A user preference analysis method and device for equity products

Abstract

The present application provides a user preference analysis method and device for equity products, which belongs to the field of intelligent recommendation algorithms and is used to solve the problem of low subscription rate of equity products in related technologies. The method includes obtaining behavioral feature data of a target user, wherein the behavioral feature data carries a timestamp and an equity product identifier, and the behavioral feature data includes a feature value of each behavioral feature dimension; substituting the behavioral feature data of the target user into a self-developed preference analysis model to obtain the target user's preference data for each equity product; and being able to intelligently and reasonably analyze the target user's preference data for each equity product based on the target user's behavioral feature data, thereby achieving accurate analysis of the target user's quantitative preference for equity products at a lower cost and a higher efficiency, thereby facilitating improving the subscription rate of equity products at a lower cost and a higher efficiency.

Description

Technical Field

[0001] The present application relates to the field of intelligent recommendation algorithms, and in particular to a method and device for analyzing user preferences for equity products. Background Art

[0002] Equity products are an important supplement to traditional operators' businesses. Such services can significantly increase user stickiness and loyalty. By providing additional value-added services such as video memberships, music subscriptions, etc., users are more likely to continue using the operator's services, reduce churn rates, and enhance brand loyalty.

[0003] Existing technologies generally implement vertical membership areas, such as food and travel, to enable users to enjoy privileged use of vertical benefits. However, how users become vertical members requires operators to conduct in-depth mining of users' past browsing, consumption, collection, and use of benefit products before determining whether to guide users to open vertical memberships or order vertical benefit products. The cost of operators guiding and mining users to open vertical memberships or order vertical benefit products is high and inefficient, which also directly affects the subscription rate of benefit products. Summary of the invention

[0004] The present application provides a user preference analysis method and device for equity products, which can analyze and calculate the recent behavior of existing users to determine the quantitative preferences of existing users for equity products, which is conducive to improving the subscription rate of equity products at a lower cost and higher efficiency.

[0005] In a first aspect, the present application provides a user preference analysis method for equity products. The method comprises:

[0006] Obtaining the target user's behavior characteristic data, wherein the behavior characteristic data carries a timestamp and an equity product identifier, and the behavior characteristic data includes a characteristic value of each behavior characteristic dimension;

[0007] Substitute the target user's behavioral characteristic data into the pre-built preference analysis model to obtain the target user's preference data for each equity product;

[0008] The preference analysis model includes:

[0009] ,

[0010] In the formula, Represents the preference data for an equity product, is a pre-constructed activation function, is the preset basic preference, is the preset characteristic influence coefficient, is the eigenvalue of the i-th behavioral feature dimension among the pre-constructed n feature dimensions, is the feature weight of the i-th behavioral feature dimension, The total amount obtained for the target users The j-th time influence coefficient of the behavior feature data is negatively correlated to the duration between the timestamp carried by the behavior feature data and the current moment.

[0011] By adopting the above technical solution, it is possible to intelligently and reasonably analyze the target user's preference data for each equity product based on the target user's behavioral characteristic data, thereby achieving accurate analysis of the target user's quantitative preference for equity products at a lower cost and higher efficiency, which is conducive to improving the subscription rate of equity products at a lower cost and higher efficiency.

[0012] Furthermore, the method for obtaining the time influence coefficient includes:

[0013] , where is the time influence coefficient corresponding to the behavior characteristic data, λ is the preset attenuation intensity coefficient, The duration between the timestamp carried by the behavior feature data and the current moment.

[0014] Furthermore, the method further comprises:

[0015] Obtain behavioral characteristic data of users other than the target user;

[0016] Determining the behavioral similarity between the target user and each other user based on the behavioral characteristic data;

[0017] Determine a preset number of other users with the highest behavioral similarity as similar users of the target user;

[0018] Substituting the behavioral characteristic data of similar users into the preference analysis model to obtain the preference data of similar users for each equity product;

[0019] Based on the equity product targeted, the recommendation data of the equity product to the target user is calculated according to the preference data of the target user and similar users, and the recommendation data is positively correlated with the preference data of the target user and similar users.

[0020] Further, determining the behavior similarity between the target user and each other user according to the behavior characteristic data includes:

[0021] Substituting the behavior feature data of the target user and each other user into a pre-built feature vectorization model to obtain the behavior feature vector of the target user or other users;

[0022] Calculating the behavior similarity between the target user and each other user based on the behavior feature vector;

[0023] The feature vectorization model includes:

[0024] , where is the vectorized behavioral feature data of the input target user or other users. The behavioral feature data is n-dimensional. is the behavior feature vector of the corresponding target user or other users, is the number of input behavioral feature data, is the time influence coefficient of the corresponding behavior characteristic data;

[0025] The behavior similarity is equal to the vector similarity between the behavior feature vectors of the target user and other users.

[0026] Furthermore, the target-based equity product, based on the preference data of the target user and similar users, calculates the recommendation data of the equity product to the target user, including:

[0027] Assume that for a certain equity product, the target user’s preference data is , there are u similar users, and the preference data of the i-th similar user is , the recommendation data of the equity product is z, then , where Calculate the weight of the recommendation relative to the preset target user. Calculate the weight for the pre-obtained recommendation of the i-th similar user.

[0028] Furthermore, the recommendation calculation weight of similar users is positively correlated with the behavior similarity of the target users of the similar users.

[0029] Furthermore, a specified number of equity products with the highest recommendation data are recommended to the target users.

[0030] In a second aspect, the present application provides a user preference analysis device for equity products. The device is used to execute any one of the methods described in the first aspect above.

[0031] In summary, this application at least has the following beneficial effects:

[0032] A method and device for analyzing user preferences for equity products are provided, which realizes low-cost, high-efficiency and high-accuracy analysis of target users' preferences for each equity product through the constructed intelligent model, which is conducive to accurate recommendation strategies and improving the subscription rate of equity products;

[0033] Comprehensively analyzing the recommendation data of equity products for target users in combination with similar users will facilitate more accurate recommendations.

[0034] It should be understood that the contents described in the Summary of the Invention are not intended to limit the key or important features of the embodiments of the present application, nor are they intended to limit the scope of the present application. Other features of the present application will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] The above and other features, advantages and aspects of the embodiments of the present application will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. In the accompanying drawings, the same or similar reference numerals represent the same or similar elements, wherein:

[0036] Figure 1 A flow chart of a method for analyzing user preferences for equity products in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0037] 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 creative work are within the scope of protection of this application.

[0038] In addition, the term "and / or" in this article is only a description of the association relationship between the associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0039] The present application provides a user preference analysis method and device for equity products, which can analyze user preferences for equity products at low cost, high efficiency and high accuracy, so as to improve the subscription rate of equity products.

[0040] In a first aspect, the present application provides a user preference analysis method for equity products. The method can be executed by a server.

[0041] Figure 1 A flow chart of a method for analyzing user preferences for equity products in an embodiment of the present application is shown.

[0042] Reference Figure 1 , the method specifically comprises the following steps:

[0043] S1: Obtaining the target user's behavior characteristic data, wherein the behavior characteristic data carries a timestamp and an equity product identifier, and the behavior characteristic data includes a characteristic value of each behavior characteristic dimension.

[0044] Specifically, the behavioral data of all users (including target users and other users) will be stored and recorded. The behavioral data may include the user's channels, items, behavioral content, pages visited and time, transaction history, evaluation feedback, advertising clicks, etc.

[0045] In the method of this step, based on all equity products, feature construction engineering is performed on the user's behavior data after preprocessing (preprocessing is general deduplication, type conversion, missing value processing and other public means, which are not expanded), and the user's behavior data is processed into specific behavior feature data. Each behavior feature data has n behavior feature dimensions and feature values ​​of each behavior feature dimension, and the behavior feature data carries a timestamp, which reflects the occurrence time of the corresponding behavior data of the behavior feature data. The behavior feature data also carries an equity product identifier, and the equity product identifier reflects the equity product associated with the behavior feature data.

[0046] S2: Substitute the target user’s behavioral characteristic data into the pre-built preference analysis model to obtain the target user’s preference data for each equity product.

[0047] In the method of this step, the preference analysis model includes:

[0048] ,

[0049] In the formula, Represents the preference data for an equity product, is a pre-constructed activation function, is the preset basic preference, is the preset characteristic influence coefficient, is the eigenvalue of the i-th behavioral feature dimension among the pre-constructed n feature dimensions, is the feature weight of the i-th behavioral feature dimension, The total amount obtained for the target users The j-th time influence coefficient of the behavior feature data is negatively correlated to the duration between the timestamp carried by the behavior feature data and the current moment.

[0050] The method for obtaining the time influence coefficient includes: , where is the time influence coefficient corresponding to the behavior characteristic data, λ is the preset attenuation intensity coefficient, It is the time length between the timestamp carried by the behavior feature data and the current time. The activation function is specifically selected as the Sigmoid function.

[0051] In combination with the above content, preference data can be used to directly reflect the target user's preference for equity products, and equity products can be directly recommended to the target user based on the preference data, for example, several equity products with higher preference data can be recommended to the user.

[0052] Of course, in order to more accurately analyze the target user's preference for equity products, the method also includes: obtaining behavioral feature data of other users other than the target user; determining the behavioral similarity between the target user and each other user based on the behavioral feature data; determining a preset number of other users with the highest behavioral similarity as similar users of the target user; substituting the behavioral feature data of similar users into the preference analysis model to obtain the preference data of similar users for each equity product; based on the equity product, calculating the recommendation data of the equity product for the target user according to the preference data of the target user and similar users, and the recommendation data is positively correlated with the preference data of the target user and similar users.

[0053] Specifically, determining the behavior similarity between the target user and each other user according to the behavior feature data includes: substituting the behavior feature data of the target user and each other user into a pre-built feature vector quantization model to obtain a behavior feature vector of the target user or other users; calculating the behavior similarity between the target user and each other user according to the behavior feature vector;

[0054] Wherein, the feature vectorization model includes: , where is the vectorized behavioral feature data of the input target user or other users. The behavioral feature data is n-dimensional. is the behavior feature vector of the corresponding target user or other users, is the number of input behavioral feature data, is the time influence coefficient of the corresponding behavior feature data. The specific method for obtaining the time influence coefficient here can refer to the aforementioned method for obtaining the time influence coefficient, and no repeated description will be made.

[0055] The behavior similarity is equal to the vector similarity between the behavior feature vectors of the target user and other users.

[0056] In the above content, the target equity product based on the target user and similar users' preference data to calculate the recommendation data of the equity product to the target user includes: assuming that for a equity product, the target user's preference data is , there are u similar users, and the preference data of the i-th similar user is , the recommendation data of the equity product is z, then , where Calculate the weight of the recommendation relative to the preset target user. Calculate the weight for the pre-obtained recommendation of the i-th similar user.

[0057] Among them, the recommendation calculation weight of similar users is positively correlated with the behavior similarity of the target user of the similar users. For example, let the behavior similarity between the i-th similar user and the target user be ,but , where c is the preset similarity calculation coefficient.

[0058] Based on the above content, the target user's recommendation data for each equity product can be calculated in combination with similar users. The recommendation data directly reflects the degree of recommendation of the equity product to the target user. In this way, a specified number of equity products with the highest recommendation data can be recommended to the target user, thereby achieving accurate recommendations for each target user and increasing the target user's subscription rate for equity products.

[0059] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described order of actions, because according to the embodiments of the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily required by the present application.

[0060] In a second aspect, the present application provides a user preference analysis device for equity products, which is used to execute any one of the methods disclosed in the first aspect above.

[0061] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the described device can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0062] In summary, this application at least has the following beneficial effects:

[0063] 1. A user preference analysis method and device for equity products are provided. Through the constructed intelligent model, the method realizes low-cost, high-efficiency and high-accuracy analysis of the target user's preference for each equity product, which is conducive to accurate recommendation strategies and improving the subscription rate of equity products;

[0064] 2. Comprehensively analyzing the recommendation data of equity products for target users by combining similar users will facilitate more accurate recommendations.

[0065] The above description is only a preferred embodiment of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of disclosure involved in the present application is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the aforementioned disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in this application (but not limited to) by each other to form a technical solution.

Claims

1. A user preference analysis method for equity products, characterized in that: include: Obtaining the target user's behavior characteristic data, wherein the behavior characteristic data carries a timestamp and an equity product identifier, and the behavior characteristic data includes a characteristic value of each behavior characteristic dimension; Substitute the target user's behavioral characteristic data into the pre-built preference analysis model to obtain the target user's preference data for each equity product; The preference analysis model includes: ， In the formula, Represents the preference data for an equity product, is a pre-constructed activation function, is the preset basic preference, is the preset characteristic influence coefficient, is the eigenvalue of the i-th behavioral feature dimension among the pre-constructed n feature dimensions, is the feature weight of the i-th behavioral feature dimension, The total amount obtained for the target users A time influence coefficient of the jth behavior feature data among the behavior feature data, wherein the time influence coefficient is negatively correlated to the time length between the timestamp carried by the behavior feature data and the current moment; The method for obtaining the time influence coefficient includes: , where is the time influence coefficient corresponding to the behavior characteristic data, λ is the preset attenuation intensity coefficient, The duration between the timestamp carried by the behavior feature data and the current moment; The method further comprises: Obtain behavioral characteristic data of users other than the target user; Determining the behavioral similarity between the target user and each other user based on the behavioral characteristic data; Determine a preset number of other users with the highest behavioral similarity as similar users of the target user; Substituting the behavioral characteristic data of similar users into the preference analysis model to obtain the preference data of similar users for each equity product; Based on the equity product targeted, the recommendation data of the equity product to the target user is calculated according to the preference data of the target user and similar users, and the recommendation data is positively correlated with the preference data of the target user and similar users.

2. The method according to claim 1, characterized in that Determining the behavior similarity between the target user and each other user according to the behavior characteristic data includes: Substituting the behavior feature data of the target user and each other user into a pre-built feature vectorization model to obtain the behavior feature vector of the target user or other users; Calculating the behavior similarity between the target user and each other user based on the behavior feature vector; The feature vectorization model includes: ， In the formula, is the vectorized behavioral feature data of the input target user or other users. The behavioral feature data is n-dimensional. is the behavior feature vector of the corresponding target user or other users, is the number of input behavioral feature data, is the time influence coefficient of the corresponding behavior characteristic data; The behavior similarity is equal to the vector similarity between the behavior feature vectors of the target user and other users.

3. The method according to claim 1, characterized in that: The target-based equity product, calculating the recommendation data of the equity product to the target user according to the preference data of the target user and similar users, includes: Assume that for a certain equity product, the target user’s preference data is , there are u similar users, and the preference data of the i-th similar user is , the recommendation data of the equity product is z, then , where Calculate the weight of the recommendation relative to the preset target user. Calculate the weight for the pre-obtained recommendation of the i-th similar user.

4. The method according to claim 3, characterized in that: The recommendation calculation weight of similar users is positively correlated with the behavior similarity of the target users of similar users.

5. The method according to any one of claims 1 to 4, characterized in that: The method further comprises: Recommend a specified number of equity products with the highest recommendation data to target users.

6. A user preference analysis device for equity products, characterized in that: The device is used to execute the method according to any one of claims 1 to 5.

Images