CN115760196A - Activity recommendation method, device, terminal and program product for long-tail users - Google Patents

Abstract

The application provides an activity recommendation method, device, terminal and program product for a long-tail user. Relates to the technical field of big data processing. Grouping the obtained long-tail users to obtain a plurality of user groups; respectively establishing a recommendation potential model for each user group, and acquiring a user group potential list by adopting the recommendation potential model; and performing activity recommendation on the users not participating in the activity according to the user group potential name lists matched with the user groups. The method comprises the steps of obtaining a plurality of user groups by grouping long-tail users, establishing a recommendation potential model by taking users who have participated in activities as samples for each user group, and obtaining lists containing potential values of the users who do not participate in the activities for participating in the long-tail activities through the established models, so that the estimation condition of the users who do not participate in the activities for participating in the long-tail activities can be obtained according to the potential lists of the user groups, accurate and efficient recommendation can be performed on the users according to the potential lists of the user groups, and the viscosity of the activities and the long-tail users is enhanced.

Description

Activity recommendation method, device, terminal and program product for long-tail users

Technical Field

The application relates to the technical field of big data processing, in particular to an activity recommendation method, device, terminal and program product for long-tail users.

Background

For a long time, most financial institutions define mainstream according to "two eight law", i.e., 80% of performance and profit are contributed by 20% of large customers. This traditional concept has caused most financial institutions to attach importance only to high-end customers, and neglect other 80% of the middle and low-end customer population, i.e., "long-tailed users".

In recent years, with the development of mobile internet, the potential of the long-tail client is increasing, and in this background, it is very valuable to analyze, mine and operate the potential of the long-tail client. However, at present, the number of long-tailed customers is huge, the behaviors are disordered, and the customers with larger potential can not be accurately positioned; in addition, the service radius of off-line outlets of the bank and the number of customer managers are limited, and all customers cannot be covered. Therefore, accurate activity recommendation cannot be carried out on the long-tailed customers at present so as to realize potential mining on the long-tailed customers.

Disclosure of Invention

The application provides an activity recommendation method, an activity recommendation device, a terminal and a program product for a long-tail user, which are used for realizing activity recommendation for the long-tail user.

In a first aspect, the present application provides an activity recommendation method for a long-tailed user, including: grouping the obtained long-tail users to obtain a plurality of user groups, wherein each user group corresponds to different major activities;

respectively establishing a recommendation potential model for each user group, and acquiring a user group potential list by adopting the recommendation potential model, wherein the user group potential list comprises potential values of users who do not participate in activities and participate in large-class activities;

and performing activity recommendation on users not participating in the activity according to the user group potential name lists matched with the user groups.

In a second aspect, the present application provides an activity recommendation apparatus for a long-tailed user, including: the grouping module is used for grouping the obtained long-tail users to obtain a plurality of user groups, wherein each user group corresponds to different large-class activities;

the system comprises a user group potential list acquisition module, a recommendation potential model generation module and a recommendation potential model generation module, wherein the user group potential list acquisition module is used for respectively establishing a recommendation potential model for each user group and acquiring a user group potential list by adopting the recommendation potential model, and the user group potential list comprises potential values of users not participating in activities and participating in large-class activities;

and the activity recommendation module is used for recommending the activity of the users not participating in the activity according to the user group potential name lists matched with the user groups.

In a third aspect, the present application provides a terminal, including: a processor, and a memory communicatively coupled to the processor;

the memory stores computer execution instructions;

the processor executes computer-executable instructions stored by the memory to implement the methods described herein.

In a fourth aspect, the present application provides a computer-readable storage medium having stored thereon computer-executable instructions for implementing the method described herein when executed by a processor.

In a fifth aspect, the present application provides a computer program product comprising a computer program which, when executed by a processor, performs the method described herein.

According to the activity recommendation method, the activity recommendation device, the terminal and the program product for the long-tail users, a plurality of user groups are obtained by grouping the long-tail users, each user group corresponds to different major activities, the users who participate in the activities are used as samples for establishing a recommendation potential model for each user group, the list containing the potential values of the users who do not participate in the activities and participate in the major activities is obtained through the established model, and therefore the estimation conditions of the users who do not participate in the activities and participate in the major activities can be obtained according to the potential list of the user groups, accurate and efficient recommendation can be conducted on the users according to the potential list of the user groups, and the viscosity of the activities and the long-tail users is enhanced.

Drawings

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

Fig. 1 is a flowchart of an activity recommendation method for a long-tailed user according to an embodiment of the present application;

fig. 2 is a flowchart of an activity recommendation method for a long-tailed user according to a second embodiment of the present application;

fig. 3 is a schematic structural diagram of an activity recommendation device for a long-tailed user according to a third embodiment of the present application;

fig. 4 is a schematic structural diagram of a terminal according to an embodiment of the present application.

With the above figures, there are shown specific embodiments of the present application, which will be described in more detail below. These drawings and written description are not intended to limit the scope of the inventive concepts in any manner, but rather to illustrate the inventive concepts to those skilled in the art by reference to specific embodiments.

Detailed Description

Reference will now be made in detail to the exemplary embodiments, examples of which are illustrated in the accompanying drawings. The following description refers to the accompanying drawings in which the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present application, as detailed in the appended claims.

The following describes the technical solutions of the present application and how to solve the above technical problems with specific embodiments. The following several specific embodiments may be combined with each other, and details of the same or similar concepts or processes may not be repeated in some embodiments. Embodiments of the present application will be described below with reference to the accompanying drawings. In the technical scheme of the application, the data acquisition, storage, use, processing and the like all conform to relevant regulations of national laws and regulations

Fig. 1 is a flowchart of an activity recommendation method for a long-tailed user according to an embodiment of the present application. As shown in fig. 1, the method comprises the following steps:

and step S101, grouping the obtained long-tail users to obtain a plurality of user groups.

Optionally, before grouping the obtained long-tailed users to obtain a plurality of user groups, the method further includes: acquiring asset management scale values of current operation users; and taking the operation user with the asset management scale value in the specified range as the long-tail user.

Particularly, as the user terminal is currently operated by the financial institution, the scale of the long-tail users, namely the middle-low end users, is very large, and therefore, the efficient and accurate activity recommendation for the users can significantly increase the overall benefit of the financial institution. Therefore, before activity recommendation is performed for the long-tailed user, the long-tailed user needs to be screened from the current operation users, and since the asset management specification value can represent the total assets hosted by each operation user in the financial institution, the screening can be performed specifically with reference to the asset management specification value. For example, the asset management specification value of each current operation user is obtained, and the operation users with the asset management specification values within 0-20 ten thousand are taken as long-tailed users. Of course, the long-tailed users may also be screened according to the activity of the operation users, and the specific screening method of the long-tailed users is not limited in this embodiment.

Optionally, the grouping the obtained long-tail users to obtain a plurality of user groups includes: extracting user designated characteristics of the long-tail user, wherein the user designated characteristics comprise a payment mode, activity preference and activity participation time; and grouping the long-tail users according to the specified characteristics of the users to obtain a plurality of user groups.

After the long-tail users are obtained, targeted grouping breakthrough can be selected for the large-population long-tail users, user designated features of the long-tail users can be extracted when the long-tail users are grouped, the user designated features comprise payment modes, activity preference and activity participation time, and therefore the long-tail users with the activity preference can be clustered into a user group when the long-tail users are grouped according to the user designated features. In addition, since the long-tailed user has a case of not participating in the activity, the acquired user group includes a user who has participated in the activity and a user who has not participated in the activity, and the major activities attended by the users who have participated in the activity are the same, but the specific minor activities attended may be different.

For example, a user group a and a user group B are obtained by grouping long-tailed users, and the user group a includes users who purchase financial products a and users who do not purchase financial products a, wherein the users who purchase financial products may specifically be different subclasses of financial products a1, a2, a3, a4 and the like under the purchased financial products a, and the users who do not purchase financial products a have similar payment modes and other characteristics to the purchased financial products a in the user group a, so that the probability that the users who do not purchase financial products a in the user group a purchase financial products a is relatively high; the user group B includes users participating in the B-type periodic deposit and users not participating in the B-type periodic deposit, wherein the users participating in the B-type periodic deposit may specifically participate in different types of subclasses of periodic deposits B1, B2, B3, and the like, and the users not participating in the B-type periodic deposit have a similar payment mode and the like with the users already participating in the B-type periodic deposit in the user group B, so that the probability that the users not participating in the B-type periodic deposit in the user group B participate in the B-type periodic deposit is relatively high.

And S102, respectively establishing a recommendation potential model for each user group, and acquiring a potential list of the user groups by adopting the recommendation potential model.

Optionally, the establishing a recommendation potential model for each user group includes: respectively sampling data for each user group to obtain users who have participated in the activity; performing characteristic screening according to users participating in the activity aiming at each user group to obtain target user group characteristics; and training according to the characteristics of the target user group aiming at each user group, and establishing a recommendation potential model matched with the user group.

Specifically, in the embodiment, users who have participated in activities in each user group are used as samples, and training is performed according to the samples to obtain the recommendation potential models matched with the user groups, so that the recommendation potential models correspond to the user groups, that is, the number of the constructed recommendation potential models is consistent with that of the user groups. And for each user group, predicting the users who do not participate in the activities by adopting the matched recommendation potential model to obtain the potential values of the users who do not participate in the activities for participating in the large-class activities, and eliminating the users who do not participate in the activities from the highest potential values to the lowest potential values to obtain a user group potential list, so that the potential list of the users who do not participate in the activities for participating in the large-class activities comprises the potential values of the users who do not participate in the activities.

In a specific implementation, aiming at a user group A, a corresponding major activity is to purchase a financial product a, a recommendation potential model X is constructed by using users who have purchased the financial product a in the user group A, the users who have not purchased the financial product a in the user group A are input into the model X, a potential value of the users who have not purchased the financial product a for purchasing the financial product a is obtained, and the users who have not purchased the financial product a are sequenced according to the potential value from large to small to obtain a user group potential list aiming at the user group A. Of course, in this embodiment, the user group potential list of the user group a is only obtained for illustration, and the method for obtaining the user group potential list of other user groups is substantially the same, and the details are not repeated in this embodiment.

And step S103, recommending the activity of the users who do not participate in the activity according to the user group potential name lists matched with the user groups.

Optionally, before performing activity recommendation on a user who does not participate in an activity according to the user group potential roster matched with each user group, the method further includes: acquiring subclass activities corresponding to users who have participated in the activities in each user group, wherein the major activities comprise at least two subclass activities; determining the participation times of each subclass activity, and sequencing the subclass activities in each user group according to the sequence of the participation times from high to low to obtain an activity sequence; and acquiring the activity to be recommended aiming at the user group according to the activity sequence.

It should be noted that, in the present embodiment, the activity to be recommended for the user group is also determined according to the users already participating in the activity in the user group. For example, for the user group a, the user who has purchased the financial product a may specifically be a minor financial product a1, a2, a3 or a4 under the financial product, the purchase frequency of each minor financial product is calculated, the activity sequence 1 is obtained by excluding each minor financial product in the user group a in the order from high purchase frequency to low purchase frequency, and the activity to be recommended for the user group a is obtained according to the activity sequence 1.

When determining the activities to be recommended of the user group a according to the activity sequence 1, the number of the activities may be preset, and a specified number of the small financial products may be screened in the order from front to back in the activity sequence 1 according to the number of the activities, for example, the specified number may be specifically 3, that is, the first 3 small financial products a2, a3, and a4 in the activity sequence 1 are screened as the activities to be recommended for the user group a. Of course, in this embodiment, only the to-be-recommended activities of the user group a are obtained as an example, and the way of obtaining the to-be-recommended activities of other user groups is substantially the same as this, and details are not repeated in this embodiment.

Optionally, the activity recommendation is performed on the users who do not participate in the activity according to the user group potential roster matched with each user group, where the activity recommendation includes: determining an interval range in which the potential value of each user in the user group potential list matched with each user group is located; determining recommendation modes according to the interval ranges, wherein different interval ranges respectively correspond to different recommendation modes; recommending the activities to be recommended to the users who do not participate in the activities by adopting a recommending mode.

After the potential lists of the user groups matched with the user groups are obtained, a recommendation mode can be determined according to the range of the potential values, for example, for the potential list of the user group corresponding to the user group A, users with potential values in the first 10% range actively carry out customer demand mining, add enterprise WeChat to the greatest extent, and carry out one-to-one service in an online mode; adopting an outbound mode in the first 20-30% interval of the potential value; and the potential value can be reminded by a short message within the range of 30-60%. Therefore, a more direct recommendation mode is adopted for users who purchase the financial products a with a higher probability, and a more graceful recommendation mode is adopted for users who purchase the financial products a with a lower probability, so that the user aversion is avoided.

It should be noted that, after determining the recommendation manner for the users who do not participate in the activity in each user group, specifically, the to-be-recommended activity determined by each user group is recommended to the users who do not participate in the activity according to the recommendation manner, for example, for the user group a, the to-be-recommended activities are recommended to the users who do not participate in the activity according to the determined recommendation manner according to the small financial products a2, a3, and a4 which are purchased frequently among the users who have purchased the financial products, so that the recommendation accuracy can be further increased, that is, the probability of purchasing the small financial products a2, a3, and a4 for the users who participate in the activity is increased.

According to the method and the device, the long-tail users are grouped to obtain the plurality of user groups, each user group corresponds to different large-class activities, the recommendation potential model is established by taking the users who participate in the activities as samples for each user group, the list containing the potential values of the users who do not participate in the activities and participate in the large-class activities is obtained through the established model, the estimation conditions of the users who do not participate in the activities and participate in the large-class activities can be obtained according to the potential list of the user groups, therefore, accurate and efficient recommendation can be conducted on the users according to the potential list of the user groups, and the stickiness of the activities and the long-tail users is enhanced.

Example two

Fig. 2 is a fourth flowchart of an activity recommendation method for a long-tailed user according to this embodiment. The embodiment mainly describes specifically that, in the first embodiment, a recommendation potential model is respectively established for each user group, as shown in fig. 2, the method includes:

step S201, data sampling is performed for each user group to obtain users who have participated in the activity.

Specifically, in the embodiment, for each user group, data sampling is performed on users in the user group, specifically, users who have participated in an activity are extracted from the user group, and the users who have participated in the activity are used as sample data, so that the model can be conveniently constructed according to the sample data.

It should be noted that, when data sampling is performed, there may be a case where the users are seriously out of order in positive and negative proportion, for example, the number of users already participating in an activity in the user group is significantly less than the number of users not participating in the activity, or the number of users already participating in the activity is significantly greater than the number of users not participating in the activity. At this time, in order to ensure the accuracy of model construction, oversampling and undersampling are adopted to make the proportion of the two as close as possible. The oversampling is to repeatedly select a few types and sample a few positive samples with a put back; and the undersampling is to randomly select a small amount of more negative samples and to simply and randomly sample more negative samples. By the method, the sample data obtained by sampling is more reasonable, so that the model constructed according to the sample data is more accurate.

Step S202, the characteristics of the target user group are obtained by screening the characteristics of the users participating in the activity aiming at each user group.

Optionally, the screening of the user group characteristics according to the users who have participated in the activity for each user group includes: performing characteristic screening according to services for users who have participated in activities in each user group to obtain a first characteristic set of the user group; screening out the characteristics with missing fields in the first characteristics of the user group to obtain a second characteristic set of the user group; calculating the information value of each feature in the second feature set of the user group, screening out the features of which the information values are smaller than a preset threshold value, and acquiring a third feature set of the user group; and acquiring the target user group characteristics according to the third characteristic set of the user group aiming at each user group.

Optionally, the obtaining, for each user group, the target user group feature according to the third feature set of the user group includes: screening the third feature set of the user group through the preliminary classification model to obtain a fourth feature set of the user group according with the preliminary classification model; obtaining the correlation among all the features in the fourth feature set of the user group, and screening the fourth features of the user group according to the correlation to obtain a fifth feature set of the user group; and taking the characteristics contained in the fifth characteristic set of the user group as the characteristics of the target user group.

Specifically, after users who have participated in the activity are obtained by sampling for each user group, the users who have participated in the activity are used as sample data, and the sample data is subjected to feature screening to obtain target user group features for the user group, so that the recommendation potential model can be conveniently established according to the screened target user group features.

The method comprises the steps of carrying out multi-layer screening when the characteristic screening is carried out on sample data, specifically, carrying out the characteristic screening according to a service to obtain a first characteristic set of a user group, and screening out the characteristics which are strongly related or unrelated to the large-class activities corresponding to the user group in the sample data according to the service screening, so that the interference of the characteristics which are strongly related to a target on model construction is avoided. After the first feature set of the user group is obtained, the missing condition of each feature field needs to be analyzed, and for the partial fields with higher missing rate, features with serious missing feature fields need to be deleted according to the selected threshold value to obtain a second feature set of the user group, so that the integrity of the extracted features is ensured. And using an Information Value (IV) in the obtained second feature set to represent the degree of contribution of the feature to the target prediction, i.e. the prediction capability of the feature, and generally, the higher the IV value is, the stronger the prediction capability of the feature is, the higher the information contribution degree is. While WOE, i.e., evidence weight value, may be used in calculating the IV value for each feature. Where the WOE representation describes the relationship between one predictable variable and two categorical variables. The specific principle of calculating the IV value based on the WOE is not important in the present application, and therefore, the description thereof will not be repeated in this embodiment. The present embodiment is merely an example, and a specific mode of calculating the IV value of each feature is not limited. And after the IV value of each feature in the second feature set of the user group is obtained, screening out the features of which the IV values are smaller than a preset threshold value, thereby obtaining a third feature set of the user group.

It should be noted that, in this embodiment, after the third feature set of the user group is obtained, the third feature set of the user group is screened through the preliminary classification model. When the preliminary classification model is adopted for screening, the characteristics of the non-mold entering are screened out, and the characteristics of the mold entering are reserved. And then adding normal distribution random numbers into the reserved features, checking the variable sequence of each modeling result through 5-fold cross validation, and deleting all the features ranked behind the random numbers to obtain a fourth feature set of the user group conforming to the preliminary classification model. And after a fourth feature set of the user group is obtained, calculating the correlation between the features, when the correlation between the two features is larger than 0.6, determining the importance sequence of the two features in the preliminary classification model according to the IV value, and screening out the features with lower importance, so as to screen out the features with high correlation and relatively lower importance in the fourth feature set to obtain a fifth feature set, and taking the obtained fifth feature set as the target user group feature. Of course, this embodiment is merely an example, and the specific filtering method of the target user characteristics is not limited.

Step S203, training is carried out according to the characteristics of the target user group aiming at each user group, and a recommendation potential model matched with the user group is established.

After the target user group characteristics are obtained for each user group, training can be performed according to the target user group characteristics to obtain a recommendation potential model matched with the user group. During training, GBDT gradient boosting decision tree modeling can be specifically adopted, and model parameters are adjusted through GridSearch. The GBDT is a Boosting algorithm, and the main idea is that a serial method is adopted during training of the base classifiers, the base classifiers are overlapped layer by layer, and each layer gives higher weight to a sample with a wrong base classifier in the previous layer during training. The specific operation principle of GBDT is not the focus of the present application, and therefore, the detailed description thereof is omitted in this embodiment.

According to the method and the device, the long-tail users are grouped to obtain the plurality of user groups, each user group corresponds to different large-class activities, the recommendation potential model is established by taking the users who participate in the activities as samples for each user group, the list containing the potential values of the users who do not participate in the activities and participate in the large-class activities is obtained through the established model, the estimation conditions of the users who do not participate in the activities and participate in the large-class activities can be obtained according to the potential list of the user groups, therefore, accurate and efficient recommendation can be conducted on the users according to the potential list of the user groups, and the stickiness of the activities and the long-tail users is enhanced. And according to users who have participated in activities in each user group, more accurate target user group characteristics are obtained through multi-layer characteristic screening, so that a recommendation potential model constructed according to the target user group characteristics is more accurate.

EXAMPLE III

Fig. 3 is a schematic structural diagram of an activity recommendation device for a long-tailed user according to an embodiment of the present application. As shown in fig. 3, the apparatus includes: a grouping module 310, a user group potential list obtaining module 320 and an activity recommending module 330.

A clustering module 310, configured to cluster the obtained long-tailed users to obtain multiple user groups, where each user group corresponds to a different large category of activities;

the user group potential list acquiring module 320 is configured to respectively establish a recommendation potential model for each user group, and acquire a user group potential list by using the recommendation potential model, where the user group potential list includes potential values of users who do not participate in activities and participate in the major-category activities;

and the activity recommendation module 330 is configured to perform activity recommendation on users who do not participate in the activity according to the user group potential name lists matched with the user groups.

Optionally, the device further includes a long-tail user obtaining module, configured to obtain asset management model values of the current operation users;

and taking the operation user with the asset management scale value in the specified range as the long-tail user.

Optionally, the clustering module is configured to extract user-specified features of the long-tailed user, where the user-specified features include a payment method, activity preferences, and activity participation time;

and grouping the long-tail users according to the specified characteristics to obtain a plurality of user groups.

Optionally, the user group potential list obtaining module further includes a recommendation potential model establishing sub-module, configured to perform data sampling for each user group to obtain users who have participated in the activity;

performing characteristic screening according to users participating in the activity aiming at each user group to obtain target user group characteristics;

and training according to the characteristics of the target user group aiming at each user group, and establishing a recommendation potential model matched with the user group.

Optionally, the recommendation potential model building submodule is configured to perform feature screening according to the service for the users who have participated in the activity in each user group to obtain a first feature set of the user group;

screening out the characteristics with missing fields in the first characteristics of the user group to obtain a second characteristic set of the user group;

calculating the information value of each feature in the second feature set of the user group, and screening out the features with the information values smaller than a preset threshold value to obtain a third feature set of the user group;

and acquiring the target user group characteristics according to the third characteristic set of the user group aiming at each user group.

Optionally, the recommendation potential model establishing sub-module is further configured to screen the third feature set of the user group through the preliminary classification model, and obtain a fourth feature set of the user group conforming to the preliminary classification model;

obtaining the correlation among all the features in the fourth feature set of the user group, and screening the fourth features of the user group according to the correlation to obtain a fifth feature set of the user group;

and taking the characteristics contained in the fifth characteristic set of the user group as the characteristics of the target user group.

Optionally, the apparatus further includes an activity to be recommended obtaining module, configured to obtain sub-activities corresponding to users who have participated in the activities in each user group, where a large-scale activity includes at least two sub-activities;

determining the participation times of each subclass activity, and sequencing the subclass activities in each user group according to the sequence of the participation times from high to low to obtain an activity sequence;

and acquiring the activity to be recommended aiming at the user group according to the activity sequence.

Optionally, the activity recommendation module is configured to determine an interval range in which the potential value of each user in the user group potential list matched with each user group is located;

determining recommendation modes according to the interval ranges, wherein different interval ranges respectively correspond to different recommendation modes;

recommending the activities to be recommended to the users who do not participate in the activities by adopting a recommending mode.

The activity recommendation device for the long-tailed user provided by the embodiment of the application can be used for executing the technical scheme of the activity recommendation method for the long-tailed user in the embodiment, and the implementation principle and the technical effect are similar, and are not described again.

It should be noted that the division of the modules of the above apparatus is only a logical division, and the actual implementation may be wholly or partially integrated into one physical entity, or may be physically separated. And these modules can be realized in the form of software called by processing element; or may be implemented entirely in hardware; and part of the modules can be realized in the form of calling software by the processing element, and part of the modules can be realized in the form of hardware. In addition, all or part of the modules can be integrated together or can be independently realized. The processing element here may be an integrated circuit with signal processing capabilities. In implementation, each step of the above method or each module above may be implemented by an integrated logic circuit of hardware in a processor element or an instruction in the form of software.

Example four

FIG. 4 shows a schematic block diagram of an electronic device 10 that may be used to implement an embodiment of the invention. Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device may also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be exemplary only, and are not meant to limit implementations of the inventions described and/or claimed herein.

As shown in fig. 4, the electronic device 10 includes at least one processor 11, and a memory communicatively connected to the at least one processor 11, such as a Read Only Memory (ROM) 12, a Random Access Memory (RAM) 13, and the like, wherein the memory stores a computer program executable by the at least one processor, and the processor 11 can perform various suitable actions and processes according to the computer program stored in the Read Only Memory (ROM) 12 or the computer program loaded from a storage unit 18 into the Random Access Memory (RAM) 13. In the RAM13, various programs and data necessary for the operation of the electronic apparatus 10 can also be stored. The processor 11, the ROM12, and the RAM13 are connected to each other via a bus 14. An input/output (I/O) interface 15 is also connected to bus 14.

A number of components in the electronic device 10 are connected to the I/O interface 15, including: an input unit 16 such as a keyboard, a mouse, or the like; an output unit 17 such as various types of displays, speakers, and the like; a storage unit 18 such as a magnetic disk, an optical disk, or the like; and a communication unit 19 such as a network card, modem, wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information/data with other devices via a computer network such as the internet and/or various telecommunication networks.

The processor 11 may be a variety of general and/or special purpose processing components having processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), various specialized Artificial Intelligence (AI) computing chips, various processors running machine learning model algorithms, a Digital Signal Processor (DSP), and any suitable processor, controller, microcontroller, or the like. The processor 11 performs the various methods and processes described above, such as activity recommendation methods for long-tailed users.

In some embodiments, the activity recommendation method for the long-tailed user may be implemented as a computer program tangibly embodied in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and/or installed onto the electronic device 10 via the ROM12 and/or the communication unit 19. When the computer program is loaded into the RAM13 and executed by the processor 11, one or more steps of the activity recommendation method for a long-tailed user described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured in any other suitable way (e.g. by means of firmware) as an activity recommendation method for a long-tailed user.

Various implementations of the systems and techniques described here above may be implemented in digital electronic circuitry, integrated circuitry, field Programmable Gate Arrays (FPGAs), application Specific Integrated Circuits (ASICs), application Specific Standard Products (ASSPs), system on a chip (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may be implemented

The method comprises the following steps: implemented in one or more computer programs that are executable and/or interpretable on a programmable system 5 including at least one programmable processor, which may be special or general purpose, receiving data and instructions from, and transmitting data and instructions to, a storage system, at least one input device, and at least one output device.

A computer program for implementing the methods of the present invention may be written in any combination of 0 of one or more programming languages. These computer programs may be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus, such that the computer programs, when executed by the processor, cause the functions/acts specified in the flowchart and/or block diagram block or blocks to be performed. A computer program can execute entirely on a machine, partly on a machine, as a stand-alone software package partly on a machine and partly on a remote machine or entirely on a remote machine or server.

In the context of the present invention, a computer-readable storage medium may be a tangible medium that can contain, or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. A computer readable storage medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or

Any suitable combination of the above. Alternatively, the computer readable storage medium may be a machine readable signal 0 medium. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

5 to provide for interaction with a user, the systems and techniques described here may be implemented on an electronic device,

the electronic device has: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to a user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which a user can provide input to the electronic device.

Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user can be 0 in any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form, including acoustic, speech, or tactile input.

The systems and techniques described here can be implemented in a computing system that includes a back-end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back-end, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local Area Networks (LANs), wide Area Networks (WANs), blockchain networks, and the internet.

The computing system may include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server can be a cloud server, also called a cloud computing server or a cloud host, and is a host product in a cloud computing service system, so that the defects of high management difficulty and weak service expansibility in the traditional physical host and VPS service are overcome.

EXAMPLE five

Embodiments of the present invention further provide a computer program product, including a computer program, where the computer program, when executed by a processor, implements an activity recommendation method for a long-tailed user as provided in any of the embodiments of the present application.

Computer program product in implementing the computer program code for carrying out operations of the present invention may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, smalltalk, C + +, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet service provider).

It is to be noted that the foregoing description is only exemplary of the invention and that the principles of the technology may be employed. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments illustrated herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, although the present invention has been described in greater detail by the above embodiments, the present invention is not limited to the above embodiments, and may include other equivalent embodiments without departing from the spirit of the present invention, and the scope of the present invention is determined by the scope of the appended claims.

Claims (15)

1. An activity recommendation method for a long-tailed user is characterized by comprising the following steps:

grouping the obtained long-tail users to obtain a plurality of user groups, wherein each user group corresponds to different major activities;

respectively establishing a recommendation potential model for each user group, and acquiring a user group potential list by adopting the recommendation potential model, wherein the user group potential list comprises potential values of users who do not participate in activities and participate in large-class activities;

and performing activity recommendation on users not participating in the activity according to the user group potential name lists matched with the user groups.

2. The method of claim 1, wherein before the clustering the obtained long-tailed users to obtain a plurality of user groups, further comprising:

acquiring asset management scale values of current operation users;

and taking the operation user with the asset management scale value in the specified range as the long-tail user.

3. The method of claim 2, wherein the clustering the obtained long-tailed users to obtain a plurality of user groups comprises:

extracting user specified characteristics of the long-tail user, wherein the user specified characteristics comprise a payment mode, activity preference and activity participation time;

and grouping the long-tail users according to the user designated characteristics to obtain a plurality of user groups.

4. The method of claim 1, wherein the separately building a recommendation potential model for each user group comprises:

respectively sampling data for each user group to obtain users who have participated in the activity;

performing characteristic screening according to the users participating in the activity aiming at each user group to obtain target user group characteristics;

and training according to the target user group characteristics aiming at each user group, and establishing a recommendation potential model matched with the user group.

5. The method of claim 4, wherein the performing user group feature filtering based on the users participating in the activity for each user group comprises:

performing characteristic screening according to services for users who have participated in activities in each user group to obtain a first characteristic set of the user group;

screening out the characteristics with missing fields in the first characteristics of the user group to obtain a second characteristic set of the user group;

calculating the information value of each feature in the second feature set of the user group, and screening out the features with the information values smaller than a preset threshold value to obtain a third feature set of the user group;

and aiming at each user group, acquiring the target user group characteristics according to the user group third characteristic set.

6. The method according to claim 5, wherein the obtaining the target user group characteristics according to the third characteristic set for each user group comprises:

screening the third feature set of the user group through a preliminary classification model to obtain a fourth feature set of the user group according with the preliminary classification model;

obtaining the correlation among the characteristics in the fourth characteristic set of the user group, and screening the fourth characteristics of the user group according to the correlation to obtain a fifth characteristic set of the user group;

and taking the characteristics contained in the fifth characteristic set of the user group as the characteristics of the target user group.

7. The method of claim 4, wherein before the activity recommendation of the non-active users according to the user group potential roster matched with each user group, further comprising:

obtaining subclass activities corresponding to the users who have participated in the activities in each user group, wherein the major activities comprise at least two of the subclass activities;

determining the participation times of each subclass activity, and sequencing the subclass activities in each user group according to the sequence of the participation times from high to low to obtain an activity sequence;

and acquiring the activity to be recommended aiming at the user group according to the activity sequence.

8. The method of claim 7, wherein the activity recommendation for the non-active users according to the user group potential roster matched with each user group comprises:

determining an interval range in which the potential value of each user in the user group potential list matched with each user group is located;

determining recommendation modes according to the interval ranges, wherein different interval ranges respectively correspond to different recommendation modes;

and recommending the activity to be recommended to the user not participating in the activity by adopting the recommending mode.

9. An activity recommendation device for a long-tailed user, comprising:

the grouping module is used for grouping the obtained long-tail users to obtain a plurality of user groups, wherein each user group corresponds to different large-class activities;

the system comprises a user group potential list acquisition module, a recommendation potential model generation module and a recommendation potential model generation module, wherein the user group potential list acquisition module is used for respectively establishing a recommendation potential model for each user group and acquiring a user group potential list by adopting the recommendation potential model, and the user group potential list comprises potential values of users not participating in activities and participating in large-class activities;

and the activity recommendation module is used for recommending the activity of the users not participating in the activity according to the user group potential name lists matched with the user groups.

10. The apparatus according to claim 9, further comprising a long-tailed user obtaining module, configured to obtain asset management model values of current operation users;

and taking the operation user with the asset management scale value in the specified range as the long-tail user.

11. The apparatus of claim 10, wherein the clustering module is configured to extract user-specified features of the long-tailed user, wherein the user-specified features include payment methods, activity preferences, and activity participation times;

and grouping the long-tail users according to the specified characteristics to obtain a plurality of user groups.

12. The apparatus of claim 9, wherein the user group potential list obtaining module further comprises a recommendation potential model building submodule for performing data sampling separately for each user group to obtain users who have participated in the activity;

performing characteristic screening according to the users participating in the activity aiming at each user group to obtain target user group characteristics;

and training according to the target user group characteristics aiming at each user group, and establishing a recommendation potential model matched with the user group.

13. A terminal, comprising: a processor, and a memory communicatively coupled to the processor;

the memory stores computer-executable instructions;

the processor executes computer-executable instructions stored by the memory to implement the method of any of claims 1-8.

14. A computer-readable storage medium having computer-executable instructions stored therein, which when executed by a processor, are configured to implement the method of any one of claims 1-8.

15. A computer program product, characterized in that it comprises a computer program which, when being executed by a processor, carries out the method of any one of claims 1-8.

Images