CN116186417B

CN116186417B - Recommendation method, recommendation device, computer equipment and storage medium

Info

Publication number: CN116186417B
Application number: CN202310453131.4A
Authority: CN
Inventors: 刘璇恒; 刘永威; 刘思喆
Original assignee: Beijing Apoco Blue Technology Co ltd
Current assignee: Beijing Apoco Blue Technology Co ltd
Priority date: 2023-04-25
Filing date: 2023-04-25
Publication date: 2023-07-04
Anticipated expiration: 2043-04-25
Also published as: CN116186417A

Abstract

The application relates to a recommendation method, a recommendation device, a computer device and a storage medium. The method comprises the following steps: acquiring the use frequency of a target object corresponding to a target user under a plurality of cycle types respectively; for any period type, determining at least one alternative resource corresponding to the period type from the alternative resources corresponding to the period type of the target object according to the use frequency of the target object corresponding to the period type; for any alternative resource, determining the predicted interaction probability of the target user for the alternative resource according to the alternative resource and the use frequency of the target object corresponding to each cycle type; for any period type, determining a target resource corresponding to the period type from at least one candidate resource corresponding to the period type according to the predicted interaction probability corresponding to each candidate resource; and recommending the resources to the target user according to the target resources corresponding to the cycle types. The method improves the interaction probability of the target user and the target resource.

Description

Recommendation method, recommendation device, computer equipment and storage medium

Technical Field

The present disclosure relates to the field of artificial intelligence technologies, and in particular, to a recommendation method, apparatus, computer device, and storage medium.

Background

In daily life, users often interact with resources that match the needs of the user. In order to improve the interaction rate of the target user, the target resource which possibly meets the user requirement is often screened out from the selectable resource set, and the target resource is recommended to the target user.

Specifically, the current recommendation technology determines a target user group where a target user is located by acquiring user information and/or resource interaction information of the target user and a preset user group division policy. Then, based on the target user group where the target user is located and the corresponding relation between the preset user group and the resources, determining the target resources corresponding to the target user, and recommending the target resources to the target user. The preset user group division strategy and the corresponding relation between the preset user group and the resource are stored in the terminal in advance. Illustratively, the user group partitioning policies include dimension partitioning from the age of the user, dimension partitioning from the registration duration, dimension partitioning from the interacted resources of the user, and the like.

However, the current recommendation technique determines the target resource based on a preset user group division policy and a corresponding relationship between the preset user group and the resource. Therefore, the predetermined user group division policy and the corresponding relation between the user group and the resource are difficult to estimate the dynamically changing and complex and diverse user demands, so that the target resource is difficult to match the user demands, and the interaction probability of the target user and the target resource is further reduced.

Disclosure of Invention

In view of the foregoing, it is desirable to provide a recommendation method, apparatus, computer device, computer readable storage medium, and computer program product that can improve the interaction probability of a target user with a target resource.

In a first aspect, the present application provides a recommendation method. The method comprises the following steps:

acquiring the use frequency of a target object corresponding to a target user under a plurality of cycle types respectively;

for any period type, determining at least one alternative resource corresponding to the period type from the alternative resources corresponding to the period type according to the use frequency of the target object corresponding to the period type;

for any alternative resource, determining the predicted interaction probability of the target user for the alternative resource according to the alternative resource and the use frequency of the target object corresponding to each cycle type;

for any cycle type, determining a target resource corresponding to the cycle type from at least one alternative resource corresponding to the cycle type according to the predicted interaction probability corresponding to each alternative resource;

and recommending the resources to the target user according to the target resources corresponding to the cycle types.

In one embodiment, the obtaining the usage frequency of the target object corresponding to the target user under the multiple period types includes:

determining a plurality of target dates according to the current date and the preset duration;

for any one of the target dates, determining the use frequency of the target user for each target object of the target objects on each of the target dates according to the cycle duration corresponding to each of the cycle types and the use frequency of the target user for each of the target objects on each of the target dates;

and according to any period type, determining the use frequency of the target object corresponding to the period type according to the use frequency of the target user for each target object of the target object on each target date under the period type.

In one embodiment, the selectable resource has a corresponding number of target object usages; according to the usage frequency of the target object corresponding to the period type, determining at least one alternative resource corresponding to the period type from the alternative resources corresponding to the period type by the target object, including:

And selecting the optional resource, corresponding to the period type, of which the usable frequency of the corresponding target object is greater than or equal to the usable frequency of the target object corresponding to the period type, as an alternative resource.

In one embodiment, the determining, for any one of the candidate resources, the predicted interaction probability of the target user for the candidate resource according to the candidate resource and the usage frequency of the target object corresponding to each cycle type includes:

and determining, for any one of the cycle types, a predicted interaction probability of the target user for the alternative resources according to the alternative resources and the target object usage frequency corresponding to each cycle type, when the number of the alternative resources corresponding to the cycle type is greater than or equal to a preset number.

In one embodiment, the method further comprises:

and regarding any cycle type, taking the alternative resource corresponding to the cycle type as the target resource corresponding to the cycle type under the condition that the number of the alternative resources corresponding to the cycle type is smaller than the preset number.

In one embodiment, the determining, according to the predicted interaction probability corresponding to each of the candidate resources, the target resource corresponding to the cycle type from at least one candidate resource corresponding to the cycle type includes:

determining initial resources corresponding to the periodic type from at least one of the alternative resources corresponding to the periodic type according to the predicted interaction probability corresponding to each of the alternative resources;

acquiring historical interaction data of the target user for each candidate resource, determining interacted resources from the candidate resources according to the historical interaction data, and determining interaction times of the target user and the interacted resources;

and taking the initial resource as a target resource corresponding to the period type under the condition that the interacted resource does not exist or the initial resource is included in the interacted resource.

In one embodiment, the method further comprises:

under the condition that the interacted resources exist and the initial resources are not included in the interacted resources, determining the prediction weight of each interacted resource according to the interaction times of the target user and each interacted resource;

For any interacted resource, determining a weighted interaction probability corresponding to the interacted resource according to the predicted weight of the interacted resource and the predicted interaction probability corresponding to the interacted resource;

and determining the target resource corresponding to the periodic type according to the weighted interaction probability corresponding to each interacted resource and the predicted interaction probability corresponding to the initial resource.

In one embodiment, the method further comprises:

constructing a training set, and training an interaction probability prediction model based on the training set to obtain a trained interaction probability prediction model, wherein the training set comprises a plurality of sample groups, the sample groups comprise the use frequency of target objects, sample selectable resources and labels of the sample selectable resources, which are respectively corresponding to sample users under each cycle type, when each sample is in a date, and the labels of the sample selectable resources are used for representing interaction conditions of the sample users and the sample selectable resources;

the determining, according to the candidate resources and the usage frequency of the target object corresponding to each cycle type, the predicted interaction probability of the target user for the candidate resources includes:

And inputting the resource data corresponding to the alternative resources and the use frequency of the target object corresponding to each cycle type into the trained interaction probability prediction model to obtain the predicted interaction probability of the target user for the alternative resources.

In one embodiment, the training the interaction probability prediction model based on the training set to obtain a trained interaction probability prediction model includes:

for any sample group, inputting the sample selectable resource and the use frequency of the target object corresponding to each sample user under each cycle type when each sample date is used to an interaction probability prediction model to obtain the predicted interaction probability of the sample selectable resource;

and training the interaction probability prediction model according to the predicted interaction probability of each sample selectable resource and the label of each sample selectable resource to obtain a trained interaction probability prediction model.

In one embodiment, the recommending the resource to the target user according to the target resource corresponding to each cycle type includes:

for any target resource, determining a basic adjustment coefficient for the target resource according to the usable times of a target object corresponding to the target resource and the use frequency of the target object of the periodic type corresponding to the target resource;

For any target resource, determining a target adjustment coefficient according to the basic adjustment coefficient, a preset coefficient adjustment strategy and a preset coefficient adjustment limiting strategy of the target resource, and determining target interaction resource data for the target resource according to preset interaction resource data corresponding to the target object, the target object usable times corresponding to the target resource and the target adjustment coefficient;

and recommending the resources to the target users according to the target resources corresponding to the cycle types and the target interaction resource data of the target resources.

In one embodiment, the preset coefficient adjustment limit policy includes a limit range; the preset coefficient adjustment strategy comprises an up-regulation range; the determining the target adjustment coefficient according to the basic adjustment coefficient, the preset coefficient adjustment policy and the preset coefficient adjustment limiting policy of the target resource includes:

determining a coefficient to be up-regulated according to the limiting range and the basic regulating coefficient;

determining an up-regulation value according to the up-regulation range, and determining a coefficient to be limited according to the up-regulation value and the coefficient to be up-regulated;

And determining the target adjustment coefficient according to the coefficient to be limited and the preset coefficient adjustment limiting strategy comprising the limiting range.

In one embodiment, the determining the coefficient to be up-regulated according to the limiting range and the basic adjustment coefficient includes:

judging whether the basic adjustment coefficient of the target resource is within the limit range;

taking the basic adjustment coefficient as a coefficient to be up-regulated under the condition that the basic adjustment coefficient is in the limit range; or,

taking the lower limit value of the limiting range as the coefficient to be up-regulated under the condition that the basic regulating coefficient is smaller than the lower limit value of the limiting range; or,

and taking the upper limit value of the limiting range as the coefficient to be up-regulated when the basic regulating coefficient is larger than the upper limit value of the limiting range.

In a second aspect, the present application further provides a recommendation device. The device comprises:

the acquisition module is used for acquiring the use frequency of the target object corresponding to the target user under a plurality of cycle types;

the first determining module is used for determining at least one alternative resource corresponding to the period type from the alternative resources corresponding to the period type of the target object according to the use frequency of the target object corresponding to the period type for any period type;

The prediction module is used for determining the predicted interaction probability of the target user for any alternative resource according to the alternative resource and the use frequency of the target object corresponding to each cycle type;

the second determining module is used for determining target resources corresponding to the cycle type from at least one alternative resource corresponding to the cycle type according to the predicted interaction probability corresponding to each alternative resource for any cycle type;

and the recommending module is used for recommending the resources to the target user according to the target resources corresponding to the cycle types.

In one embodiment, the acquiring module is specifically configured to:

In one embodiment, the selectable resource has a corresponding number of target object usages; the first determining module is specifically configured to:

In one embodiment, the prediction module is specifically configured to:

In one embodiment, the recommending device further includes:

and a third determining module, configured to, for any of the cycle types, use the candidate resource corresponding to the cycle type as the target resource corresponding to the cycle type when the number of candidate resources corresponding to the cycle type is less than the preset number.

In one embodiment, the second determining module is specifically configured to:

In one embodiment, the recommending device further includes:

a fourth determining module, configured to determine, when the interacted resource exists and the interacted resource does not include the initial resource, a predicted weight of each interacted resource according to the interaction times of the target user and each interacted resource;

a fifth determining module, configured to determine, for any of the interacted resources, a weighted interaction probability corresponding to the interacted resource according to the predicted weight of the interacted resource and the predicted interaction probability corresponding to the interacted resource;

And a sixth determining module, configured to determine a target resource corresponding to the period type according to the weighted interaction probability corresponding to each interacted resource and the predicted interaction probability corresponding to the initial resource.

In one embodiment, the recommending device further includes:

the construction module is used for constructing a training set and training an interaction probability prediction model based on the training set to obtain a trained interaction probability prediction model, wherein the training set comprises a plurality of sample groups, the sample groups comprise the use frequency of target objects, sample selectable resources and labels of the sample selectable resources, which are respectively corresponding to sample users under each period type, at each sample period, and the labels of the sample selectable resources are used for representing interaction conditions of the sample users and the sample selectable resources;

the prediction module is specifically configured to:

In one embodiment, the construction module is specifically configured to:

In one embodiment, the recommendation module is specifically configured to:

In one embodiment, the preset coefficient adjustment limit policy includes a limit range; the preset coefficient adjustment strategy comprises an up-regulation range; the recommendation module is specifically configured to:

In one embodiment, the recommendation module is specifically configured to:

In a third aspect, the present application also provides a computer device. The computer device comprises a memory storing a computer program and a processor implementing the steps of the first aspect when the processor executes the computer program.

In a fourth aspect, the present application also provides a computer-readable storage medium. The computer readable storage medium having stored thereon a computer program which, when executed by a processor, realizes the steps as described in the first aspect.

In a fifth aspect, the present application also provides a computer program product. The computer program product comprising a computer program which, when executed by a processor, performs the steps of the first aspect.

The recommendation method, the recommendation device, the computer equipment, the storage medium and the computer program product are characterized in that the use frequency of the target object respectively corresponding to the target user under a plurality of cycle types is obtained; for any period type, determining at least one alternative resource corresponding to the period type from the alternative resources corresponding to the period type of the target object according to the use frequency of the target object corresponding to the period type; for any alternative resource, determining the predicted interaction probability of the target user for the alternative resource according to the alternative resource and the use frequency of the target object corresponding to each cycle type; and determining a target resource corresponding to the cycle type from at least one alternative resource corresponding to the cycle type according to the predicted interaction probability corresponding to each alternative resource aiming at any cycle type. In the method, the alternative resources are obtained through coarse screening among the alternative resources corresponding to the period types according to the use frequency of the target object corresponding to the period type of the target user. Then, for any alternative resource, the predicted interaction probability of the alternative resource is determined according to the alternative resource and the use frequency of all target objects, and further the target resource is further obtained by screening from the alternative resources corresponding to the same period type according to the predicted interaction probability of the alternative resource. It can be understood that the usage frequency of the target object is used for representing the usage frequency of the target object by the target user in the period duration corresponding to the period type, and the usage frequency of the target object by the target user is dynamically changed along with the user demand, so that the matching degree of the target resource determined based on the usage frequency of the target object dynamically changed along with the user demand and the user demand is higher, and the interaction probability of the target user and the target resource is further improved. In addition, the target resource is further screened from the alternative resources corresponding to the same period type based on the predicted interaction probability of the alternative resources, so that the target resource with higher predicted interaction probability in the alternative resources can be obtained, the matching degree of the target user and the target resource is further improved, and the interaction probability of the target user and the target resource is further improved.

Drawings

FIG. 1 is a flow chart of a recommendation method in one embodiment;

FIG. 2 is a flowchart of a method for obtaining a usage frequency of a target object according to an embodiment;

FIG. 3 is a flow diagram of a method of determining predicted interaction probabilities and target resources in one embodiment;

FIG. 4 is a flow chart illustrating a method for determining a target resource according to another embodiment;

FIG. 5 is a flow chart illustrating a method for determining a target resource according to another embodiment;

FIG. 6 is a flow diagram of a method for determining target interaction resource data in one embodiment;

FIG. 7 is a flow chart illustrating a method for determining a target adjustment coefficient according to one embodiment;

FIG. 8 is a flowchart of a method for determining a coefficient to be up-regulated according to an embodiment;

FIG. 9 is a block diagram of a recommender in one embodiment;

fig. 10 is an internal structural view of a computer device in one embodiment.

Detailed Description

In order to make the objects, technical solutions and advantages of the present application more apparent, the present application will be further described in detail with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the present application.

In one embodiment, as shown in fig. 1, a recommendation method is provided, where the embodiment is applied to a terminal to illustrate, it is understood that the method may also be applied to a server, and may also be applied to a system including the terminal and the server, and implemented through interaction between the terminal and the server. In this embodiment, the method includes the steps of:

step 102, obtaining the use frequency of the target object corresponding to the target user under a plurality of cycle types.

In the embodiment of the application, for any period type, the terminal obtains the use frequency of the target object corresponding to the target user under the period type. The target object use frequency is used for representing the use times of the target object in the period duration corresponding to the period type, the resource data corresponding to the selectable resources comprise the period duration and the target object use times, different selectable resources can be included in the same period type, the period durations corresponding to the different selectable resources are the same, the period durations corresponding to the period type are the period durations corresponding to the period type, but the corresponding target object use times are different, and the resource data corresponding to the selectable resources are prestored in the terminal.

By way of example, the period duration corresponding to each period type may include 3 days, 5 days, 7 days, 10 days, 15 days, 20 days, 25 days, and 30 days, the target object usable times include 2 times, 3 times, 4 times, 5 times, 7 times, 10 times, 15 times, 20 times, and 30 times, the resource data corresponding to the optional resource include (3 days, 2 times), (3 days, 3 times), (3 days, 4 times), (5 days, 3 times), (5 days, 4 times), and (5 days, 5 times), and the like. Specifically, for any period type, the terminal obtains the use frequency of each target date corresponding to the period type for the target object of the target object, and takes the average value of the use frequency of each target date corresponding to the same period type for the target object of the target object as the use frequency of the target object corresponding to the period type. Or, for any period type, the terminal acquires a target object use frequency corresponding to the period type, and obtains the target object use frequency corresponding to the period type.

Step 104, for any period type, determining at least one alternative resource corresponding to the period type from the alternative resources corresponding to the period type according to the use frequency of the target object corresponding to the period type.

In the embodiment of the application, for any period type, the terminal determines, according to the usage frequency of the target object corresponding to the period type, an optional resource corresponding to the target object usable frequency matched with the usage frequency of the target object from the optional resources corresponding to the target object under the period type, and takes the optional resource corresponding to the target object usable frequency matched with the usage frequency of the target object as an optional resource. Optionally, the alternative resource may be an alternative resource whose corresponding target object usable frequency is greater than or equal to the target object usable frequency, or an alternative resource whose sum of the corresponding target object usable frequency and the preset frequency is greater than or equal to the target object usable frequency; the preset times are preset in the terminal, and the preset times can be negative numbers or non-negative numbers.

And 106, determining the predicted interaction probability of the target user for any alternative resource according to the alternative resource and the use frequency of the target object corresponding to each cycle type.

In the embodiment of the application, for any alternative resource, the terminal predicts the predicted interaction probability of the target user for the alternative resource according to the resource data corresponding to the alternative resource and the target object use frequency corresponding to all cycle types. The predicted interaction probability is used for representing the possibility of interaction between the target user and the alternative resource, and the higher the predicted interaction probability is, the higher the possibility of interaction between the target user and the alternative resource is, whereas the lower the predicted interaction probability is, the lower the possibility of interaction between the target user and the alternative resource is, and the range of the predicted interaction probability is [0,1].

Step 108, for any period type, determining a target resource corresponding to the period type from at least one candidate resource corresponding to the period type according to the predicted interaction probability corresponding to each candidate resource.

In the embodiment of the application, the terminal obtains the candidate resource with the highest predicted interaction probability from at least one candidate resource corresponding to each cycle type according to the predicted interaction probability corresponding to each candidate resource of the same cycle type, and takes the candidate resource with the highest predicted interaction probability as the target resource corresponding to the cycle type. By way of example, the resource data of the candidate resource of the cycle type 3 days includes { (3 days, 3 times), (3 days, 4 times) }, where the predicted interaction probability corresponding to the candidate resource of (3 days, 3 times) is 0.6 and the predicted interaction probability corresponding to the candidate resource of (3 days, 4 times) is 0.8, the terminal judges that 0.8 is greater than 0.6, and regards the candidate resource of the cycle type 0.8, that is, the candidate resource of (3 days, 4 times), as the target resource of the cycle type 3 days.

If the number of the candidate resources (called target candidate resources for convenience of distinction) with the highest predicted interaction probability is a plurality of, the terminal takes the target candidate resource with the highest usable frequency of the target object as the target resource. By way of example, the resource data of the candidate resource of the cycle type 3 days includes { (3 days, 3 times), (3 days, 4 times) }, where the predicted interaction probability corresponding to the candidate resource of (3 days, 3 times) is 0.6 and the predicted interaction probability corresponding to the candidate resource of (3 days, 4 times) is 0.6, then the terminal judges that the target candidate resource of the cycle type 3 days includes the candidate resource of (3 days, 3 times) and the candidate resource of (3 days, 4 times), and continues to judge that 4 times is greater than 3 times, and then the target object can use the candidate resource of the number of times 4 times (3 days, 4 times) as the target resource.

And 110, recommending the resources to the target user according to the target resources corresponding to each cycle type.

In the embodiment of the application, the terminal recommends resources to the target user according to the target resources corresponding to each period type. In one embodiment, the terminal sorts the target resources according to the cycle type corresponding to the target resources, obtains sorted target resources, and recommends the sorted target resources to the target user. Illustratively, the shorter (or longer) the period duration corresponding to the period type, the earlier the target resource is in the ordered target resource. In another embodiment, the terminal sorts the target resources according to the predicted interaction probability corresponding to the target resources, obtains sorted target resources, and recommends the sorted target resources to the target user. Illustratively, the higher the predicted interaction probability, the earlier the target resource is in the ordered target resources.

In the recommendation method, firstly, the alternative resources are obtained through coarse screening among the alternative resources corresponding to the period types according to the use frequency of the target object corresponding to the period type of the target user. Then, for any alternative resource, the predicted interaction probability of the alternative resource is determined according to the alternative resource and the use frequency of all target objects, and further the target resource is further obtained by screening from the alternative resources corresponding to the same period type according to the predicted interaction probability of the alternative resource. It can be understood that the usage frequency of the target object is used for representing the usage frequency of the target object by the target user in the period duration corresponding to the period type, and the usage frequency of the target object by the target user is dynamically changed along with the user demand, so that the matching degree of the target resource determined based on the usage frequency of the target object dynamically changed along with the user demand and the user demand is higher, and the interaction probability of the target user and the target resource is further improved. In addition, the target resource is further screened from the alternative resources corresponding to the same period type based on the predicted interaction probability of the alternative resources, so that the target resource with higher predicted interaction probability in the alternative resources can be obtained, the matching degree of the target user and the target resource is further improved, and the interaction probability of the user and the target resource is further improved.

In one embodiment, as shown in fig. 2, obtaining the usage frequency of the target object corresponding to the target user under the multiple cycle types includes:

step 202, determining a plurality of target dates according to the current date and the preset duration.

In the embodiment of the application, the terminal acquires the current date, and calculates the target date according to the current date and the preset duration. Wherein the target date is earlier than the current date. For example, the current date may be taken as an expiration date, a date of a preset time period from the current date before the expiration date is taken as a start date, and the target date is determined from among a plurality of dates between the start date and the expiration date, for example, a date between the start date and the expiration date is taken as a target date.

Step 204, for any one of the target dates, determining the frequency of use of the target user for each target object on the target date under each of the cycle types according to the cycle duration corresponding to each cycle type and the number of use of the target user for each target object on each of the target dates.

In the embodiment of the application, for any target date, the terminal determines each target period according to the period duration and the target date corresponding to each period type. Specifically, for any target date, the terminal takes the target date as the end date of the target period, and calculates the start date of the target period based on the period duration of the target date corresponding to each period type. It will be appreciated that the target period corresponds to a period of time equal to the period of time. For any target period of the target date, the terminal counts the total use times of the target user for the target object in the target period (namely, the use frequency of the target user for the target object in the target date when the target date is reached).

For example, the period duration corresponding to the period type 1 is 3 days, the period duration corresponding to the period type 2 is 5 days, and the target date is 3 months and 20 days, and for the period type 1, the terminal calculates the target period 1 based on 3 months and 20 days to obtain [3 months and 18 days, 3 months and 20 days ]; for cycle type 2, the terminal calculates the target period 2 based on 3 months and 20 days [3 months and 16 days, 3 months and 20 days ]. Assume that the number of times the target user uses the target object per day is shown in table 1. It will be appreciated that the data in table 1 is merely illustrative of the present solution and is not limiting of the data in practical applications.

TABLE 1

For the target period 1[3 month 18 day, 3 month 20 day ] of 3 month 20 day, the terminal obtains the respective usage times of the target user for the target object at 3 month 18 day, 3 month 19 day and 3 month 20 day, obtains 3 times, 4 times and 5 times, and calculates the total usage times of the target user for the target object at 3 month 18 day, 3 month 19 day and 3 month 20 day, that is, 3+4+5=12 times. The terminal uses 12 times as the target object use frequency of the target object when the target date is 3 months and 20 days. Similarly, for the target period 2[3 months 16 days, 3 months 20 days ] of 3 months 20 days, the terminal calculates that the total usage frequency of the target user for the target object in the target period 2 is 1+2+3+4+5=15 times. The terminal uses 15 times as the target object use frequency of the target object when the target date is 3 months and 20 days.

Step 206, for any period type, determining the usage frequency of the target object corresponding to the period type according to the usage frequency of the target user for each target object of the target object on each target date under the period type.

In the embodiment of the application, for any period type, the terminal determines the use frequency of the target object corresponding to the period type according to the use frequency of the target object of the target user for each target object in the target period corresponding to each target date under the period type. Specifically, the terminal uses the average value of the total use times of the target user for the target object in the target period corresponding to each target date under the period type as the use frequency of the target object corresponding to the period type.

For example, period type 1 corresponds to a period duration of 3 days, period type 2 corresponds to a period duration of 5 days, and the target date includes 3 months 20 days and 3 months 21 days. Referring to step 204, it can be seen that at day 3 month 21, the total number of uses of the target user for the target object is 4+5+6=15 times in the case of cycle type 1 (i.e., the target period is [ day 3 month 19, day 3 month 21 ]); on day 3 month 21, the total number of uses of the target user for the target object in the case of cycle type 1 (i.e., target period is [ day 3 month 17, day 3 month 21 ]) is 2+3+4+5+6=20 times. Then, under the cycle type 1, the terminal acquires the total usage number of 12 times for the target period (i.e., [3 month 18 day, 3 month 20 day ]) corresponding to the target date of 3 month 20 day, the total usage number of 15 times for the target period (i.e., [3 month 19 day, 3 month 21 day ]) corresponding to the target date of 3 month 21 day, and calculates the target object usage frequency of (12+15)/2=13.5 times for the cycle type 1. Similarly, under the period type 2, the terminal acquires the total usage number of 15 times for the target period (i.e., [3 month 16 day, 3 month 20 day ]) corresponding to the target date of 3 month 20 day, the total usage number of 20 times for the target period (i.e., [3 month 17 day, 3 month 21 day) corresponding to the target date of 3 month 21 day), and calculates the usage frequency of the target object corresponding to the period type 2 as (15+20)/2=17.5 times.

In this embodiment, a target date is determined according to a current date, and then, for any period type, the target object use frequency of a target user for a target object in a target period corresponding to each target date is counted, so as to obtain a target object use frequency corresponding to the period type. Therefore, the method and the device can determine the target object use frequency corresponding to the period type based on the target object use frequency of the target objects on a plurality of target dates under the same period type, so that data errors caused by one target date can be reduced, and the accuracy of the target object use frequency is improved.

In one embodiment, the selectable resources have corresponding target object numbers of times that can be used; according to the use frequency of the target object corresponding to the period type, determining at least one alternative resource corresponding to the period type from the alternative resources corresponding to the target object under the period type, wherein the method comprises the following steps:

and selecting the optional resource with the using frequency of the corresponding target object, which is larger than or equal to the using frequency of the target object corresponding to the period type, from the optional resources corresponding to the period type as an optional resource.

In the embodiment of the present application, the terminal may use, as the alternative resource, an alternative resource whose number of times of usage of the corresponding target object is greater than or equal to the number of times of usage of the corresponding target object of the period type, among the alternative resources corresponding to the period type. Illustratively, the resource data of the selectable resource of the cycle type 1 includes { (3 days, 2 times), (3 days, 3 times), (3 days, 4 times) }, and the target object corresponding to the cycle type is used 2.4 times. For the optional resources of the cycle type 1, the terminal judges that the usable times of the target object are 3 times and the usable times of the target object are 4 times respectively larger than the usable times of the target object by 2.4 times, and takes the corresponding optional resources (3 days and 3 times) and the corresponding optional resources (3 days and 4 times) as the optional resources of the cycle type 1. It can be understood that the determination manner of the alternative resources corresponding to the other period types is similar to that of the alternative resources of the period type 1, and will not be described again.

In this embodiment, the selectable resource with the target object usable frequency greater than or equal to the target object usable frequency is used as the candidate resource, that is, the target object usable frequency of the candidate resource is guaranteed to be sufficiently large, and because the target resource is determined based on the high target object usable frequency, the matching degree of the target resource and the target user is improved, and the interaction probability of the target user and the target resource is further improved.

In one embodiment, as shown in fig. 3, for any one of the alternative resources, determining, according to the alternative resource and the usage frequency of the target object corresponding to each cycle type, a predicted interaction probability of the target user for the alternative resource includes:

step 302, for any cycle type, when the number of the alternative resources corresponding to the cycle type is greater than or equal to the preset number, for any alternative resource corresponding to the cycle type, determining the predicted interaction probability of the target user for the alternative resource according to the alternative resource and the target object use frequency corresponding to each cycle type.

The method further comprises the steps of:

step 304, for any cycle type, determining a target resource corresponding to the cycle type from at least one candidate resource corresponding to the cycle type according to the predicted interaction probability corresponding to each candidate resource when the number of candidate resources corresponding to the cycle type is greater than or equal to the preset number. Or,

Step 306, regarding any cycle type, taking the alternative resource corresponding to the cycle type as the target resource corresponding to the cycle type when the number of the alternative resources corresponding to the cycle type is smaller than the preset number.

In the embodiment of the application, for any period type, the terminal compares the number of the alternative resources corresponding to the period type with a preset number. The preset number is a number preset and stored in the terminal, and the specific value can be determined by a person skilled in the art according to requirements, for example: the preset number may be 1 or 2, etc., where the preset number may be an upper limit value of the number of target resources for one cycle type.

It can be understood that when the number of the alternative resources corresponding to the same period type is greater than or equal to the preset number, the number of the alternative resources is excessive, so that the target resources need to be screened from the alternative resources, thereby reducing the number of the resources recommended to the target user and improving the accuracy of the recommendation; the number of the alternative resources corresponding to the same period type is smaller than the preset number, namely the number of the alternative resources is just smaller or the alternative resources are fewer, so that the target resources do not need to be screened from the alternative resources. In case that the number of alternative resources corresponding to the cycle type is greater than or equal to the preset number, the terminal performs step 302. In case that the number of alternative resources corresponding to the cycle type is smaller than the preset number, the terminal executes step 306.

For step 302, specifically, for any candidate resource corresponding to the cycle type, the terminal inputs the candidate resource and the usage frequency of the target object corresponding to all cycle types into the trained interaction probability prediction model, so as to obtain the predicted interaction probability of the target user for the candidate resource. The terminal then performs step 304. Specifically, for any period type, the terminal compares the predicted interaction probabilities of the alternative resources corresponding to the same period type, and takes the alternative resource corresponding to the highest predicted interaction probability among the alternative resources corresponding to the same period type as the target resource corresponding to the period type.

For step 306, the terminal uses the alternative resource corresponding to the period type as the target resource corresponding to the period type.

In this embodiment, when the number of the candidate resources corresponding to the same period type is smaller than the preset number, the candidate resource corresponding to the period type is directly used as the target resource corresponding to the period type. Therefore, under the condition that the number of the alternative resources corresponding to the same period type is smaller than the preset number, the determination of the target resources can not depend on the predicted interaction probability of the alternative resources, so that the processing procedure for determining the predicted interaction probability of the alternative resources is saved, and the determination efficiency of the target resources is improved.

In one embodiment, as shown in fig. 4, according to the predicted interaction probability corresponding to each candidate resource, determining the target resource corresponding to the cycle type from at least one candidate resource corresponding to the cycle type includes:

step 402, determining initial resources corresponding to the cycle type from at least one alternative resource corresponding to the cycle type according to the predicted interaction probability corresponding to each alternative resource.

In the embodiment of the application, the terminal compares the predicted interaction probabilities of the alternative resources corresponding to the same period type, and takes the alternative resource corresponding to the highest predicted interaction probability among the alternative resources corresponding to the same period type as the initial resource corresponding to the period type.

Step 404, obtaining historical interaction data of the target user for each candidate resource, determining interacted resources from the candidate resources according to the historical interaction data, and determining the interaction times of the target user and the interacted resources.

In the embodiment of the application, the terminal acquires historical interaction data of the target user aiming at each alternative resource. The historical interaction data comprise data generated when the target user interacts with the alternative resource, and the historical interaction data comprise data such as interaction date, interaction times and the like. And aiming at any alternative resource, under the condition that the historical interaction data of the target user aiming at the alternative resource exists, the terminal takes the alternative resource as an interacted resource and counts the interaction times of the interacted resource. In case that there is no interacted resource or an initial resource is included in the interacted resource, the terminal performs step 406. Alternatively, in the case that there are interacted resources and initial resources are not included in the interacted resources, the terminal performs step 502.

And step 406, taking the initial resource as a target resource corresponding to the period type when the interacted resource does not exist or the initial resource is included in the interacted resource.

In the embodiment of the application, the terminal takes the initial resource as the target resource corresponding to the period type under the condition that the interacted resource does not exist or the initial resource is included in the interacted resource. It can be understood that the interacted resource is an optional resource interacted by the target user, which indicates that the interacted resource has higher matching degree with the user requirement. When determining the target resource, the initial resource determined based on the predicted interaction probability is considered, and the interacted resource with higher matching degree with the user demand is considered, so that the target resource determined based on the multi-dimension has higher matching degree with the user demand.

For example, when there is no interacted resource, the initial resource is the optional resource that is most matched with the target user under the currently determined period type, so the initial resource can be used as the target resource corresponding to the period type, or when there is interacted resource and the interacted resource includes the initial resource, the initial resource is considered to be the optional resource that is most matched with the target user currently, so the initial resource can be used as the target resource corresponding to the period type.

In this embodiment, the target resource is determined together according to the predicted interaction probability corresponding to the candidate resource and the historical interaction data of the candidate resource, so that the target resource determined from multiple dimensions can reduce errors caused by a single dimension, thereby improving the interaction probability of the target resource.

In one embodiment, as shown in fig. 5, the method further comprises:

step 502, under the condition that interacted resources exist and initial resources are not included in the interacted resources, determining the prediction weight of each interacted resource according to the interaction times of the target user and each interacted resource.

In the embodiment of the application, under the condition that the interacted resource exists and the interacted resource does not include the initial resource, the initial resource is indicated not to be interacted with the target user in the history interaction process, and the matching degree of the interacted resource which is interacted with the target user in the history is higher than the target user requirement, so that the target resource can be determined from the interacted resource and the initial resource.

At this time, the predicted interaction probability corresponding to the interacted resource is lower than that of the initial resource, so that in order to better measure the adaptation condition of the interacted resource and the initial resource to the user, the predicted interaction probability of the interacted resource can be adjusted based on the interaction times, and as the interaction times can represent the adaptation degree of the target user and the interacted resource to a certain extent, that is, the more the interaction times are, the more the interest of the target user in the interacted resource is, that is, the higher the adaptation degree of the target user and the interacted resource is, the predicted interaction probability of the interacted resource can be adjusted for any interacted resource by the terminal according to the interaction times of the target user and the interacted resource, for example: the predicted weight of the interacted resource can be determined based on the interaction times of the target object and the interacted resource, the predicted interaction probability of the interacted resource is up-regulated based on the predicted weight, the predicted weight is larger than 1 and is positively related to the interaction times, namely, the more the interaction times are, the larger the corresponding weight is.

For example: the product of the interaction times and the preset proportion can be used as a weight adjustment value of the interacted resource. And taking the sum of the weight reference value and the weight adjustment value as the predicted weight of the interacted resource. For example: the weight basic value is 1, and the preset proportion is 10%. The preset proportion is preset in the terminal and is used for mapping the interaction times into weight adjustment values. It can be appreciated that the number of interactions of the interacted resource may reflect the preference degree of the target user for the interacted resource, and the interaction probability corresponding to the interacted resource with high preference degree is larger than the interaction probability corresponding to the resource with low preference degree or 0 preference degree. Therefore, the method and the device can determine the weight adjustment value through the interaction times, and can pay more attention to the interacted resources with high preference degree.

Step 504, for any interacted resource, determining a weighted interaction probability corresponding to the interacted resource according to the predicted weight of the interacted resource and the predicted interaction probability corresponding to the interacted resource.

In the embodiment of the application, for any interacted resource, the terminal calculates the product of the predicted weight of the interacted resource and the predicted interaction probability corresponding to the interacted resource to obtain the weighted interaction probability corresponding to the interacted resource, and because the predicted weight is greater than 1, the predicted interaction probability corresponding to the interacted resource is subjected to up-regulation processing through the predicted weight to obtain the weighted interaction probability corresponding to the interacted resource, namely the interacted resource is added at the angle of the historical interaction preference of the user. The present solution improves the predicted interaction probability of the interacted resource by predicting the weight, and in step 506, the probability of using the interacted resource with high preference degree as the target resource may be improved. It can be understood that the interacted resource with high preference degree is used as the target resource, so that the matching degree of the target user and the target resource can be ensured, and the interaction probability of the target user and the target resource is further improved.

Step 506, determining the target resource corresponding to the period type according to the weighted interaction probability corresponding to each interacted resource and the predicted interaction probability corresponding to the initial resource.

In the embodiment of the application, the terminal takes the resource (interacted resource or initial resource) corresponding to the highest interaction probability as the target resource corresponding to the period type in the interaction probabilities (including the weighted interaction probability corresponding to each interacted resource and the predicted interaction probability corresponding to the initial resource).

In one embodiment, the method further comprises:

and constructing a training set, and training the interaction probability prediction model based on the training set to obtain the trained interaction probability prediction model.

The training set comprises a plurality of sample groups, wherein the sample groups comprise the use frequency of a target object, sample selectable resources and labels of the sample selectable resources, which are respectively corresponding to sample users under each period type, at each sample date, and the labels of the sample selectable resources are used for representing interaction conditions of the sample users and the sample selectable resources. Optionally, the label of the sample optional resource may be used to characterize whether the sample user has interacted with the sample optional resource, and the label of the sample data may also be used to characterize the number of interactions of the sample user with the sample optional resource.

According to the candidate resources and the use frequency of the target object corresponding to each cycle type, determining the predicted interaction probability of the target user for the candidate resources comprises the following steps:

In an embodiment of the application, the terminal constructs a sample set and constructs a training set based on at least one sample set. Wherein the sample objects differ from sample group to sample group and/or the sample dates differ from sample group to sample group. Taking the construction of one sample group as an example, the construction of a plurality of sample groups is similar to the construction of a plurality of sample groups and will not be described in detail. The terminal acquires a plurality of sample dates, and determines the total use times of the sample user for the target object under each cycle type when the sample dates are determined according to the cycle duration corresponding to each cycle type and the use times of the sample user for the target object under each sample date aiming at the same sample date. It will be appreciated that, the method for determining the total number of uses of the target object by the sample user in each cycle type is similar to the method for determining the total number of uses of the target object by the target user in each cycle type, and specific reference may be made to step 204.

For any period type, the terminal determines the use frequency of the target object corresponding to the sample user under the period type according to the total use times of the sample user for the target object under each sample date and the period duration corresponding to the period type. For convenience of distinction, the use frequency of the target object corresponding to the sample user under the period type is called as the sample use frequency. It will be appreciated that the method for determining the frequency of use of the sample is similar to the method for determining the frequency of use of the target object, and reference is specifically made to step 206.

The terminal constructs a sample group based on the sample use frequency, one sample optional resource and the label of the sample optional resource of each period type corresponding to each sample date. And the terminal trains the interaction probability prediction model based on the training set to obtain the trained interaction probability prediction model. And for any alternative resource, the terminal inputs the resource data corresponding to the alternative resource and the use frequency of the target object corresponding to all cycle types into the trained interaction probability prediction model to obtain the predicted interaction probability of the target user for the alternative resource.

In this embodiment, a trained interaction probability prediction model is obtained based on training of a training set, and a predicted interaction probability of each candidate resource is determined based on the trained interaction probability prediction model, so that a precondition is provided for determining a target resource based on the predicted interaction probability of each candidate resource.

In one embodiment, training the interaction probability prediction model based on the training set, resulting in a trained interaction probability prediction model, comprises:

for any sample group, inputting sample selectable resources and the use frequency of target objects respectively corresponding to sample users under each period type at each sample date into an interaction probability prediction model to obtain the predicted interaction probability of the sample selectable resources; and training the interaction probability prediction model according to the predicted interaction probability of the selectable resources of each sample and the labels of the selectable resources of each sample to obtain a trained interaction probability prediction model.

In the embodiment of the application, for any sample group, the terminal inputs resource data corresponding to the sample selectable resource and the use frequency of all sample objects in the sample group into the interaction probability prediction model to obtain the predicted interaction probability of the sample selectable resource. It will be appreciated that the resource data corresponding to the sample selectable resources is the same as the resource data corresponding to the selectable resources, and the selectable resources in the model training stage are referred to as sample selectable resources only for distinguishing the model training stage from the actual model application stage. And training the interaction probability prediction model by the terminal according to the predicted interaction probability of the selectable resources of each sample and the labels of the selectable resources of each sample to obtain the loss value of the interaction probability prediction model. Wherein the loss function may be, but is not limited to, a cross entropy loss function and a maximum likelihood loss function. And the terminal determines the precision of the interaction probability prediction model according to the loss value of the interaction probability prediction model. And the terminal updates the parameters of the precision of the interaction probability prediction model by adopting a gradient descent algorithm according to the precision of the interaction probability prediction model until the precision of the interaction probability prediction model reaches a preset precision condition, so as to obtain the trained interaction probability prediction model.

In this embodiment, the interaction probability prediction model is trained through the predicted interaction probability of the sample selectable resource and the predicted interaction probability of the sample selectable resource, so as to obtain a trained interaction probability prediction model, and further, a precondition is provided for determining the predicted interaction probability of the candidate resource based on the trained interaction probability prediction model.

In one embodiment, as shown in fig. 6, according to the target resources corresponding to each cycle type, resource recommendation is performed to the target user, including:

step 602, for any target resource, determining a basic adjustment coefficient for the target resource according to the usable times of the target object corresponding to the target resource and the use frequency of the target object of the periodic type corresponding to the target resource.

In the embodiment of the application, for any target resource, the terminal calculates a quotient of the use frequency of the target object of the period type corresponding to the target resource and the usable frequency of the target object corresponding to the target resource, so as to obtain a basic adjustment coefficient of the target resource. For example, if the resource data corresponding to the target resource is (3 days, 4 times), the cycle type corresponding to the target resource is 3 days, and if the use frequency of the target object with the cycle type of 3 days is 2.4 times, the terminal calculates the quotient (i.e. 0.6) of the use frequency of the target object 2.4 times and the usable frequency of the target object 4 times, so as to obtain the basic adjustment coefficient with 0.6 as the target resource.

Step 604, for any target resource, determining a target adjustment coefficient according to the basic adjustment coefficient, the preset coefficient adjustment policy and the preset coefficient adjustment limit policy of the target resource, and determining target interaction resource data for the target resource according to preset interaction resource data corresponding to the target object, the usable times of the target object corresponding to the target resource and the target adjustment coefficient.

The preset coefficient adjustment limiting strategy includes a limiting range, and the preset coefficient adjustment limiting strategy is used for enabling the coefficient (including the coefficient to be up-regulated or the target adjustment coefficient) to be within the limiting range, and in one embodiment, the limiting range is [0.3,0.9]. The preset coefficient adjustment strategy includes an up-scaling range, which is used to up-scale the base adjustment coefficient, in one embodiment [0,0.2].

In the embodiment of the application, for any target resource, the terminal determines the coefficient to be up-regulated according to a preset coefficient regulation limiting strategy and a basic regulation coefficient of the target resource, determines the coefficient to be limited according to the coefficient to be up-regulated and the preset coefficient regulation strategy, and then determines the target regulation coefficient according to the preset coefficient regulation limiting strategy and the coefficient to be limited. The specific determination of the target adjustment coefficient refers to steps 702 to 706.

For any target resource, the terminal calculates the product of preset interaction resource data corresponding to the target object and the usable times of the target object corresponding to the target resource to obtain original interaction resource data, and calculates the product of the original interaction resource data and a target adjustment coefficient to obtain target interaction resource data for the target resource. The method comprises the steps that interaction resource data are preset and used for representing how much interaction resource data of a target user needs to interact with a terminal to use a target object once, original interaction resource data are used for representing how much interaction resource data of the target user needs to interact with the terminal to use the target object for a plurality of times, and the target interaction resource data are used for representing how much interaction resource data of the target user needs to interact with the terminal to obtain target resources.

For example, assuming that the resource data corresponding to the target resource is (3 days, 4 times), that is, the number of times that the target object corresponding to the target resource can be used is 4 times, the target adjustment coefficient corresponding to the target resource is 0.6, the unit of the resource data is "unit", the preset interactive resource data is x unit/time, the terminal calculates the product of the number of times that the target object can be used 4 times and the preset interactive resource data x unit/time to obtain the original interactive resource data 4x unit, and calculates the product of the original interactive resource data 4x unit and the target adjustment coefficient 0.6 to obtain the target interactive resource data 2.4x unit of the target resource.

Step 606, recommending the resources to the target users according to the target resources corresponding to each cycle type and the target interaction resource data of each target resource.

In the embodiment of the application, the terminal recommends the resources to the target user according to the target resources corresponding to each period type and the target interaction resource data of the target resources. Specifically, the resource recommendation method is similar to the resource recommendation method in step 110, and will not be described again, except that the target resources sequenced in step 606 based on step 110 include target interaction resource data of the target resources.

In this embodiment, a target resource and target interaction resource data corresponding to the target resource are recommended to a target user, wherein the target interaction resource data is determined based on a target adjustment coefficient, and the target adjustment coefficient is determined based on an adjustment coefficient, a preset coefficient adjustment policy and a preset coefficient adjustment limiting policy. The preset coefficient adjustment limiting strategy can enable the target adjustment coefficient to be within the limiting range, meanwhile, the preset coefficient adjustment strategy can adjust the basic adjustment coefficient upwards to obtain the target adjustment coefficient, so that the target adjustment coefficient can be reasonably increased (namely kept within the limiting range) while being increased on the basis of the basic adjustment coefficient, reasonable target interaction resource data can be recommended for target users, and the interaction probability of target resources is further improved.

In one embodiment, as shown in fig. 7, the preset coefficient adjustment limit policy includes a limit range; the preset coefficient adjustment strategy comprises an up-regulation range; the determining the target adjustment coefficient according to the basic adjustment coefficient, the preset coefficient adjustment policy and the preset coefficient adjustment limiting policy of the target resource includes:

step 702, determining a coefficient to be up-regulated according to the limit range and the basic regulation coefficient.

In the embodiment of the application, the terminal judges whether the basic adjustment coefficient belongs to the limiting range or not, and determines the coefficient to be adjusted upwards based on the judging result. Wherein the limiting range is a closed range (i.e. includes an upper limit value of the limiting range and a lower limit value of the limiting range), and the coefficient to be up-regulated is within the limiting range.

Step 704, determining an up-regulation value according to the up-regulation range, and determining a coefficient to be limited according to the up-regulation value and the coefficient to be up-regulated.

In the embodiment of the application, the terminal selects a number value as an up-regulation value in the up-regulation range, calculates the sum of the coefficient to be up-regulated and the up-regulation value, and obtains the coefficient to be limited. It is understood that the method of selecting a value in the up-regulation range as the up-regulation value is not limited in the present application, and any method of selecting a value in the up-regulation range as the up-regulation value is within the scope of protection of the present application. Illustratively, the terminal randomly selects a value in the up-regulation range as the up-regulation value.

Step 706, determining the target adjustment coefficient according to the coefficient to be limited and the preset coefficient adjustment limiting strategy including the limiting range.

In the embodiment of the application, the terminal judges whether the coefficient to be limited of the target resource is within the limiting range. And if the coefficient to be limited is within the limiting range, the terminal takes the coefficient to be limited as a target adjustment coefficient. Or if the coefficient to be limited is smaller than the lower limit value of the limiting range, the terminal takes the lower limit value of the limiting range as the target adjustment coefficient. Or if the coefficient to be limited is larger than the upper limit value of the limiting range, the terminal takes the upper limit value of the limiting range as a target adjustment coefficient. Wherein the target adjustment coefficient is within a limit.

In this embodiment, the coefficient to be adjusted up is determined according to the limiting range and the basic adjustment coefficient, and the coefficient to be limited is determined based on the coefficient to be adjusted up and the adjusting range, and then the target adjustment coefficient is determined according to the coefficient to be limited and the limiting range. It can be understood that the target adjustment coefficient after the processing is adjusted upwards as much as possible above the basic adjustment coefficient and also belongs to the limiting range, so that the scheme ensures that the target adjustment coefficient is not excessively adjusted while being adjusted upwards, and the benefit of a user is protected to a certain extent.

In one embodiment, as shown in fig. 8, the determining the coefficient to be up-regulated according to the limiting range and the basic adjustment coefficient includes:

step 802, determining whether the basic adjustment coefficient of the target resource is within the limit range.

In the embodiment of the present application, the terminal determines whether the basic adjustment coefficient of the target resource is within the limiting range, and if the basic adjustment coefficient is within the limiting range, the terminal executes step 804; if the basic adjustment coefficient is smaller than the lower limit value of the limiting range, the terminal executes step 806; if the base adjustment coefficient is greater than the upper limit of the limit range, the terminal performs step 808.

In step 804, the basic adjustment coefficient is used as the coefficient to be up-adjusted when the basic adjustment coefficient is within the limit range. Or,

in the embodiment of the application, the terminal takes the basic adjustment coefficient as the coefficient to be adjusted upwards under the condition that the basic adjustment coefficient is in the limit range. The terminal then performs step 704.

In step 806, in the case that the basic adjustment coefficient is smaller than the lower limit value of the limiting range, the lower limit value of the limiting range is taken as the coefficient to be adjusted. Or,

in the embodiment of the present application, when the basic adjustment coefficient is smaller than the lower limit value of the limiting range, the terminal takes the lower limit value of the limiting range as the coefficient to be adjusted. The terminal then performs step 704.

In step 808, in the case where the basic adjustment coefficient is greater than the upper limit value of the limit range, the upper limit value of the limit range is taken as the coefficient to be adjusted.

In the embodiment of the present application, when the basic adjustment coefficient is greater than the upper limit value of the limiting range, the terminal takes the upper limit value of the limiting range as the coefficient to be adjusted. The terminal then performs step 704.

In this embodiment, the target adjustment coefficient is determined according to the basic adjustment coefficient, the limiting range and the up-adjustment range, which provides a precondition for determining the target interaction resource data based on the target adjustment coefficient.

In order for those skilled in the art to better understand the embodiments of the present disclosure, the embodiments of the present disclosure will be described below with reference to the field of sharing bicycles.

In the step of selecting the target city, the terminal first acquires an initial city set. Wherein the set of initial cities comprises at least one initial city, which refers to a city for which the recommendation method is intended. Aiming at any initial city, the terminal acquires the use records of all the shared bicycles in the initial city, and calculates the single crown block effect of the initial city based on the use records. The single-sky vehicle effect refers to the average vehicle effect of each shared vehicle in the initial city in one day. In one embodiment, single overhead = number of uses of shared bicycles available for the day of the initial city/number of shared bicycles currently available for the initial city. Aiming at any initial city, the terminal calculates and obtains the average vehicle effect of the initial city according to at least one single-day vehicle effect. The average vehicle effect refers to the average vehicle effect of each sharing bicycle in the initial city every day.

Illustratively, assume that the single day vehicle effect of initial city 1 is shown in table 2 between 16 days 3 months and 21 days 3 months. It will be appreciated that the data in table 2 is merely illustrative of the present solution and is not limiting of the data in practical applications.

TABLE 2

Assuming that the average vehicle effect is determined based on only the single day vehicle effect between 16 days of 3 months and 21 days of 3 months, the terminal calculates the average vehicle effect of the initial city 1 to be (1+2+3+4+5+6)/6=3.5.

For any initial city, if the average vehicle effect corresponding to the initial city is within the preset vehicle effect range, the terminal takes the initial city as a target city, and adopts a recommendation method in the target city. The preset vehicle effect range is stored in the terminal in advance, and is used for measuring whether the average vehicle effect corresponding to the initial city is too low, too high or proper. It can be understood that if the average vehicle effect corresponding to the initial city is smaller than the lower limit value of the preset vehicle effect range, the average vehicle effect of the initial city is too low, which means that the shared bicycle of the initial city is in a stopped or almost stopped state, that is, the number of times of riding the shared bicycle by the person in the initial city is not large. If the average vehicle effect corresponding to the initial city is larger than the upper limit value of the preset vehicle effect range, the average vehicle effect of the initial city is too high, which means that the shared single vehicle of the initial city is in a saturated state. The aim of the scheme is to improve the interaction probability based on the recommendation method, namely the use probability of the shared bicycle, so that compared with an initial city with too high or too low average vehicle effect, the probability of improving the interaction probability of the initial city with moderate average vehicle effect is higher.

In the step of obtaining the usage frequency of the target object, the specific execution process of the terminal may refer to steps 202 to 206, taking table 1 as an example, where the terminal calculates that the usage frequency of the target object of the target user riding the shared bicycle (i.e. the target object) under the cycle type 1 is 12 times, and the usage frequency of the target object of the target user riding the shared bicycle (i.e. the target object) under the cycle type 2 is 15 times.

In the step of determining the alternative resource, the terminal uses the riding card (equivalent to the alternative resource) of the shared bicycle corresponding to the cycle type, wherein the corresponding available times (equivalent to the available times of the target object) of the shared bicycle is larger than or equal to the available times of the target object corresponding to the cycle type, as the alternative riding card (alternative resource).

For example, the cycle type 1 corresponding riding card includes a 3 day 2 card, a 3 day 3 card and a 3 day 4 card, the target object usage frequency of the target user a riding the shared bicycle under the cycle type 1 is 2.4 times (i.e. 2.4 times in every 3 days on average), and then the terminal determines that the usable number 2 corresponding to the 3 day 2 card is less than the target object usage frequency of 2.4, and then the 3 day 2 card is not used as the alternative riding card. Similarly, the terminal takes the 3-day 3-time card and the 3-day 4-time card as alternative riding cards.

In the step of determining the predicted interaction probability of the alternative resources, for any alternative riding card, the terminal inputs resource data (namely, several times in several days) corresponding to the alternative riding card and the use frequency of the target object corresponding to each cycle type into the trained interaction probability prediction model, and outputs the predicted interaction probability of the target user for the alternative riding card.

In the step of determining the target resource, for each candidate riding card corresponding to the same cycle type, the terminal determines the target riding card (corresponding to the target resource) corresponding to the cycle type according to the predicted interaction probability corresponding to each candidate riding card.

In the step of resource recommendation, the terminal recommends a target riding card to a target user according to the target riding card corresponding to each cycle type. Specifically, for any target riding card, the terminal calculates the quotient of the use frequency of the target object of the cycle type corresponding to the target riding card and the usable frequency of the shared bicycle corresponding to the target riding card, and obtains the lowest discount (equivalent to a basic adjustment coefficient) of the target riding card. For example, the target riding card is a 3 day 4 time card, and the target object with a period duration of 3 days corresponding to the period type uses 2.4 times, the terminal calculates 2.4/4=0.6, that is, the 6-fold is the lowest discount of the 3 day 4 time card.

The terminal determines whether the lowest discount falls within the discount limit range (equivalent to the limit range). If the lowest discount belongs to the discount limit range, the terminal takes the lowest discount as the discount to be up-regulated (equivalent to the coefficient to be up-regulated); if the lowest discount is smaller than the lower limit value of the discount limit range, the terminal takes the lower limit value of the discount limit range as the discount to be adjusted upwards; if the lowest discount is larger than the upper limit value of the discount limit range, the terminal takes the upper limit value of the discount limit range as the discount to be adjusted. In one embodiment, the limit is [0.3,0.9].

And the terminal determines an up-regulation value according to a preset turnbuckle up-regulation range (equivalent to the up-regulation range), calculates the sum of the discount to be up-regulated and the up-regulation value, and obtains the discount to be limited (equivalent to the coefficient to be limited). In one embodiment, the limit is [0,0.2].

The terminal determines whether the discount to be limited falls within a discount limit range (equivalent to a limit range). If the discount to be limited belongs to the discount limiting range, the terminal takes the discount to be limited as a target discount (corresponding to a target adjustment coefficient); if the discount to be limited is smaller than the lower limit value of the discount limiting range, the terminal takes the lower limit value of the discount limiting range as a target discount; if the discount to be limited is larger than the upper limit value of the discount limiting range, the terminal takes the upper limit value of the discount limiting range as a target discount.

The terminal calculates the product of the price of the passenger corresponding to the shared bicycle (corresponding to the preset interactive resource data) and the usable times of the target object corresponding to the target riding card, obtains the original price of the target riding card (corresponding to the original interactive resource data), and calculates the product of the original price and the target discount, so as to obtain the selling price (corresponding to the target interactive resource data) of the target riding card. Wherein, the guest price refers to the original selling price of the shared bicycle for single riding. Exemplary, the target riding card is 3 days 4 times card, the guest price is 2 yuan/time, the target discount is 0.8, the terminal calculates the original price of the 3 days 4 times card to be 4×2=8 yuan, and calculates the selling price of the 3 days 4 times card to be 8×0.8=6.4 yuan.

And the terminal recommends the target riding card containing the selling price corresponding to each period type to the target user.

It should be understood that, although the steps in the flowcharts related to the embodiments described above are sequentially shown as indicated by arrows, these steps are not necessarily sequentially performed in the order indicated by the arrows. The steps are not strictly limited to the order of execution unless explicitly recited herein, and the steps may be executed in other orders. Moreover, at least some of the steps in the flowcharts described in the above embodiments may include a plurality of steps or a plurality of stages, which are not necessarily performed at the same time, but may be performed at different times, and the order of the steps or stages is not necessarily performed sequentially, but may be performed alternately or alternately with at least some of the other steps or stages.

Based on the same inventive concept, the embodiment of the application also provides a recommendation device for realizing the recommendation method. The implementation of the solution provided by the device is similar to the implementation described in the above method, so the specific limitation of one or more embodiments of the recommendation device provided below may be referred to the limitation of the recommendation method hereinabove, and will not be repeated here.

In one embodiment, as shown in fig. 9, there is provided a recommending apparatus including:

the obtaining module 902 is configured to obtain usage frequencies of target objects corresponding to the target users under a plurality of period types respectively;

a first determining module 904, configured to determine, for any cycle type, at least one alternative resource corresponding to the cycle type from alternative resources corresponding to the cycle type, according to the usage frequency of the target object corresponding to the cycle type;

a prediction module 906, configured to determine, for any candidate resource, a predicted interaction probability of the target user for the candidate resource according to the candidate resource and a target object usage frequency corresponding to each cycle type;

a second determining module 908, configured to determine, for any cycle type, a target resource corresponding to the cycle type from at least one candidate resource corresponding to the cycle type according to the predicted interaction probability corresponding to each candidate resource;

And a recommending module 910, configured to recommend resources to the target user according to the target resources corresponding to each cycle type.

In the recommendation device, the target object use frequency is used for representing the use times of the target user on the target object in the period duration corresponding to the period type, and the use times of the target user on the target object are dynamically changed along with the user demand, so that the matching degree of the target resource and the user demand, which is determined based on the target object use frequency along with the user demand, is higher, and the interaction probability of the target user and the target resource is further improved. In addition, the target resource is further screened from the alternative resources corresponding to the same period type based on the predicted interaction probability of the alternative resources, so that the target resource with higher predicted interaction probability in the alternative resources can be obtained, the matching degree of the target user and the target resource is further improved, and the interaction probability of the target user and the target resource is further improved.

In one embodiment, the obtaining module 902 is specifically configured to:

for any target date, determining the use frequency of the target user for each target object on each cycle type according to the cycle duration corresponding to each cycle type and the use frequency of the target user for each target object on each target date;

And aiming at any period type, determining the use frequency of the target object corresponding to the period type according to the use frequency of the target user aiming at each target object of the target object on each target date under the period type.

In one embodiment, the selectable resources have corresponding target object numbers of times that can be used; the first determining module 904 is specifically configured to:

In one embodiment, the prediction module 906 is specifically configured to:

and under the condition that the number of the alternative resources corresponding to the cycle type is larger than or equal to the preset number, according to the alternative resources and the target object use frequency corresponding to each cycle type, determining the predicted interaction probability of the target user for the alternative resources for any alternative resource corresponding to the cycle type.

In one embodiment, the recommendation device further comprises:

and the third determining module is used for taking the alternative resources corresponding to the cycle type as target resources corresponding to the cycle type when the number of the alternative resources corresponding to the cycle type is smaller than the preset number for any cycle type.

In one embodiment, the second determining module 908 is specifically configured to:

determining initial resources corresponding to the periodic type from at least one alternative resource corresponding to the periodic type according to the predicted interaction probability corresponding to each alternative resource;

acquiring historical interaction data of a target user aiming at each alternative resource, determining interacted resources from the alternative resources according to the historical interaction data, and determining interaction times of the target user and the interacted resources;

In one embodiment, the recommendation device further comprises:

a fourth determining module, configured to determine, when the interacted resources exist and the interacted resources do not include the initial resources, a predicted weight of each interacted resource according to the interaction times of the target user and each interacted resource;

a fifth determining module, configured to determine, for any interacted resource, a weighted interaction probability corresponding to the interacted resource according to a predicted weight of the interacted resource and a predicted interaction probability corresponding to the interacted resource;

In one embodiment, the recommendation device further comprises:

the system comprises a building module, a training set and a sample selection module, wherein the building module is used for building the training set and training an interaction probability prediction model based on the training set to obtain a trained interaction probability prediction model, the training set comprises a plurality of sample groups, the sample groups comprise the use frequency of target objects, sample selection resources and labels of the sample selection resources, which are respectively corresponding to sample users under each cycle type, at each sample date, and the labels of the sample selection resources are used for representing interaction conditions of the sample users and the sample selection resources;

the prediction module 906 is specifically configured to:

In one embodiment, the building block is specifically configured to:

for any sample group, inputting sample selectable resources and the use frequency of target objects respectively corresponding to sample users under each period type at each sample date into an interaction probability prediction model to obtain the predicted interaction probability of the sample selectable resources;

and training the interaction probability prediction model according to the predicted interaction probability of the selectable resources of each sample and the labels of the selectable resources of each sample to obtain a trained interaction probability prediction model.

In one embodiment, the recommendation module 910 is specifically configured to:

for any target resource, determining a basic adjustment coefficient for the target resource according to the usable times of the target object corresponding to the target resource and the use frequency of the target object of the periodic type corresponding to the target resource;

for any target resource, determining a target adjustment coefficient according to a basic adjustment coefficient, a preset coefficient adjustment strategy and a preset coefficient adjustment limiting strategy of the target resource, and determining target interaction resource data for the target resource according to preset interaction resource data corresponding to a target object, the usable times of the target object corresponding to the target resource and the target adjustment coefficient;

and recommending the resources to the target users according to the target resources corresponding to each cycle type and the target interaction resource data of each target resource.

In one embodiment, the preset coefficient adjustment limit policy includes a limit range; the preset coefficient adjustment strategy comprises an up-regulation range; the recommendation module 910 is specifically configured to:

determining a coefficient to be up-regulated according to the limiting range and the basic adjustment coefficient;

And determining a target adjustment coefficient according to the coefficient to be limited and a preset coefficient adjustment limiting strategy comprising a limiting range.

In one embodiment, the recommendation module 910 is specifically configured to:

judging whether the basic adjustment coefficient of the target resource is within a limit range;

under the condition that the basic adjustment coefficient is in a limiting range, taking the basic adjustment coefficient as a coefficient to be adjusted upwards; or,

under the condition that the basic adjustment coefficient is smaller than the lower limit value of the limiting range, taking the lower limit value of the limiting range as the coefficient to be adjusted upwards; or,

and when the basic adjustment coefficient is larger than the upper limit value of the limiting range, taking the upper limit value of the limiting range as the coefficient to be adjusted.

The respective modules in the above recommendation device may be implemented in whole or in part by software, hardware, and combinations thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory in the computer device, so that the processor may call and execute operations corresponding to the above modules.

In one embodiment, a computer device is provided, which may be a terminal, and an internal structure diagram thereof may be as shown in fig. 10. The computer device includes a processor, a memory, an input/output interface, a communication interface, a display unit, and an input means. The processor, the memory and the input/output interface are connected through a system bus, and the communication interface, the display unit and the input device are connected to the system bus through the input/output interface. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The input/output interface of the computer device is used to exchange information between the processor and the external device. The communication interface of the computer device is used for carrying out wired or wireless communication with an external terminal, and the wireless mode can be realized through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. The computer program is executed by a processor to implement a recommendation method. The display unit of the computer device is used for forming a visual picture, and can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer equipment can be a touch layer covered on the display screen, can also be a key, a track ball or a touch pad arranged on the shell of the computer equipment, and can also be an external keyboard, a touch pad or a mouse and the like.

It will be appreciated by those skilled in the art that the structure shown in fig. 10 is merely a block diagram of some of the structures associated with the present application and is not limiting of the computer device to which the present application may be applied, and that a particular computer device may include more or fewer components than shown, or may combine certain components, or have a different arrangement of components.

In an embodiment, there is also provided a computer device comprising a memory and a processor, the memory having stored therein a computer program, the processor implementing the steps of the method embodiments described above when the computer program is executed.

In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored which, when executed by a processor, carries out the steps of the method embodiments described above.

In an embodiment, a computer program product is provided, comprising a computer program which, when executed by a processor, implements the steps of the method embodiments described above.

It should be noted that, the user information (including, but not limited to, user equipment information, user personal information, etc.) and the data (including, but not limited to, data for analysis, stored data, presented data, etc.) referred to in the present application are information and data authorized by the user or sufficiently authorized by each party, and the collection, use and processing of the related data are required to comply with the related laws and regulations and standards of the related countries and regions.

Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above. Any reference to memory, database, or other medium used in the various embodiments provided herein may include at least one of non-volatile and volatile memory. The nonvolatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical Memory, high density embedded nonvolatile Memory, resistive random access Memory (ReRAM), magnetic random access Memory (Magnetoresistive RandomAccess Memory, MRAM), ferroelectric Memory (Ferroelectric Random Access Memory, FRAM), phase change Memory (PhaseChange Memory, PCM), graphene Memory, and the like. Volatile memory can include random access memory (Random Access Memory, RAM) or external cache memory, and the like. By way of illustration, and not limitation, RAM can be in the form of a variety of forms, such as static random access memory (StaticRandom Access Memory, SRAM) or dynamic random access memory (Dynamic Random Access Memory, DRAM), and the like. The databases referred to in the various embodiments provided herein may include at least one of relational databases and non-relational databases. The non-relational database may include, but is not limited to, a blockchain-based distributed database, and the like. The processors referred to in the embodiments provided herein may be general purpose processors, central processing units, graphics processors, digital signal processors, programmable logic units, quantum computing-based data processing logic units, etc., without being limited thereto.

The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.

The above examples only represent a few embodiments of the present application, which are described in more detail and are not to be construed as limiting the scope of the present application. It should be noted that it would be apparent to those skilled in the art that various modifications and improvements could be made without departing from the spirit of the present application, which would be within the scope of the present application. Accordingly, the scope of protection of the present application shall be subject to the appended claims.

Claims

1. A recommendation method, the method comprising:

recommending resources to the target user according to the target resources corresponding to the cycle types;

determining, from at least one candidate resource corresponding to the cycle type, a target resource corresponding to the cycle type according to the predicted interaction probability corresponding to each candidate resource, including:

comparing the predicted interaction probability of each alternative resource corresponding to the period type, and taking the alternative resource corresponding to the highest predicted interaction probability in each alternative resource corresponding to the period type as an initial resource corresponding to the period type;

2. The method of claim 1, wherein the obtaining the usage frequency of the target object corresponding to the target user under the plurality of cycle types includes:

3. A method according to claim 1 or 2, wherein the selectable resources have a corresponding number of target object usages; according to the usage frequency of the target object corresponding to the period type, determining at least one alternative resource corresponding to the period type from the alternative resources corresponding to the period type by the target object, including:

4. The method according to claim 1 or 2, wherein the determining, for any one of the candidate resources, the predicted interaction probability of the target user for the candidate resource according to the candidate resource and the usage frequency of the target object corresponding to each of the cycle types includes:

5. The method according to claim 4, wherein the method further comprises:

6. The method according to claim 1, wherein the method further comprises:

7. The method according to claim 1 or 2, characterized in that the method further comprises:

8. The method of claim 7, wherein training the interaction probability prediction model based on the training set results in a trained interaction probability prediction model, comprising:

9. The method according to claim 1 or 2, wherein said recommending resources to the target user according to the target resources corresponding to each of the cycle types includes:

10. The method of claim 9, wherein the preset coefficient adjustment limit policy includes a limit range; the preset coefficient adjustment strategy comprises an up-regulation range; the determining the target adjustment coefficient according to the basic adjustment coefficient, the preset coefficient adjustment policy and the preset coefficient adjustment limiting policy of the target resource includes:

11. The method of claim 10, wherein the determining the coefficients to be up-regulated based on the limit range and the base adjustment coefficients comprises:

12. A recommendation device, the device comprising:

the first determining module is used for determining at least one alternative resource corresponding to any cycle type from the alternative resources corresponding to the cycle type of the target object according to the use frequency of the target object corresponding to the cycle type;

the recommending module is used for recommending resources to the target user according to the target resources corresponding to the cycle types;

the second determining module is specifically configured to:

comparing the predicted interaction probability of each alternative resource corresponding to the period type, and taking the alternative resource corresponding to the highest predicted interaction probability in each alternative resource corresponding to the period type as an initial resource corresponding to the period type; acquiring historical interaction data of the target user for each candidate resource, determining interacted resources from the candidate resources according to the historical interaction data, and determining interaction times of the target user and the interacted resources; and taking the initial resource as a target resource corresponding to the period type under the condition that the interacted resource does not exist or the initial resource is included in the interacted resource.

13. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the steps of the method of any one of claims 1 to 11 when the computer program is executed.

14. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any of claims 1 to 11.