as one category to obtain the classification results; wherein the centroid label vector is used to characterize the category features of all goods associated with all users belonging to the same target centroid; the clustering analysis of each of the users based on the user label vectors corresponding to each of the users includes: From the user label vectors corresponding to each of the users, select multiple user label vectors as the centroid label vectors of each current centroid; cluster analysis process: for each of the users, calculate the first distance values ​​between the user label vector and the centroid label vectors of each current centroid; from each of the first distance values, select the smallest first distance value; divide the users into a class of the current centroids corresponding to the smallest first distance value; for each current centroid, use each of the user label vectors of the same class as the current centroid to update the centroid label vector of the current centroid to obtain an updated centroid label vector; if it is determined that the updated centroid label vector is different from the centroid label vector of the current centroid, repeat the cluster analysis process until the centroid label vector no longer changes, obtain each of the target centroids, and complete the cluster analysis; A determination module, used to determine, from each of the set category features, a target set category feature that is closest to the feature of the rewarded product; The recommendation module is used to take each member user in the user set corresponding to the target set category feature as a recommended target user.

8. An electronic device, characterized in that: include: A processor, a memory and a communication bus, wherein the processor and the memory communicate with each other via the communication bus; the memory is used to store computer programs; and the processor is used to execute the programs stored in the memory to implement the target user recommendation method described in any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method for recommending a target user according to any one of claims 1 to 6 is implemented.

Images