Recommendation method and device based on portrait similarity, electronic equipment and storage medium

Abstract

The invention provides a recommendation method and device based on portrait similarity, electronic equipment and a storage medium, the method is applied to the field of data processing, and the method comprises the following steps: based on a first user feature of a target user, determining a target user set to which the target user belongs, the target user set is a set in which the similarity between collective user features in the target user set and first user features of target users is greater than a similarity threshold, and the collective user features are used for describing basic attributes of a plurality of candidate users in the target user set; the first user feature of the target user and the set feature of the target user set are spliced to obtain the target user feature of the target user, the set feature is used for describing the preferences of the multiple candidate users to the vehicle-mounted equipment, and the target user feature is used for describing the preferences of the target user to the vehicle-mounted equipment; and recommending the vehicle-mounted equipment to the target user based on the target user features. The method can relieve the cold start problem of the target user based on the target user set.

Description

Technical Field

[0001] The present application relates to the field of data processing, and more specifically, to a recommendation method, device, electronic device and storage medium based on portrait similarity in the field of data processing. Background Art

[0002] With the explosive growth of Internet information, personalized recommendations are playing an increasingly important role in e-commerce, social networks, content distribution and other fields. However, traditional personalized recommendations based on user needs cannot meet the diverse needs of users. For example, in the off-road vehicle modification equipment market, when consumers choose modification equipment, they will face the problem of insufficient equipment selection diversity and untimely recommended equipment updates.

[0003] In the related art, the construction and generation of user portraits based on the user's historical data has been widely used. However, when the amount of user historical data is too small or the user has no historical data, the user portrait generated based on the user's historical data is unreliable, and therefore it is impossible to provide users with more accurate personalized recommendation services, which brings a bad experience to users. Summary of the invention

[0004] The present application provides a recommendation method, device, electronic device and storage medium based on portrait similarity. The method can recommend vehicle-mounted equipment preferred by the target user to the target user based on the set characteristics of the target user set to which the target user belongs, thereby alleviating the cold start problem of the target user.

[0005] In a first aspect, a recommendation method based on portrait similarity is provided, the method comprising:

[0006] Obtaining a first user feature of a target user, where the first user feature is used to describe a basic attribute of the target user;

[0007] Based on the first user feature of the target user, determine the target user set to which the target user belongs, the target user set being a set in which the similarity between the collective user features in the target user set and the first user feature of the target user is greater than a similarity threshold, the collective user features being used to describe basic attributes of multiple candidate users in the target user set;

[0008] The first user feature of the target user and the set feature of the target user set are concatenated to obtain the target user feature of the target user, the set feature is used to describe the preference of multiple candidate users in the target user set for the vehicle-mounted equipment, and the target user feature is used to describe the preference of the target user for the vehicle-mounted equipment;

[0009] Recommend vehicle-mounted equipment to target users based on their characteristics.

[0010] Through the above technical solution, the first user feature of the target user is obtained, and the first user feature is used to describe the basic attributes of the target user. Since the target user does not have data corresponding to the vehicle-mounted equipment, the target user set to which the target user belongs is determined based on the first user feature of the target user. The target user set is a set whose similarity between the collective user feature in the target user set and the first user feature of the target user is greater than the similarity threshold. The collective user feature is used to describe the basic attributes of multiple candidate users in the target user set. The first user feature of the target user and the collective feature of the target user set are spliced ​​to obtain the target user feature of the target user. The collective feature is used to describe the preference of multiple candidate users in the target user set for the vehicle-mounted equipment. The target user feature is used to describe the preference of the target user for the vehicle-mounted equipment, and the vehicle-mounted equipment is recommended to the target user based on the target user feature. That is, even if the target user does not have data corresponding to the vehicle-mounted equipment, the target user set whose similarity between the collective feature and the first user feature of the target user is greater than the similarity threshold can still be recommended to the target user based on the target user set whose preference for the vehicle-mounted equipment is greater than the similarity threshold, thereby recommending the vehicle-mounted equipment to the target user in a personalized manner.

[0011] In combination with the first aspect, in some possible implementations, multiple candidate user sets are obtained;

[0012] Determine the similarity between the collective user features in the plurality of candidate user sets and the first user feature of the target user;

[0013] A user set whose similarity between the collective user features in the plurality of candidate user sets and the first user feature of the target user is greater than a similarity threshold is determined as the target user set to which the target user belongs.

[0014] In conjunction with the first aspect, in some possible implementations, a similarity between a first user feature of a target user and second user features of a plurality of candidate users is determined, where the second user feature of each candidate user is used to describe a basic attribute of each candidate user;

[0015] Determine the candidate users whose second user characteristics among the multiple candidate users have a similarity with the first user characteristics of the target user greater than a similarity threshold as similar users of the target user;

[0016] A user set consisting of multiple similar users of the target user is determined as the target user set to which the target user belongs.

[0017] In combination with the first aspect, in some possible implementations, second user features of multiple candidate users and behavior data of multiple candidate users with respect to vehicle-mounted equipment are obtained, where the behavior data of the candidate users with respect to the vehicle-mounted equipment is data generated by the candidate users operating a page displaying the vehicle-mounted equipment;

[0018] Processing the second user features of the plurality of candidate users and the behavior data of the plurality of candidate users with respect to the vehicle-mounted equipment to obtain the preference levels of the plurality of candidate users with respect to the vehicle-mounted equipment and the user portraits of the plurality of candidate users;

[0019] The set features of the target user set are generated based on the preference levels of multiple candidate users for the vehicle-mounted equipment and the user portraits of the multiple candidate users.

[0020] In combination with the first aspect, in some possible implementations, the second user features of the plurality of candidate users and the behavior data of the plurality of candidate users with respect to the vehicle-mounted equipment are clustered to obtain the preference levels of the plurality of candidate users with respect to the vehicle-mounted equipment;

[0021] Determining first weight values ​​of second user characteristics of a plurality of candidate users, the first weight values ​​being used to indicate the influence of the second user characteristics on the preference degree of the in-vehicle equipment;

[0022] Calculating second weight values ​​of the preference of the plurality of candidate users for the vehicle-mounted equipment based on the generation time of the first weight value, the second weight value being used to indicate the preference of the plurality of candidate users for the vehicle-mounted equipment over time;

[0023] Calculating an average of the second weight values ​​of the preference of the plurality of candidate users for the vehicle-mounted equipment to obtain third weight values ​​of the user portraits of the plurality of candidate users, where the third weight values ​​are used to describe the credibility of the user portraits of the plurality of candidate users;

[0024] User portraits of the multiple candidate users are generated based on the first weight value, the second weight value, the third weight value, the second user features of the multiple candidate users, and the preference levels of the multiple candidate users for the vehicle-mounted equipment.

[0025] In combination with the first aspect, in some possible implementations, based on the target user feature, determining whether there is a target reference user among the multiple reference users, the target reference user being a reference user whose user feature has a similarity with the target user feature that is greater than or equal to a similarity threshold;

[0026] When there is a target reference user among the multiple reference users, the vehicle-mounted equipment preferred by the target reference user is recommended to the target user.

[0027] In combination with the first aspect, in some possible implementations, when there is no target reference user among multiple reference users, the target vehicle equipment is determined from a set of vehicle equipment preferred by multiple similar users in the target user set based on the target user characteristics;

[0028] Recommend target vehicle-mounted equipment to target users.

[0029] In a second aspect, an electronic device is provided, comprising a memory and a processor. The memory is used to store executable program code, and the processor is used to call and run the executable program code from the memory, so that the electronic device executes the method executed by the above-mentioned recommendation method based on portrait similarity.

[0030] In a third aspect, a computer program product is provided, which includes: a computer program code, which, when executed on a computer, enables the computer to execute the method executed by the above-mentioned recommendation method based on portrait similarity.

[0031] In a fourth aspect, a computer-readable storage medium is provided, which stores a computer program code. When the computer program code runs on a computer, the computer executes the method executed by the above-mentioned recommendation method based on portrait similarity. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a schematic diagram of an implementation environment of a recommendation method based on portrait similarity provided in an embodiment of the present application;

[0033] Figure 2 is a schematic flow chart of a recommendation method based on portrait similarity provided in an embodiment of the present application;

[0034] Figure 3 is a schematic flow chart of another recommendation method based on portrait similarity provided in an embodiment of the present application;

[0035] Figure 4 is a schematic diagram of describing collective characteristic information of a target user set provided by an embodiment of the present application;

[0036] Figure 5 is a schematic flow chart of a method for determining a target user set provided in an embodiment of the present application;

[0037] Figure 6 is a schematic diagram of a user portrait of a candidate user provided in an embodiment of the present application;

[0038] Figure 7 It is a structural schematic diagram of a recommendation device based on portrait similarity provided in an embodiment of the present application;

[0039] Figure 8 It is a structural schematic diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0040] The technical solution in the present application will be described clearly and in detail below in conjunction with the accompanying drawings. In the description of the embodiments of the present application, unless otherwise specified, " / " means or, for example, A / B can mean A or B: "and / or" in the text is only a description of the association relationship of associated objects, indicating that there can be three relationships, for example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, in the description of the embodiments of the present application, "multiple" means two or more than two.

[0041] In the following, the terms "first" and "second" are used for descriptive purposes only and are not to be understood as suggesting or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features.

[0042] In the off-road vehicle modification equipment market, users have a growing demand for personalized vehicle equipment. Users often face difficulties in finding suitable modification equipment. Therefore, it is particularly important to improve the recommendation method for off-road vehicle modification equipment.

[0043] In the related art, the method of implementing recommendation based on the analysis of the user's historical data can more accurately determine the user's user portrait than the traditional recommendation method, and fully provide personalized services for the user. However, in the field of off-road vehicle modification equipment, there is currently little data on the research of users who modify off-road vehicles. Therefore, conducting research from the perspective of understanding the consumer's image and modification preferences will achieve twice the result with half the effort. However, since new users do not have historical data or the new users have less historical data, it is difficult to provide new users with more accurate personalized recommendation services, insufficient equipment selection diversity, and untimely updates of recommended information.

[0044] Figure 1 This is a schematic diagram of the implementation environment of a recommendation method based on portrait similarity provided in an embodiment of the present application.

[0045] For example, Figure 1 As shown, the implementation environment includes a server 101 and a terminal 102 .

[0046] The server 101 is a cloud server for providing cloud database, cloud computing and cloud communication. For example, the server 101 is a cloud server for collecting user preferences for vehicle-mounted equipment. In some embodiments, a user portrait generation model is deployed on the server 101, and the user portrait generation model is used to generate a user portrait corresponding to the user.

[0047] The terminal 102 is a device for users to interact with the computer system. In some embodiments, the terminal 102 is used to instruct the user to input data related to the vehicle-mounted equipment. In some embodiments, the terminal 102 is connected to the server 101, and the terminal 102 sends the data related to the vehicle-mounted equipment to the server 101, and the server 101 performs corresponding operations based on the data. For example, the user's rating and purchase of the vehicle-mounted equipment generates historical data corresponding to the user, and the server 101 analyzes the historical data to generate a user portrait corresponding to the user.

[0048] Figure 2 It is a schematic flowchart of a recommendation method based on portrait similarity provided in an embodiment of the present application.

[0049] For example, Figure 2 As shown, taking the execution subject as a server as an example, the method 200 includes:

[0050] Step 201: Obtain a first user feature of a target user, where the first user feature is used to describe a basic attribute of the target user.

[0051] The target user is a user with little relevant behavior data for the vehicle-mounted equipment, or the target user is a new user. The first user feature is used to classify the target user, and the first user feature includes information such as the target user's gender, age, occupation, income, education, marital status, and place of residence.

[0052] It should be understood that the target user is classified into multiple types of target users based on the first user feature. For example, if the first user feature of the target user is that the target user is male and 20 years old, then the target user is determined to be a young male based on the first user feature, thereby facilitating the determination of the young male's preference for vehicle-mounted equipment.

[0053] Step 202: Based on the first user feature of the target user, determine the target user set to which the target user belongs. The target user set is a set in which the similarity between the collective user features in the target user set and the first user feature of the target user is greater than a similarity threshold. The collective user features are used to describe basic attributes of multiple candidate users in the target user set.

[0054] It should be understood that since the target user has less data corresponding to the vehicle-mounted equipment, or the target user is a new user, it is impossible to recommend the vehicle-mounted equipment that meets the target user's needs to the target user based on the target user's historical behavior data for the vehicle-mounted equipment. In order to bring a better personalized service experience to the target user, based on the first user feature of the target user, it is determined whether there is a target user set similar to the target user.

[0055] The target user set includes multiple candidate users, and the similarity between the second user features of the multiple candidate users and the first user features of the target user is greater than the similarity threshold. The set user features include the second user features of multiple candidate users. When the similarity between the collective user features in the target user set and the first user features of the target user is greater than the similarity threshold, the target user is determined to be similar to multiple candidate users in the target user. For example, the set user features include multiple age feature vectors of males aged 18-25. When the target user is a 20-year-old male, the similarity between the age feature vector of the first user feature of the target user and the age feature vector in the target user set to which the target user belongs is determined. If the target user set includes males aged 18-25, it is determined that the similarity between the first user feature of the target user and the collective user feature vector of the target user set to which the target user belongs is greater than the similarity threshold, thereby determining the target user set to which the target user belongs.

[0056] It should be understood that the similarity threshold is a threshold automatically determined by the server. The embodiment of the present application does not limit the similarity threshold. For example, the similarity threshold may be 0.8.

[0057] Step 203, concatenate the first user feature of the target user and the set feature of the target user set to obtain the target user feature of the target user, the set feature is used to describe the preference degree of multiple candidate users in the target user set for the vehicle-mounted equipment, and the target user feature is used to describe the preference degree of the target user for the vehicle-mounted equipment.

[0058] Among them, since the target user has less data or no data related to the vehicle-mounted equipment, and the target user set includes the preferences of multiple candidate users for the vehicle-mounted equipment, the first user feature of the target user and the set feature in the target user set are spliced ​​to obtain the target user feature of the target user, thereby obtaining the target user's preference for the vehicle-mounted equipment.

[0059] It should be understood that the splicing of the first user feature of the target user and the set feature in the target user set is to merge the first user feature with the set feature. For example, the first user feature is used to describe the basic attributes of the target user, and the set feature is used to describe the preferences of multiple candidate users for vehicle-mounted equipment. The first user feature and the set feature are horizontally spliced ​​to generate a complete data set including the first user feature and the preferences of the target user for vehicle-mounted equipment.

[0060] Step 204: recommending vehicle-mounted equipment to the target user based on the target user's characteristics.

[0061] The target user characteristics are used to describe the target user's preference for vehicle-mounted equipment. Vehicle-mounted equipment is equipment used to modify off-road vehicles. It should be understood that since the target user characteristics describe the target user's preference for vehicle-mounted equipment, the target user is recommended vehicle-mounted equipment that the target user may like based on the target user characteristics, with high accuracy.

[0062] The embodiment of the present application provides a recommendation method based on portrait similarity, which obtains the first user feature of the target user, and the first user feature is used to describe the basic attributes of the target user. Since the target user does not have data corresponding to the vehicle-mounted equipment, the target user set to which the target user belongs is determined based on the first user feature of the target user. The target user set is a set whose similarity between the collective user feature in the target user set and the first user feature of the target user is greater than the similarity threshold. The collective user feature is used to describe the basic attributes of multiple candidate users in the target user set. The first user feature of the target user and the collective feature of the target user set are spliced ​​to obtain the target user feature of the target user. The collective feature is used to describe the preference of multiple candidate users in the target user set for the vehicle-mounted equipment. The target user feature is used to describe the preference of the target user for the vehicle-mounted equipment, and the vehicle-mounted equipment is recommended to the target user based on the target user feature. That is, even if the target user does not have data corresponding to the vehicle-mounted equipment, the target user set whose similarity between the collective feature and the first user feature of the target user is greater than the similarity threshold can still recommend the target user the preference of the target user for the vehicle-mounted equipment, thereby recommending the vehicle-mounted equipment to the target user in a personalized manner.

[0063] Figure 3 It is a schematic flowchart of another recommendation method based on portrait similarity provided in an embodiment of the present application.

[0064] It should be noted that the above steps 201-203 are a simple description of a recommendation method based on portrait similarity provided in an embodiment of the present application. The following will provide a more detailed description of a recommendation method based on portrait similarity provided in an embodiment of the present application in combination with some examples. Figure 3 Taking the execution subject as a server as an example, the method includes the following steps.

[0065] Step 301: Acquire a first user feature of a target user, where the first user feature is used to describe a basic attribute of the target user.

[0066] The basic attributes are information used to classify target users.

[0067] For example, if the first user feature of the target user includes that the target user is male and is 20 years old, the target user is classified as a young male based on the first user feature. The server determines the preference of the young male group for vehicle-mounted equipment by analyzing multiple young males.

[0068] It can be understood that the specific implementation of obtaining the first user feature of the target user can refer to the related description of the aforementioned step 201.

[0069] In some embodiments, attribute information of the target user is obtained based on the page displaying the vehicle-mounted equipment.

[0070] For example, the page displaying the vehicle-mounted equipment is a page for selling the vehicle-mounted equipment, and the target user can submit the attribute information corresponding to the target user on the page displaying the vehicle-mounted equipment.

[0071] In some embodiments, the attribute information corresponding to the target user is analyzed to obtain a first user feature corresponding to the target user.

[0072] In this implementation, a first user feature of the target user is obtained, and the first user feature is used to describe the basic attributes of the target user, that is, based on the attribute information corresponding to the target user, the category of the target user is determined to facilitate determining candidate users similar to the target user.

[0073] Step 302: Based on the first user feature of the target user, determine the target user set to which the target user belongs. The target user set is a set in which the similarity between the collective user features in the target user set and the first user feature of the target user is greater than a similarity threshold. The collective user features are used to describe basic attributes of multiple candidate users in the target user set.

[0074] It should be understood that since the target user does not have data corresponding to the vehicle-mounted equipment, it is impossible to determine the target user's preference for the vehicle-mounted equipment based on the data. In the case of determining the first user feature of the target user, multiple candidate users similar to the target user are determined based on the first user feature of the target user.

[0075] In a possible implementation, multiple candidate user sets are obtained; similarities between collective user features in the multiple candidate user sets and a first user feature of a target user are determined; and a user set whose similarity between the collective user features in the multiple candidate user sets and the first user feature of the target user is greater than a similarity threshold is determined as the target user set to which the target user belongs.

[0076] The multiple candidate user sets are used to determine the target user set to which the target user belongs. The target user set is a user set whose similarity between the set user feature and the first user feature is greater than a similarity threshold.

[0077] In some embodiments, the similarity between the set user features in the plurality of candidate user sets and the first user feature of the target user is determined based on modified cosine similarity.

[0078] The corrected cosine similarity is used to correct the deviation caused by the size difference between the first user feature and the set user feature by removing the mean.

[0079] In some embodiments, the first user feature is converted into a first numerical vector, the first numerical vector is encoded as a one-hot encoding (One-Hot Encoding), and a feature vector corresponding to the first user feature including multiple dimensions is generated.

[0080] In some embodiments, the collective user features in the plurality of candidate user sets are converted into a second numerical vector, the second numerical vector is encoded as a one-hot encoding, and a feature vector corresponding to the collective user features and including multiple dimensions is generated.

[0081] It should be understood that cosine similarity is a commonly used vector similarity measurement method, which calculates the cosine value of the angle between the first user feature and the set user feature, and the cosine value of the angle is between [-1,1]. If the cosine value of the angle is closer to 1, it means that the first user feature and the set user feature are more similar. If the cosine value of the angle is closer to 0, it means that the first user feature and the set user feature are less related. If the cosine value of the angle is a negative value, it means that the feature vector of the first user feature is in the opposite direction to the feature vector of the set user feature.

[0082] In some embodiments, when the similarity between the first user feature and the collective user features in multiple candidate user sets is greater than a similarity threshold, the user set in which the similarity between the collective user features in the multiple candidate user sets and the first user feature of the target user is greater than the similarity threshold is determined as the target user set to which the target user belongs.

[0083] For example, when the similarity threshold is 0.8 and the similarity between the first user feature and the collective user features in multiple candidate user sets is greater than 0.8, the user set whose similarity between the collective user features in the multiple candidate user sets and the first user feature of the target user is greater than the similarity threshold is determined as the target user set to which the target user belongs.

[0084] In this implementation manner, the similarity between the set user features in the plurality of candidate user sets and the first user feature of the target user is determined based on the modified cosine similarity, with high accuracy.

[0085] In some embodiments, the similarity between the set user features in the plurality of candidate user sets and the first user feature of the target user is determined by Jaccard similarity.

[0086] It should be understood that the Jaccard similarity is used to determine the similarity of Boolean data, wherein the Boolean data is used to represent two logical states of true or false.

[0087] In some embodiments, if the characteristic vector between the first user feature and the collective user features in the plurality of candidate user sets includes only two possible states, the characteristic vector between the first user feature and the collective user features in the plurality of candidate user sets is represented by 0 or 1.

[0088] In some embodiments, the first user feature is converted into a set of features. For example, if the first user feature includes the target user's gender, age, occupation, income, education, and marital status, the first user feature is converted into a first user feature set.

[0089] In some embodiments, a feature set is constructed based on the target user and multiple candidate user sets. The Jaccard similarity is determined by dividing the size of the first user feature set of the target user and the feature set constructed from the multiple candidate user sets by the size of the union.

[0090] In some embodiments, the similarities between each candidate user set and the first user feature of the target user are compared, and the set with the highest similarity is selected as the set most similar to the target user.

[0091] In this implementation manner, the similarity between the set user features in the plurality of candidate user sets and the first user feature of the target user is determined by using the Jaccard similarity, which has high accuracy.

[0092] It should be understood that since the first user characteristic data of the target user is Boolean data, it is more accurate to determine multiple candidate users similar to the target user based on the Jaccard similarity algorithm.

[0093] In this implementation, multiple candidate user sets are obtained; the similarity between the collective user features in the multiple candidate user sets and the first user features of the target user is determined; and the user set whose similarity between the collective user features in the multiple candidate user sets and the first user features of the target user is greater than a similarity threshold is determined as the target user set to which the target user belongs. That is, in the case where the target user's preference for the vehicle-mounted equipment cannot be obtained, the target user set similar to the target user is determined based on the similarity between the first user feature and the collective user feature, with high accuracy.

[0094] In a possible implementation, the similarity between a first user feature of a target user and second user features of a plurality of candidate users is determined, where the second user feature of each candidate user is used to describe a basic attribute of each candidate user; candidate users among the plurality of candidate users whose similarity between the second user feature and the first user feature of the target user is greater than a similarity threshold are determined as similar users of the target user; and a user set consisting of the plurality of similar users of the target user is determined as a target user set to which the target user belongs.

[0095] Among them, the multiple candidate users are users who have a preference for the vehicle-mounted equipment. The target user set includes multiple users who are similar to the target user.

[0096] It is understandable that the specific implementation of determining the similarity between the first user feature of the target user and the second user features of multiple candidate users may refer to the above description of determining the similarity between the first user feature of the target user and the collective user features.

[0097] In some embodiments, a similarity between a first user feature of a target user and second user features of a plurality of candidate users is determined.

[0098] In some embodiments, when the similarity between a first user feature of a target user and a second user feature among multiple candidate users is greater than a similarity threshold, candidate users whose similarity with the first user feature of the target user is greater than the similarity threshold are determined as similar users of the target user.

[0099] It should be understood that, since the similarity between the first user feature of the target user and the second user feature of the multiple candidate users is greater than the similarity threshold, the multiple candidate users are similar users to the target user.

[0100] In some embodiments, when the similarity between a first user feature of a target user and a second user feature of multiple candidate users is less than or equal to a similarity threshold, candidate users whose similarity with the first user feature of the target user is less than or equal to the similarity threshold are determined as dissimilar users of the target user.

[0101] It should be understood that since no similarity between the second user feature and the first user feature of the target user is greater than the similarity threshold among the multiple candidate users, it is determined that there is a user dissimilar to the target user among the multiple candidate users.

[0102] In this embodiment, the similarity between the first user feature of the target user and the second user features of multiple candidate users is determined, and the second user features of each candidate user are used to describe the basic attributes of each candidate user; the candidate users whose second user features among the multiple candidate users have a similarity greater than a similarity threshold with the first user feature of the target user are determined as similar users of the target user; and the user set consisting of multiple similar users of the target user is determined as the target user set to which the target user belongs. That is, in the case where the target user's preference for the vehicle-mounted equipment cannot be obtained, the target user set similar to the target user is determined based on the similarity between the first user feature and the second user features of multiple candidate users, with high accuracy.

[0103] Step 303: Acquire the collective features of multiple candidate users in the target user set, where the collective features are used to describe the preferences of multiple candidate users in the target user set for the vehicle-mounted equipment.

[0104] It should be understood that since the similarity between the collective user feature in the target user set and the first user feature of the target user is greater than the similarity threshold, the target user set includes multiple candidate users similar to the target user, and the multiple candidate users similar to the target user include multiple collective features.

[0105] The target user set includes a set feature and second user features of multiple candidate users. The set feature is used to describe the preference of multiple candidate users in the target user set for the vehicle-mounted equipment.

[0106] In a possible implementation, second user features of multiple candidate users and behavior data of the multiple candidate users with respect to the in-vehicle equipment are obtained, and collective features of the multiple candidate users are determined based on the second user features and the behavior data of the multiple candidate users with respect to the in-vehicle equipment.

[0107] The candidate user's behavior data on the vehicle-mounted equipment is data generated by the candidate user's operation on the page displaying the vehicle-mounted equipment. For example, the candidate user's behavior data on the vehicle-mounted equipment includes data generated by the user's browsing, evaluation or purchase of the vehicle-mounted equipment.

[0108] In some embodiments, attribute information of multiple candidate users and behavior data of multiple candidate users with respect to the in-vehicle equipment are obtained based on a page displaying the in-vehicle equipment.

[0109] In some embodiments, the attribute information of the multiple candidate users is analyzed to obtain second user features corresponding to the multiple candidate users.

[0110] Among them, the second user characteristics are used to classify the target users, and the second user characteristics include information such as the candidate user's gender, age, occupation, income, education, marital status and place of residence.

[0111] In some embodiments, duplicate data in the second user features of multiple candidate users and the behavioral data of multiple candidate users regarding in-vehicle equipment are removed, and blank or missing values ​​in the second user features and the behavioral data are processed; keywords are extracted from the second user features and the behavioral data, and collective features are created based on the activity index, consumption frequency, etc. of the multiple candidate users in the behavioral data.

[0112] For example, when the second user feature is male and the age is 20 years old student, the behavior data of the candidate user on the vehicle-mounted equipment is processed to obtain the collective feature of the candidate user as liking black hip-hop style vehicle-mounted equipment.

[0113] In this implementation, the second user features of multiple candidate users and the behavior data of multiple candidate users on the vehicle-mounted equipment are obtained, and the collective features of the multiple candidate users are determined based on the second user features and the behavior data of the multiple candidate users on the vehicle-mounted equipment. That is, based on the second user features of the candidate users and the behavior data of the multiple candidate users on the vehicle-mounted equipment, the collective features of the target user set can be obtained, thereby obtaining the preference levels of the multiple candidate users on the vehicle-mounted equipment.

[0114] In some embodiments, the set features in the target user set include user portraits of each candidate user, wherein the user portraits of the candidate users are used to describe the degree of preference of the candidate users for the vehicle-mounted equipment.

[0115] In some embodiments, the set features in the target user set include three levels of tags, which are respectively a first tag, a second tag, and a third tag.

[0116] In some embodiments, Figure 4 As shown, the first tag is used to indicate the identification of the target user set, that is, the first tag is the collective user feature of the target user set. The second tag is used to describe the degree of preference of each candidate user for the vehicle-mounted equipment. The third tag is used to record the number of multiple candidate users in the target user set of the collective feature of the target user set, the prediction accuracy based on the degree of preference of each candidate user for the vehicle-mounted equipment, and the generation time of the target user set.

[0117] For example, the target user set includes 10 candidate users, and the prediction accuracy of the preference of the 10 candidate users for the vehicle-mounted equipment and the generation time of the target user set are t0.

[0118] The prediction accuracy of the preference of each candidate user for the vehicle-mounted equipment is the average of the preference of the target user set for the vehicle-mounted equipment and the accuracy of the actual purchase of the vehicle-mounted equipment among multiple candidate users.

[0119] For example, if it is predicted that 10 candidate users will purchase vehicle-mounted equipment A, and 8 candidate users actually purchase vehicle-mounted equipment A, then the prediction accuracy of the degree of preference of each candidate user for the vehicle-mounted equipment is 80%.

[0120] It should be understood that since the preferences of multiple candidate users for vehicle-mounted equipment will change over time, it is inaccurate to only obtain the set features in the target user set within a certain period of time. It is also necessary to obtain the generation time of the target user set to determine the impact of the generation time on the user's preference for vehicle-mounted equipment.

[0121] In some embodiments, the preference levels of multiple candidate users for the vehicle-mounted equipment are obtained based on a collaborative filtering algorithm, and the time of generating the preference levels of multiple candidate users for the vehicle-mounted equipment is determined as the generation time of the target user set.

[0122] In this implementation, the target user set to which the target user belongs is identified based on the first tag, the preference of each candidate user in the target user set for the vehicle-mounted equipment is determined based on the second tag, and the influence of the generation time of the target user set on the preference of each candidate user for the vehicle-mounted equipment is determined based on the third tag, so as to more accurately obtain the preference of each candidate user in the target user set for the vehicle-mounted equipment. That is, based on the first tag, the second tag and the third tag, the favorite vehicle-mounted equipment is more accurately recommended to each candidate user in the target user set.

[0123] Step 304 , concatenate the first user feature of the target user and the set feature of the target user set to obtain a target user feature of the target user, where the target user feature is used to describe the target user's preference for the vehicle-mounted equipment.

[0124] It can be understood that the specific implementation method of combining the first user feature of the target user and the set feature of the target user set to obtain the target user feature of the target user can refer to the relevant description in the aforementioned step 203.

[0125] It should be understood that since the first user feature of the target user is used to describe the basic attributes of the target user, and the collective feature of the target user set is used to describe the preference of multiple candidate users in the target user set for vehicle-mounted equipment, the first user feature of the target user and the collective feature of the target user set are not in the same dimension and cannot be spliced. In order to make the first user feature of the target user and the collective feature of the target user set have the same dimension, the feature vector of the first user feature of the target user and the feature vector of the collective feature of the target user set are aligned.

[0126] In some embodiments, the feature vector of the first user feature of the target user is aligned with the feature vector of the set feature of the target user set. For example, the feature vector of the first user feature is aligned with the feature vector of the set feature by inner join or outer join.

[0127] The method of aligning the feature vector of the first user feature of the target user with the feature vector of the set feature of the target user set includes dimensionally aligning the feature vector of the first user feature with the feature vector of the set feature and semantically aligning the feature vector of the first user feature with the feature vector of the set feature. In some embodiments, the feature vector of the first user feature of the target user and the feature vector of the set feature of the target user set are concatenated by a programming language to obtain the target user feature of the target user.

[0128] In some embodiments, the feature vector of the target user feature is checked for missing values ​​or abnormal values, and the feature vector of the target user feature is standardized and normalized.

[0129] It should be understood that since the target user feature is obtained by integrating the first user feature of the target user and the set feature of the target user set, the vehicle-mounted equipment preferred by the target user is recommended to the target user based on the target user feature.

[0130] For example, the first user characteristic of the target user is a 20-year-old male, a student, and the collective characteristic of the target user set is a preference for black and white, green, and cool-style vehicle-mounted equipment. Therefore, the target user characteristic is a preference for black and white, green, and cool-style vehicle-mounted equipment, and based on the target user characteristic, the vehicle-mounted equipment that the target user likes is recommended to the target user.

[0131] In this implementation, the first user feature of the target user and the set feature of the target user set are concatenated to obtain the target user feature of the target user, and the target user feature is used to describe the target user's preference for the vehicle-mounted equipment. That is, the accuracy of recommending vehicle-mounted equipment to the target user based on the target user feature is high. Even if the target user does not have the behavior data for the vehicle-mounted equipment and cannot recommend the personalized vehicle-mounted equipment required by the target user to the target user based on the behavior data for the vehicle-mounted equipment, the vehicle-mounted equipment can still be recommended to the target user based on the target user feature, and the recommendation accuracy is high.

[0132] Step 305 : Based on the target user feature, determine whether there is a target reference user among the multiple reference users, where the target reference user is a reference user whose user feature has a similarity with the target user feature that is greater than or equal to a similarity threshold.

[0133] The multiple reference users are candidate users having multiple second user features.

[0134] It should be understood that the target user feature is used to describe the target user's preference for the vehicle-mounted equipment, so the target user feature corresponds to the first user feature of the target user. Since the target user is a user with little or no behavior data for the vehicle-mounted equipment, there may not be a similar user similar to the target user among the multiple reference users, so it is necessary to determine whether there is a target reference user among the multiple reference users based on the target user feature.

[0135] In some embodiments, it is determined whether the similarity between the user features of the plurality of reference users and the target user features is greater than or equal to a similarity threshold based on modified cosine similarity or Jaccard similarity.

[0136] It is understandable that the specific implementation of determining whether the similarity between the user features of multiple reference users and the target user features is greater than or equal to the similarity threshold based on the modified cosine similarity or the Jaccard similarity can refer to the detailed steps in the aforementioned step 302.

[0137] It should be understood that since the target user feature is obtained by combining the first user feature and the set feature, it is determined whether the target reference user exists among the multiple reference users based on the first user feature in the target user feature.

[0138] In some embodiments, whether the similarity between the user features of the plurality of reference users and the first user feature of the target user is greater than or equal to a similarity threshold is determined based on modified cosine similarity or Jaccard similarity.

[0139] It is understandable that the specific implementation of determining whether the similarity between the user features of multiple reference users and the first user feature of the target user is greater than or equal to the similarity threshold based on the modified cosine similarity or the Jaccard similarity can refer to the detailed steps in the aforementioned step 302.

[0140] In this implementation, based on the target user characteristics, it is determined whether there is a target reference user among the multiple reference users, and the target reference user is a reference user whose user characteristics have a similarity greater than or equal to a similarity threshold. That is, based on the target user characteristics, it is determined whether there is a user similar to the target user among the multiple reference users, thereby determining the vehicle-mounted equipment preferred by the target user.

[0141] Optionally, after executing the above step 305, according to actual circumstances, the following steps may be executed or the following step 306 may be executed directly.

[0142] In a possible implementation, when there is no target reference user among multiple reference users, based on target user characteristics, target vehicle equipment is determined from a set of vehicle equipment preferred by multiple similar users in the target user set; and the target vehicle equipment is recommended to the target user.

[0143] In some embodiments, when the similarity between the user features of the multiple reference users and the first user feature of the target user is less than a similarity threshold, it is determined that the target reference user does not exist in the multiple reference users.

[0144] It should be understood that since the target user is a user without behavior data on the vehicle-mounted equipment, the target reference user may not exist among the multiple reference users, and therefore the vehicle-mounted equipment preferred by the target reference user cannot be recommended to the target user.

[0145] It should also be understood that the target user characteristics are used to describe the target user's preference for vehicle-mounted equipment. Therefore, the target vehicle-mounted equipment corresponding to the target user is determined from a set of vehicle-mounted equipment with similar user preferences in the target user set, thereby achieving personalized recommendations for the target user.

[0146] In some embodiments, the target user set includes multiple candidate users, and the multiple candidate users have different preferences for different vehicle-mounted equipment. Therefore, the target vehicle-mounted equipment is recommended to the target user based on the preference levels of the multiple candidate users for the vehicle-mounted equipment, thereby achieving personalized recommendation.

[0147] For example, the target user characteristics of the target user are a 20-year-old male student who likes cool-style car equipment, while the target user set includes 21-25-year-old males who like "street" style car equipment, and the target user set does not include 20-year-old male students. However, based on the 21-25-year-old males in the target user set, the "street" style car equipment is recommended to the target user, a 20-year-old male student.

[0148] In this implementation, when there is no target reference user among multiple reference users, the target vehicle equipment is determined from the vehicle equipment sets preferred by multiple similar users in the target user set based on the target user characteristics; and the target vehicle equipment is recommended to the target user. That is, even when there is no target reference user among multiple reference users, the target vehicle equipment in the vehicle equipment set corresponding to the target user set can be recommended to the target user based on the target user characteristics, thereby realizing personalized recommendation of vehicle equipment for the target user.

[0149] Step 306 : When there is a target reference user among the multiple reference users, recommend the vehicle-mounted equipment preferred by the target reference user to the target user.

[0150] The target reference user is used to determine the vehicle-mounted equipment preferred by the target user.

[0151] It should be understood that when there is a target reference user among multiple reference users, that is, there is a target reference user who is similar to the target user among multiple reference users, the vehicle-mounted equipment preferred by the target reference user is recommended to the target user, so that the target user can more accurately purchase the vehicle-mounted equipment that he or she may like.

[0152] In some embodiments, when the similarity between the user features of the multiple reference users and the first user feature of the target user is greater than or equal to a similarity threshold, it is determined that there is a target reference user among the multiple reference users.

[0153] In some embodiments, based on the target reference user's preference for the in-vehicle equipment, the current plurality of in-vehicle equipment are sorted, thereby generating a recommendation list corresponding to the target reference user.

[0154] In some embodiments, the vehicle-mounted equipment most relevant to the target user's interests is ranked at the front, and products that the target user may not have seen are recommended to the target user, so that the vehicle-mounted equipment recommended to the target user is richer.

[0155] In this implementation, when there is a target reference user among multiple reference users, the vehicle-mounted equipment preferred by the target reference user is recommended to the target user. That is, since there is a similar user similar to the target user among multiple reference users, the vehicle-mounted equipment preferred by the target reference user is recommended to the target user, thereby increasing the accuracy of recommending vehicle-mounted equipment for the target user.

[0156] The embodiment of the present application provides a recommendation method based on portrait similarity, which obtains the first user feature of the target user, and the first user feature is used to describe the basic attributes of the target user. Since the target user does not have data corresponding to the vehicle-mounted equipment or has very little data corresponding to the vehicle-mounted equipment, the target user set to which the target user belongs is determined based on the first user feature of the target user. The target user set is a set in which the similarity between the collective user feature in the target user set and the first user feature of the target user is greater than the similarity threshold, and the collective user feature is used to describe the basic attributes of multiple candidate users in the target user set. The collective features of multiple candidate users in the target user set are obtained, and the collective features are used to describe the degree of preference of multiple candidate users in the target user set for vehicle-mounted equipment. The first user feature of the target user and the collective features of the target user set are spliced ​​to obtain the target user feature of the target user, and the target user feature is used to describe the degree of preference of the target user for vehicle-mounted equipment. That is, the collective features in the target user set similar to the target user are spliced ​​to the first user feature of the target user, so as to describe the degree of preference of the target user for vehicle-mounted equipment through the target user feature. Based on the target user characteristics, it is determined whether there is a target reference user among the multiple reference users, and the target reference user is a reference user whose user characteristics have a similarity greater than or equal to a similarity threshold. If there is a target reference user among the multiple reference users, the vehicle equipment preferred by the target reference user is recommended to the target user, that is, the vehicle equipment preferred by the target reference user is recommended to the target user, thereby providing personalized services for the target user and alleviating the problem of cold start when purchasing vehicle equipment.

[0157] Figure 5 It is a schematic flowchart of a method for determining a target user set provided in an embodiment of the present application.

[0158] For example, Figure 5 As shown, taking the execution subject as a server as an example, the method 500 includes:

[0159] Step 501 , obtaining second user features of multiple candidate users and behavior data of multiple candidate users with respect to vehicle-mounted equipment, wherein the behavior data of the candidate users with respect to the vehicle-mounted equipment is data generated when the candidate users operate a page displaying the vehicle-mounted equipment.

[0160] In some embodiments, second user characteristics of multiple candidate users and behavior data of multiple candidate users with respect to the in-vehicle equipment are obtained based on the page displaying the in-vehicle equipment.

[0161] In some embodiments, a data collection module is embedded in the vehicle-mounted system, and the behavior data of multiple candidate users on the vehicle-mounted equipment is obtained based on the data collection module, for example, a feedback button of a user interface.

[0162] It can be understood that the specific implementation of obtaining the second user characteristics of multiple candidate users and the behavior data of multiple candidate users with respect to the in-vehicle equipment can refer to the relevant description in the aforementioned step 303.

[0163] In some embodiments, the behavior data of multiple candidate users with respect to the in-vehicle equipment are obtained within a preset time period, so that the obtained second user characteristics and the behavior data of the in-vehicle equipment are more accurate.

[0164] Among them, the preset time is the time automatically determined by the server. The embodiment of the present application does not limit this time. For example, the automatically determined time can be 30 days.

[0165] In this implementation, the second user features of multiple candidate users and the behavior data of multiple candidate users on the vehicle-mounted equipment are obtained, and the behavior data of the candidate users on the vehicle-mounted equipment is the data generated by the candidate users operating the page displaying the vehicle-mounted equipment. That is, the degree of preference of the multiple candidate users on the vehicle-mounted equipment can be determined through the behavior data.

[0166] Step 502 , processing the second user features of the plurality of candidate users and the behavior data of the plurality of candidate users with respect to the vehicle-mounted equipment, to obtain the preference levels of the plurality of candidate users with respect to the vehicle-mounted equipment and the user portraits of the plurality of candidate users.

[0167] In some embodiments, the second user features of multiple candidate users and the behavioral data of multiple candidate users regarding vehicle-mounted equipment are processed based on a collaborative filtering algorithm to obtain the preference levels of multiple candidate users for vehicle-mounted equipment, user portraits of multiple candidate users, generation time of the user portraits of multiple candidate users, and prediction accuracy of the preference levels of multiple candidate users for vehicle-mounted equipment.

[0168] In some embodiments, a similarity threshold is set based on a collaborative filtering algorithm, and each similar user who is similar to the target user is screened out from a plurality of candidate users based on the similarity threshold.

[0169] For example, the similarity threshold is 0.8, the total number of candidate users is N, and n similar users who are similar to the target user are screened out from the total number of candidate users.

[0170] In some embodiments, as shown in the following table, the user U is calculated based on the table. i . Similarity with other users.

[0171]

[0172]

[0173] Among them, U i (1≤i≤n) represents the i-th user among all n users, I j (1≤j≤m) represents the jth equipment among all m modified equipment, R ij Represents user U i Equipment I j Ratings or purchase records of a user U. i When implementing the recommendation, the vector R i = {R i1 , R i2 , R i3 , …, R im} to calculate user U i Similarity with other users. That is, calculate R i = {R i1 , R i2 , R i3 , …, R im} similarity between them.

[0174] In some embodiments, multiple candidate users have corresponding user portraits. For example, n similar users have n user portraits.

[0175] In a possible implementation, the second user features of the plurality of candidate users and the behavior data of the plurality of candidate users with respect to the in-vehicle equipment are clustered to obtain the preference degrees of the plurality of candidate users with respect to the in-vehicle equipment.

[0176] In some embodiments, data mining is performed on the behavior data of multiple candidate users with respect to the in-vehicle equipment, clusters are automatically divided based on the similarity or distance between the behavior data of multiple candidate users with respect to the in-vehicle equipment, and the average similarity of the clusters and the differences between the clusters are evaluated.

[0177] In some embodiments, a weighted sum is performed on the preference levels of the n candidate users, and the calculation formula is as follows.

[0178]

[0179] Among them, X i W represents the preference of the i-th candidate user for vehicle-mounted equipment. irepresents the primary tag weight value of the i-th similar candidate user. Then filter X sum The top five corresponding vehicle-mounted equipment in descending order are added to the preference prediction list of the set features in the target user set.

[0180] In a possible implementation, first weight values ​​of second user characteristics of a plurality of candidate users are determined, where the first weight values ​​are used to indicate the influence of the second user characteristics on the preference for the in-vehicle equipment.

[0181] In some embodiments, the first weight value of the second user feature is directly extracted from the second user features of the plurality of candidate users.

[0182] For example, if the current user is male, it is represented by the first weight value <gender, 1>, and the first weight value is 1.

[0183] In this implementation, since the second user feature is used to describe the basic attributes of multiple candidate users, the weight value of the second user feature has a greater impact on the preference degree of the vehicle-mounted equipment.

[0184] In a possible implementation, based on the generation time of the first weight value, a second weight value of the preference of multiple candidate users for the vehicle-mounted equipment is calculated, and the second weight value is used to indicate the preference of multiple candidate users for the vehicle-mounted equipment over time.

[0185] In some embodiments, the second weight value is calculated as shown in the following formula.

[0186]

[0187] Where W represents the second weight value. S represents the proportion of missing data to all data, that is, S represents the sparsity of behavior data. The reciprocal of S is used to measure the completeness of the user profile information of the candidate user. λ represents the time to generate the user profile of the candidate user.

[0188] In some embodiments, if λ is 0, it means that the candidate user's preference for the vehicle-mounted equipment will not change over time. If λ is 1, it means that the candidate user's preference for the vehicle-mounted equipment will change over time.

[0189] In this embodiment, based on the generation time of the first weight value, a second weight value of the preference of multiple candidate users for the vehicle-mounted equipment is calculated, and the second weight value is used to indicate the preference of multiple candidate users for the vehicle-mounted equipment over time. That is, the influence of the generation time of the candidate user's preference for the vehicle-mounted equipment on the generation of the candidate user's user portrait is determined, so as to obtain a more accurate user portrait of the candidate user.

[0190] In a possible implementation, the average of the second weight values ​​of the preferences of multiple candidate users for vehicle-mounted equipment is calculated to obtain the third weight values ​​of the user portraits of the multiple candidate users, and the third weight values ​​are used to describe the credibility of the user portraits of the multiple candidate users.

[0191] It should be understood that since the user portrait of the candidate user has a greater impact on the user's preference for vehicle-mounted equipment, the higher the credibility of the user portrait of the candidate user, the higher the accuracy of the vehicle-mounted equipment recommended to the user. Therefore, it is necessary to determine the credibility of the user portraits of multiple candidate users.

[0192] In this implementation, the average of the second weight values ​​of the preference of multiple candidate users for the vehicle-mounted equipment is calculated to obtain the third weight values ​​of the user portraits of the multiple candidate users, and the third weight values ​​are used to describe the credibility of the user portraits of the multiple candidate users. That is, based on the third weight value, the credibility of the user portraits of the multiple candidate users can be determined, thereby obtaining the accuracy of recommending the vehicle-mounted equipment to the user. The larger the third weight value, the higher the credibility and accuracy of the user portraits of the multiple candidate users.

[0193] It should be understood that since different candidate users have different preferences for vehicle-mounted equipment and second user characteristics, the user portraits of the candidate users are affected differently by the second user characteristics of the candidate users and the preferences of the candidate users for vehicle-mounted equipment. In order to obtain user portraits of candidate users with higher accuracy, user portraits of multiple candidate users are generated based on the first weight value, the second weight value, the third weight value, the second user characteristics of multiple candidate users, and the preferences of multiple candidate users for vehicle-mounted equipment.

[0194] In a possible implementation, user portraits of multiple candidate users are generated based on the first weight value, the second weight value, the third weight value, the second user features of the multiple candidate users, and the preference levels of the multiple candidate users for the vehicle-mounted equipment.

[0195] In some embodiments, Figure 6 As shown, the user portrait of the candidate user is shown in the figure, the first weight value represents the importance of the second user feature, the second weight value is used to represent the importance of the candidate user's preference for the vehicle-mounted equipment, and the third weight value is used to represent the importance of the user portrait of the candidate user.

[0196] In some embodiments, the second user characteristics include information such as gender, age, occupation, and education level for describing the basic attributes of the candidate user. The preferences for vehicle-mounted equipment include preferences for electric, color, performance, ventilation, warmth, safety, and personality. The commercial attributes of the preferred vehicle-mounted equipment include attributes such as brand loyalty, equipment applicable season, equipment accessories, equipment environmental adaptability, and whether the equipment is resistant to high temperatures.

[0197] In some embodiments, the first weight value, the second weight value, the second user feature, and the candidate user's preference for the in-vehicle equipment are weighted and summed to generate a reference user portrait of the candidate user.

[0198] Among them, the reference user profile of the candidate user is used to analyze and understand the preferences of the candidate user.

[0199] It should be understood that the candidate user's preference for the in-vehicle equipment affects the second weight value, and the feature vector of the reference user portrait of the candidate user can be determined based on the first weight value and the second weight value.

[0200] For example, the first weight value is 0.6, the second weight value is 0.4, the feature vector of the second user feature is (a, b, c), the candidate user's preference for vehicle-mounted equipment is 0.8, then the feature vector of the candidate user's reference user portrait is (d, e, f).

[0201] It should be understood that since the third weight value is used to describe the credibility of the user portraits of multiple candidate users, the higher the third weight value, the higher the accuracy of generating the user portrait of the candidate user. In order to more accurately analyze and understand the preferences of the candidate users, the user portrait of the candidate user is determined based on the third weight value.

[0202] In some embodiments, the product of the third weight value and the feature vector of the reference user portrait of the candidate user is determined as the feature vector of the user portrait of the candidate user, thereby determining the user portrait of the candidate user.

[0203] For example, if the third weight value is 0.1 and the feature vector of the reference user portrait of the candidate user is (d, e, f), then the feature vector of the user portrait of the candidate user is (g, h, k). In this embodiment, based on the first weight value, the second weight value, the third weight value, the second user features of the plurality of candidate users, and the preference of the plurality of candidate users for the vehicle-mounted equipment, the user portraits of the plurality of candidate users are generated, that is, the credibility of recommending the vehicle-mounted equipment to the candidate users is determined by describing the user portraits of the plurality of candidate users through the first weight value, the second weight value, and the third weight value, so that the accuracy of recommending the vehicle-mounted equipment is higher.

[0204] Step 503: determine a user set consisting of candidate users whose user portraits have similarities greater than a similarity threshold as a target user set.

[0205] It should be understood that since the user portraits of multiple candidate users are relatively scattered and the preference for the described vehicle-mounted equipment is relatively low, in order to obtain richer user portraits, the collective characteristics of the target user set are generated based on the preference levels of multiple candidate users for vehicle-mounted equipment and the user portraits of multiple candidate users, thereby enriching the dimensions of the target user set.

[0206] In some embodiments, data cleaning is performed on relevant features of user portraits of multiple candidate users, and similar users are grouped, and user portraits of candidate users with similar second user features are clustered together, thereby determining a target user set.

[0207] In some embodiments, when the similarity between the second user features of multiple candidate users is greater than a similarity threshold, the multiple candidate users are grouped into a user set, and the user set is determined as a target user set.

[0208] In this implementation, a user set consisting of candidate users whose user portraits have similarities greater than a similarity threshold is determined as a target user set, and the target user set has a high similarity.

[0209] In some embodiments, the set features of the target user set are generated based on the grouping of similar users and the user portraits of multiple candidate users.

[0210] It should be understood that since the set features in the target user set include user portraits of multiple candidate users, analyzing the differences and similarities between different groups based on the set features of the target user set can recommend more accurate and abundant vehicle-mounted equipment to users.

[0211] In this implementation, the collective characteristics of the target user set are generated based on the preference levels of multiple candidate users for vehicle-mounted equipment and the user portraits of the multiple candidate users. That is, by analyzing the differences and similarities between different target user sets based on different target user sets, the market demand for vehicle-mounted equipment can be obtained more accurately.

[0212] In some embodiments, the weight value of the set feature of the target user set is calculated as shown in the following formula.

[0213]

[0214] Among them, W qis the weight value of the set feature of the target user set, n is the user similar to the target user, N is the total number of users participating in the calculation in the server, r is the accuracy of predicting vehicle equipment, λ is the stability of the set feature, and t0 is the time for generating the set feature.

[0215] The embodiment of the present application provides a method for determining a target user set, obtaining the second user features of multiple candidate users and the behavior data of multiple candidate users on vehicle-mounted equipment, the behavior data of candidate users on vehicle-mounted equipment being data generated by the candidate users operating the vehicle-mounted equipment display page, processing the second user features of multiple candidate users and the behavior data of multiple candidate users on vehicle-mounted equipment, obtaining the preference degree of multiple candidate users on vehicle-mounted equipment and the user portraits of multiple candidate users, and determining the user set consisting of candidate users whose user portraits have a similarity greater than a similarity threshold as the target user set. That is, the user portraits of each candidate user are combined into a target user set, so that the server can analyze a target user set group and obtain a richer degree of preference of candidate users on vehicle-mounted equipment.

[0216] Figure 7 It is a structural diagram of a recommendation device based on portrait similarity provided in an embodiment of the present application.

[0217] For example, Figure 7 As shown, the device 700 includes:

[0218] An acquisition module 701 is used to acquire a first user feature of a target user, where the first user feature is used to describe a basic attribute of the target user;

[0219] A determination module 702 is used to determine, based on the first user feature of the target user, a target user set to which the target user belongs, where the target user set is a set in which the similarity between the collective user features in the target user set and the first user feature of the target user is greater than a similarity threshold, and the collective user features are used to describe basic attributes of multiple candidate users in the target user set;

[0220] A concatenation module 703 is used to concatenate the first user feature of the target user and the set feature of the target user set to obtain a target user feature of the target user, wherein the set feature is used to describe the preference of multiple candidate users in the target user set for the vehicle-mounted equipment, and the target user feature is used to describe the preference of the target user for the vehicle-mounted equipment;

[0221] The recommendation module 704 is used to recommend vehicle-mounted equipment to the target user based on the target user's characteristics.

[0222] In a possible implementation manner, the device 700 further includes:

[0223] A user set acquisition module is used to acquire multiple candidate user sets;

[0224] A similarity determination module is used to determine the similarity between the collective user features in the plurality of candidate user sets and the first user feature of the target user;

[0225] The determination module 702 is configured to determine a user set whose similarity between collective user features in a plurality of candidate user sets and the first user feature of the target user is greater than a similarity threshold as a target user set to which the target user belongs.

[0226] In a possible implementation manner, the device 700 further includes:

[0227] A similarity determination module is used to determine the similarity between a first user feature of a target user and second user features of a plurality of candidate users, wherein the second user feature of each candidate user is used to describe a basic attribute of each candidate user;

[0228] A similar user determination module is used to determine candidate users among multiple candidate users whose similarity between the second user feature and the first user feature of the target user is greater than a similarity threshold as similar users of the target user;

[0229] The determination module 702 is used to determine a user set consisting of multiple similar users of the target user as the target user set to which the target user belongs.

[0230] In a possible implementation manner, the device 700 further includes:

[0231] A candidate user data acquisition module is used to acquire second user characteristics of multiple candidate users and behavior data of multiple candidate users with respect to vehicle-mounted equipment, wherein the behavior data of candidate users with respect to vehicle-mounted equipment is data generated when the candidate users operate a vehicle-mounted equipment display page;

[0232] A processing module, used to process the second user features of the plurality of candidate users and the behavior data of the plurality of candidate users with respect to the vehicle-mounted equipment, to obtain the preference degrees of the plurality of candidate users with respect to the vehicle-mounted equipment and the user portraits of the plurality of candidate users;

[0233] The module for generating set features is used to generate set features of a target user set based on the preference levels of multiple candidate users for vehicle-mounted equipment and the user portraits of multiple candidate users.

[0234] In a possible implementation manner, the device 700 further includes:

[0235] An analysis module, used to cluster the second user features of the plurality of candidate users and the behavior data of the plurality of candidate users with respect to the vehicle-mounted equipment, and obtain the preference degree of the plurality of candidate users with respect to the vehicle-mounted equipment;

[0236] A first weight value determination module is used to determine first weight values ​​of second user characteristics of multiple candidate users, where the first weight values ​​are used to indicate the influence of the second user characteristics on the preference degree of the vehicle-mounted equipment;

[0237] A second weight value calculation module, used to calculate a second weight value of the preference of multiple candidate users for the vehicle-mounted equipment based on the generation time of the first weight value, the second weight value is used to indicate the preference of the multiple candidate users for the vehicle-mounted equipment over time;

[0238] A third weight value calculation module is used to calculate the average of the second weight values ​​of the preference of multiple candidate users for the vehicle-mounted equipment, and obtain the third weight values ​​of the user portraits of the multiple candidate users, and the third weight value is used to describe the credibility of the user portraits of the multiple candidate users;

[0239] A user portrait generation module is used to generate user portraits of multiple candidate users based on the first weight value, the second weight value, the third weight value, the second user characteristics of the multiple candidate users, and the preference levels of the multiple candidate users for the vehicle-mounted equipment.

[0240] In a possible implementation manner, the device 700 further includes:

[0241] A target reference user determination module is used to determine whether there is a target reference user among multiple reference users based on the target user characteristics, where the target reference user is a reference user whose user characteristics have a similarity greater than or equal to a similarity threshold with the target user characteristics;

[0242] The recommendation module 703 is used to recommend the vehicle-mounted equipment preferred by the target reference user to the target user when there is a target reference user among the multiple reference users.

[0243] In a possible implementation manner, the device 700 further includes:

[0244] A target vehicle equipment determination module is used to determine the target vehicle equipment from a set of vehicle equipment preferred by multiple similar users in a target user set based on target user characteristics when there is no target reference user among multiple reference users;

[0245] The recommendation module 703 is used to recommend the target vehicle-mounted equipment to the target user.

[0246] Figure 8 It is a structural schematic diagram of an electronic device provided in an embodiment of the present application.

[0247] For example, Figure 8As shown, the electronic device 800 includes: a memory 801 and a processor 802, wherein the memory 801 stores an executable program code 803, and the processor 802 is used to call and execute the executable program code 803 to perform a recommendation method based on portrait similarity.

[0248] In addition, an embodiment of the present application also protects a device, which may include a memory and a processor, wherein the memory stores executable program code, and the processor is used to call and execute the executable program code to execute a recommendation method based on portrait similarity provided in an embodiment of the present application.

[0249] In this embodiment, the functional modules of the device can be divided according to the above method example. For example, each functional module can be corresponded, or two or more functions can be integrated into one processing module. The above integrated module can be implemented in the form of hardware. It should be noted that the division of modules in this embodiment is schematic and is only a logical function division. There may be other division methods in actual implementation.

[0250] In the case of dividing each functional module according to each function, the device may also include a module for determining a target vehicle-mounted equipment, a module for determining a target reference user, and a module for generating a user portrait, etc. It should be noted that all relevant contents of each step involved in the above method embodiment can be referred to the functional description of the corresponding functional module, and will not be repeated here.

[0251] It should be understood that the device provided in this embodiment is used to execute the above-mentioned recommendation method based on portrait similarity, and thus can achieve the same effect as the above-mentioned implementation method.

[0252] In the case of an integrated unit, the device may include a processing module and a storage module. When the device is applied to an electronic device, the processing module may be used to control and manage the actions of the electronic device. The storage module may be used to support the electronic device in executing related program codes, etc.

[0253] The processing module may be a processor or a controller, which may implement or execute various exemplary logic blocks, modules and circuits shown in conjunction with the disclosure of the present application. The processor may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of digital signal processing (DSP) and a microprocessor, etc. The storage module may be a memory.

[0254] In addition, the device provided in the embodiments of the present application may specifically be a chip, component or module, and the chip may include a connected processor and memory; wherein the memory is used to store instructions, and when the processor calls and executes the instructions, the chip can execute a recommendation method based on portrait similarity provided in the above embodiments.

[0255] This embodiment also provides a computer-readable storage medium, in which a computer program code is stored. When the computer program code is executed on a computer, the computer executes the above-mentioned related method steps to implement a recommendation method based on portrait similarity provided in the above embodiment.

[0256] This embodiment also provides a computer program product. When the computer program product runs on a computer, it enables the computer to execute the above-mentioned related steps to implement a recommendation method based on portrait similarity provided in the above embodiment.

[0257] Among them, the device, computer-readable storage medium, computer program product or chip provided in this embodiment is used to execute the corresponding method provided above. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding method provided above, and will not be repeated here.

[0258] Through the description of the above implementation methods, technical personnel in the relevant field can understand that for the convenience and simplicity of description, only the division of the above-mentioned functional modules is used as an example. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0259] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of modules or units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0260] The above contents are only specific implementation methods of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A recommendation method based on portrait similarity, characterized in that: The method comprises: Acquire a first user feature of a target user, where the first user feature is used to describe a basic attribute of the target user; Based on the first user feature of the target user, determine a target user set to which the target user belongs, the target user set being a set in which a similarity between a collective user feature in the target user set and the first user feature of the target user is greater than a similarity threshold, the collective user feature being used to describe basic attributes of multiple candidate users in the target user set; concatenating the first user feature of the target user and the set feature of the target user set to obtain a target user feature of the target user, wherein the set feature is used to describe the degree of preference of multiple candidate users in the target user set for the vehicle-mounted equipment, and the target user feature is used to describe the degree of preference of the target user for the vehicle-mounted equipment; Recommending vehicle-mounted equipment to the target user based on the target user characteristics.

2. The method according to claim 1, characterized in that The determining, based on the first user feature of the target user, a target user set to which the target user belongs includes: Obtain multiple candidate user sets; Determine the similarity between the collective user features in the plurality of candidate user sets and the first user feature of the target user; A user set whose similarity between the collective user features in the multiple candidate user sets and the first user feature of the target user is greater than a similarity threshold is determined as the target user set to which the target user belongs.

3. The method according to claim 1, characterized in that The determining, based on the first user feature of the target user, a target user set to which the target user belongs includes: Determine the similarity between the first user feature of the target user and the second user features of multiple candidate users, where the second user feature of each candidate user is used to describe the basic attributes of each candidate user; Determine the candidate users whose second user characteristics among the multiple candidate users have a similarity greater than a similarity threshold with the first user characteristics of the target user as similar users of the target user; A user set consisting of multiple similar users of the target user is determined as the target user set to which the target user belongs.

4. The method according to claim 1, characterized in that: Before the first user feature of the target user is concatenated with the set feature of the target user set to obtain the target user feature of the target user, the method includes: Acquire second user features of multiple candidate users and behavior data of multiple candidate users with respect to vehicle-mounted equipment, wherein the behavior data of the candidate users with respect to the vehicle-mounted equipment is data generated by the candidate users operating a page displaying the vehicle-mounted equipment; Processing the second user features of the plurality of candidate users and the behavior data of the plurality of candidate users with respect to the vehicle-mounted equipment to obtain the preference levels of the plurality of candidate users with respect to the vehicle-mounted equipment and the user portraits of the plurality of candidate users; The set features of the target user set are generated based on the preference levels of the multiple candidate users for the vehicle-mounted equipment and the user portraits of the multiple candidate users.

5. The method according to claim 4, characterized in that The processing of the second user features of the plurality of candidate users and the behavior data of the plurality of candidate users with respect to the vehicle-mounted equipment to obtain the preference levels of the plurality of candidate users with respect to the vehicle-mounted equipment and the user portraits of the plurality of candidate users includes: Clustering the second user features of the plurality of candidate users and the behavior data of the plurality of candidate users with respect to the vehicle-mounted equipment to obtain the preference levels of the plurality of candidate users with respect to the vehicle-mounted equipment; Determining first weight values ​​of second user characteristics of the plurality of candidate users, the first weight values ​​being used to indicate the influence of the second user characteristics on the preference degree of the in-vehicle equipment; Calculating, based on the generation time of the first weight value, second weight values ​​of the preference levels of the plurality of candidate users for the in-vehicle equipment, the second weight values ​​being used to indicate the preference levels of the plurality of candidate users for the in-vehicle equipment over time; Calculating an average of the second weight values ​​of the preference levels of the multiple candidate users for the vehicle-mounted equipment to obtain third weight values ​​of the user portraits of the multiple candidate users, wherein the third weight values ​​are used to describe the credibility of the user portraits of the multiple candidate users; User portraits of multiple candidate users are generated based on the first weight value, the second weight value, the third weight value, the second user characteristics of the multiple candidate users, and the preference levels of the multiple candidate users for the vehicle-mounted equipment.

6. The method according to claim 1, characterized in that The recommending vehicle-mounted equipment to the target user based on the target user characteristics includes: Based on the target user feature, determining whether there is a target reference user among multiple reference users, the target reference user being a reference user whose user feature has a similarity greater than or equal to a similarity threshold with the target user feature; In the case that there is a target reference user among the multiple reference users, the in-vehicle equipment preferred by the target reference user is recommended to the target user.

7. The method according to claim 6, characterized in that The method further comprises: In the case that there is no target reference user among the multiple reference users, determining the target vehicle-mounted equipment from a set of vehicle-mounted equipment preferred by multiple similar users in the target user set based on the target user characteristics; The target vehicle-mounted equipment is recommended to the target user.

8. A recommendation device based on image similarity, characterized in that: The device comprises: An acquisition module, used to acquire a first user feature of a target user, where the first user feature is used to describe a basic attribute of the target user; a determination module, configured to determine, based on the first user feature of the target user, a target user set to which the target user belongs, wherein the target user set is a set in which a similarity between a collective user feature in the target user set and the first user feature of the target user is greater than a similarity threshold, and the collective user feature is used to describe basic attributes of multiple candidate users in the target user set; a splicing module, configured to splice the first user feature of the target user and the set feature of the target user set to obtain a target user feature of the target user, wherein the set feature is used to describe the preference of multiple candidate users in the target user set for the vehicle-mounted equipment, and the target user feature is used to describe the preference of the target user for the vehicle-mounted equipment; A recommendation module is used to recommend vehicle-mounted equipment to the target user based on the target user's characteristics.

9. An electronic device, characterized in that: The electronic device comprises: A memory for storing executable program codes; A processor, configured to call and run the executable program code from the memory, so that the electronic device executes the method according to 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, and when the computer program is executed, the method according to any one of claims 1 to 7 is implemented.

Images