CN112328894A

CN112328894A - Behavior guiding method and device, computer equipment and storage medium

Info

Publication number: CN112328894A
Application number: CN202011339735.9A
Authority: CN
Inventors: 孔洋洋; 王冉冉; 陈巍立; 田甘迅
Original assignee: Beijing Lexuebang Network Technology Co Ltd
Current assignee: Beijing Lexuebang Network Technology Co Ltd
Priority date: 2020-11-25
Filing date: 2020-11-25
Publication date: 2021-02-05

Abstract

The present disclosure provides a behavior guidance method, apparatus, computer device, and storage medium, the method comprising: acquiring behavior characteristic information of at least one user to be screened; the user to be screened is a user who does not execute the target behavior; screening target users meeting target behavior execution conditions from the at least one user to be screened based on the behavior characteristic information of the at least one user to be screened; acquiring labeled first personalized feature information aiming at the target user and labeled second personalized feature information aiming at a user to be matched; and determining a matching user corresponding to each target user from the users to be matched based on the acquired first characterization feature information of the target user, the second characterization feature information of the user to be matched, the behavior feature information of the target user and a first pre-estimation model, wherein the matching user is used for guiding the target user to execute the target behavior.

Description

Behavior guiding method and device, computer equipment and storage medium

Technical Field

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

Background

With the rapid development of internet technology, online education has been rapidly popularized. To encourage users to perform certain pre-set activities, such as encouraging user consumption, user stickiness is typically increased by the staff of the online education platform.

However, this approach is inefficient and less effective in directing.

Disclosure of Invention

The embodiment of the disclosure at least provides a behavior guiding method, a behavior guiding device, computer equipment and a storage medium.

In a first aspect, an embodiment of the present disclosure provides a behavior guidance method, including:

acquiring behavior characteristic information of at least one user to be screened; the user to be screened is a user who does not execute the target behavior;

screening target users meeting target behavior execution conditions from the at least one user to be screened based on the behavior characteristic information of the at least one user to be screened;

acquiring labeled first personalized feature information aiming at the target user and labeled second personalized feature information aiming at a user to be matched;

and determining a matching user corresponding to each target user from the users to be matched based on the acquired first characterization feature information of the target user, the second characterization feature information of the user to be matched, the behavior feature information of the target user and a first pre-estimation model, wherein the matching user is used for guiding the target user to execute the target behavior.

In a possible implementation manner, the screening, based on the behavior feature information of the at least one user to be screened, a target user meeting a target behavior execution condition from the at least one user to be screened includes:

determining an initial estimation result of behavior probability for representing the target behavior executed by the user to be screened based on the behavior characteristic information of the user to be screened and a second estimation model;

and screening out target users with corresponding behavior probabilities meeting preset screening conditions from the users to be screened based on the initial estimation result.

In a possible implementation manner, the behavior feature information of the user to be filtered includes at least one of the following information:

clicking a browsing behavior sequence, historical participation information aiming at target behaviors, user attribute information and interaction information.

In a possible implementation, the first personalized feature information of the target user includes: reasons why the target user did not perform the target behavior and/or personality characteristic information of the target user.

In a possible implementation manner, the second personalized feature information of the user to be matched includes: and the attribute information of the user to be matched and/or the behavior guide characteristic information of the user to be matched.

In a possible implementation manner, the determining an initial estimation result of a behavior probability for characterizing a target behavior executed by the user to be screened based on the behavior feature information of the user to be screened and the second estimation model includes:

determining a first characterization matrix corresponding to the behavior characteristic information of the user to be screened;

and inputting the first characterization matrix into the second pre-estimation model, and determining the initial pre-estimation result.

In a possible implementation manner, in a case that the behavior feature information includes a plurality of behavior feature information, the determining a first characterization matrix corresponding to the behavior feature information of the user to be filtered includes:

determining a characterization vector corresponding to each behavior feature information and a splicing sequence corresponding to the behavior feature information;

and splicing the characterization vectors corresponding to the behavior characteristic information according to the splicing sequence to obtain the first characterization matrix.

In a possible implementation manner, the screening, based on the initial estimation result, a target user whose corresponding behavior probability meets a preset screening condition from the users to be screened includes:

determining the user to be screened, of which the corresponding behavior probability is greater than the preset probability, as the target user; alternatively, the first and second electrodes may be,

and sequencing the users to be screened according to the corresponding behavior probability from large to small, and determining the users to be screened which are not ranked at the top N as the target users, wherein N is a preset positive integer.

In a possible implementation manner, the determining, based on the acquired first personalized feature information of the target user, the second personalized feature information of the user to be matched, the behavior feature information of the target user, and the first pre-estimation model, a matching user corresponding to each target user from the users to be matched includes:

determining a second characterization matrix corresponding to the first characterization feature information of the target user, the second characterization feature information of the user to be matched and the behavior feature information of the target user;

inputting the second characterization matrix into the first pre-estimation model, and determining a target pre-estimation result, wherein the target pre-estimation result is used for representing the probability of the target user executing the target behavior under the guidance of each user to be matched;

and determining a corresponding matched user for each target user based on the target estimation result.

In one possible embodiment, the method further comprises training the second predictive model according to the following method:

acquiring behavior characteristic information and first marking information of a sample user, wherein the first marking information is used for indicating whether the sample user executes the target behavior within a target time period;

determining the behavior probability of the sample user for executing the target behavior based on the behavior feature information of the sample user and a second pre-estimation model to be trained;

and training the second pre-estimation model based on the first marking information and the behavior probability.

In one possible embodiment, the method further comprises training the first predictive model according to the following method:

acquiring behavior characteristic information and second marking information of the sample user, wherein the second marking information is used for indicating whether the sample user executes the target behavior after the target time period;

screening target sample users meeting target behavior execution conditions from the sample users based on the trained second pre-estimation model and the behavior characteristic information of the sample users;

acquiring labeled first characterization feature information for the target sample user and labeled second characterization feature information for the user to be matched;

determining the probability of the target sample user executing the target behavior under the guidance of each user to be matched based on the first pre-estimation model to be trained, the first characterization feature information of the target sample user, the second characterization feature information of the user to be matched and the behavior feature information of the sample user;

and training the second pre-estimation model based on the probability that the target sample user executes the target behavior under the guidance of each user to be matched and the first marking information.

In a second aspect, an embodiment of the present disclosure further provides a behavior guiding device, including:

the first acquisition module is used for acquiring the behavior characteristic information of at least one user to be screened; the user to be screened is a user who does not execute the target behavior;

the screening module is used for screening target users meeting target behavior execution conditions from the at least one user to be screened based on the behavior characteristic information of the at least one user to be screened;

the second acquisition module is used for acquiring the labeled first personalized feature information aiming at the target user and the labeled second personalized feature information aiming at the user to be matched;

the determining module is used for determining a matching user corresponding to each target user from the users to be matched based on the acquired first characterization feature information of the target user, the second characterization feature information of the user to be matched, the behavior feature information of the target user and a first pre-estimation model, wherein the matching user is used for guiding the target user to execute the target behavior.

In a possible implementation manner, when the screening module is configured to screen, based on the behavior feature information of the at least one user to be screened, a target user meeting a target behavior execution condition from the at least one user to be screened, the screening module is configured to:

In a possible implementation manner, when determining an initial estimation result used for characterizing a behavior probability of the user to be screened for executing the target behavior based on the behavior feature information of the user to be screened and the second estimation model, the screening module is configured to:

In a possible implementation manner, when determining the first characterization matrix corresponding to the behavior feature information of the user to be screened, the screening module, when the behavior feature information includes a plurality of behavior feature information, is configured to:

In a possible implementation manner, when the target user whose corresponding behavior probability meets a preset screening condition is screened from the users to be screened based on the initial estimation result, the screening module is configured to:

In a possible implementation manner, when determining a matching user corresponding to each target user from the users to be matched based on the acquired first personalized feature information of the target user, the second personalized feature information of the user to be matched, the behavior feature information of the target user, and the first pre-estimation model, the screening module is configured to:

In a possible implementation, the behavior guiding apparatus further includes a training module, and the training module is configured to:

acquiring behavior characteristic information and first marking information of a sample user, wherein the first marking information is used for indicating whether the sample user executes the target behavior;

In a possible implementation, the training module is further configured to:

acquiring behavior characteristic information and second marking information of a sample user, wherein the second marking information is used for indicating whether the sample user executes the target behavior;

In a third aspect, this disclosure also provides a computer device, a processor, and a memory, where the memory stores machine-readable instructions executable by the processor, and the processor is configured to execute the machine-readable instructions stored in the memory, and when the machine-readable instructions are executed by the processor, the machine-readable instructions are executed by the processor to perform the steps in the first aspect or any one of the possible implementations of the first aspect.

In a fourth aspect, this disclosure also provides a computer-readable storage medium having a computer program stored thereon, where the computer program is executed to perform the steps in the first aspect or any one of the possible implementation manners of the first aspect.

According to the behavior guidance method, the behavior guidance device, the computer equipment and the storage medium, firstly, target users meeting target behavior execution conditions are screened from the users to be screened based on the acquired behavior characteristic information of the users to be screened; and determining a corresponding matched user for each target user based on the acquired first characterization feature information of the target user, the acquired second characterization feature information of the user to be matched, the acquired behavior feature information of the target user and the first estimation model. By the method, one-step screening can be performed on the basis of the behavior characteristic information of the user to be screened, and then the matched user is determined for the screened target user, so that the guidance efficiency of the target behavior can be improved, and the guidance effect of the target behavior can be improved.

In order to make the aforementioned objects, features and advantages of the present disclosure more comprehensible, preferred embodiments accompanied with figures are described in detail below.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the drawings required for use in the embodiments will be briefly described below, and the drawings herein incorporated in and forming a part of the specification illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the technical solutions of the present disclosure. It is appreciated that the following drawings depict only certain embodiments of the disclosure and are therefore not to be considered limiting of its scope, for those skilled in the art will be able to derive additional related drawings therefrom without the benefit of the inventive faculty.

Fig. 1 illustrates a flow chart of a behavior guidance method provided by an embodiment of the present disclosure;

FIG. 2 is a flow chart illustrating a training process of a second predictive model in a behavior guidance method according to an embodiment of the disclosure;

FIG. 3 is a flow chart illustrating a training process of a first predictive model in a behavior guidance method according to an embodiment of the disclosure;

fig. 4 illustrates a schematic diagram of a behavior guidance device 400 provided by an embodiment of the present disclosure;

fig. 5 shows a schematic diagram of a computer device 500 provided by an embodiment of the present disclosure.

Detailed Description

In order to make the objects, technical solutions and advantages of the embodiments of the present disclosure more clear, the technical solutions of the embodiments of the present disclosure will be described clearly and completely with reference to the drawings in the embodiments of the present disclosure, and it is obvious that the described embodiments are only a part of the embodiments of the present disclosure, not all of the embodiments. The components of embodiments of the present disclosure, as generally described and illustrated herein, may be arranged and designed in a wide variety of different configurations. Thus, the following detailed description of the embodiments of the present disclosure is not intended to limit the scope of the disclosure, as claimed, but is merely representative of selected embodiments of the disclosure. All other embodiments, which can be derived by a person skilled in the art from the embodiments of the disclosure without making creative efforts, shall fall within the protection scope of the disclosure.

It has been found that increasing the user's viscosity between the online education platform in order to encourage the user to perform certain predetermined actions, such as encouraging user consumption, is generally guided by the staff of the online education platform. However, in the related art, all users who interact with the online education platform are generally guided by staff, which is inefficient and has poor guiding effect.

Based on the above research, the behavior guidance method, the behavior guidance device, the computer device, and the storage medium provided by the embodiments of the present disclosure screen the target users that meet the target behavior execution condition from the users to be screened based on the acquired behavior feature information of the users to be screened; and determining a corresponding matched user for each target user based on the acquired first characterization feature information of the target user, the acquired second characterization feature information of the user to be matched, the acquired behavior feature information of the target user and the first estimation model. By the method, one-step screening can be performed on the basis of the behavior characteristic information of the user to be screened, and then the matched user is determined for the screened target user, so that the guidance efficiency of the target behavior can be improved, and the guidance effect of the target behavior can be improved.

The above-mentioned drawbacks are the results of the inventor after practical and careful study, and therefore, the discovery process of the above-mentioned problems and the solutions proposed by the present disclosure to the above-mentioned problems should be the contribution of the inventor in the process of the present disclosure.

It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, it need not be further defined and explained in subsequent figures.

To facilitate understanding of the present embodiment, first, a behavior guidance method disclosed in the embodiments of the present disclosure is described in detail, where an execution subject of the behavior guidance method provided in the embodiments of the present disclosure is generally a computer device with certain computing capability, and the computer device includes, for example: a terminal device, which may be a User Equipment (UE), a mobile device, a User terminal, a cellular phone, a cordless phone, a Personal Digital Assistant (PDA), a handheld device, a computing device, a vehicle mounted device, a wearable device, or a server or other processing device. In some possible implementations, the behavior guidance method may be implemented by a processor calling computer readable instructions stored in a memory.

The following describes a behavior guidance method provided by the embodiment of the present disclosure by taking an execution subject as a terminal device.

Referring to fig. 1, a flowchart of a behavior guidance method provided in an embodiment of the present disclosure is shown, where the method includes steps S101 to S104, where:

s101: acquiring behavior characteristic information of at least one user to be screened; and the user to be screened is a user who does not execute the target behavior.

S102: and screening target users meeting target behavior execution conditions from the at least one user to be screened based on the behavior characteristic information of the at least one user to be screened.

S103: and acquiring labeled first characterization feature information aiming at the target user and labeled second characterization feature information aiming at the user to be matched.

S104: and determining a matching user corresponding to each target user from the users to be matched based on the acquired first characterization feature information of the target user, the second characterization feature information of the user to be matched, the behavior feature information of the target user and a first pre-estimation model, wherein the matching user is used for guiding the target user to execute the target behavior.

The following is a detailed description of the above steps:

for S101,

In one possible application scenario, the user to be screened may be a student or a student who participates in an online classroom, which may be a recorded video classroom or an online live video classroom.

The click browsing behavior sequence can be a sequence formed by clicking and browsing behaviors of the user to be screened in a preset website/application by using the terminal equipment according to a certain arrangement sequence;

the historical participation information for the target behavior represents the participation situation of the user to be screened for the target behavior in a past preset time period, for example, if the target behavior is a web class purchase, the historical participation information for the target behavior may be audition duration, audition times, audition feedback and the like of the user for the audition class;

the user attribute information may include basic attributes of the user to be screened, such as age, gender, region and the like, and if the target behavior is to purchase a web course, the user attribute information may further include a learning stage of the user;

the interaction information represents the interaction of the user to be screened with other personnel based on the target behavior, and may include information such as the number of interactions with other personnel and the content of the interactions.

When the obtained behavior feature information of the user to be screened includes the user attribute information, in a possible implementation manner, the behavior feature information corresponding to the user to be screened may be obtained from a stored user feature information library.

Here, the user may create an account at the client of the terminal device. When the account is created, the user side can ask questions corresponding to the characteristic information, reserves a space for the user to fill in, and sends the user characteristic information to the user characteristic information base for storage after the user fills in. For example, the question list in the presentation page when the question corresponding to the user feature information is filled in may be as shown in table 1:

TABLE 1

When the behavior feature information includes a click browsing behavior sequence, and when the behavior feature information of the user to be screened is obtained, the pre-stored historical operation data of the user to be screened may be obtained, and the corresponding click browsing behavior sequence is determined based on the historical operation data, where the historical operation data includes historical browsing data and historical click data.

Taking the acquisition of the click browsing behavior sequence as an example, various click behaviors of the user may be sorted in advance, then the historical operation data of the user is sorted according to the sorting result of the click behaviors to obtain the click browsing behavior sequence, for example, whether the user clicks four operations of "advertisement 1", "link 2", "picture 3", and "video 4" may be sequentially sorted, and then a value corresponding to each item is determined based on the historical operation data of the user to obtain the click browsing behavior sequence corresponding to the historical operation data.

For example, the process of performing two classifications (i.e., "1" for yes and "0" for no) on the click-to-browse behavior sequence is taken as an example. When the user clicks "advertisement 1" and "link 2" but does not click "picture 3" and "video 4", the generated corresponding click-to-browse behavior sequence is:

[1 1 0 0]

for S102,

In a possible implementation manner, when a target user meeting a target behavior execution condition is screened from the at least one user to be screened, some screening conditions may be preset for behavior feature information, and then a user whose corresponding behavior feature information meets preset conditions among the users to be screened is determined as the target user.

In another possible implementation manner, when a target user meeting a target behavior execution condition is screened from the at least one user to be screened based on the behavior feature information of the at least one user to be screened, an initial prediction result for characterizing a behavior probability of the user to be screened for executing the target behavior may be determined based on the behavior feature information of the user to be screened and a second prediction model; and then screening out target users with corresponding behavior probabilities meeting preset screening conditions from the users to be screened based on the initial estimation result. Illustratively, the second predictive model may be a two-class Random Forest (RF) model with filtering capability.

In a possible implementation manner, when an initial estimation result used for characterizing the behavior probability of the user to be screened executing the target behavior is determined based on the behavior feature information of the user to be screened and the second estimation model, a first characterization matrix corresponding to the behavior feature information of the user to be screened may be determined first; and then, inputting the first characterization matrix into the second pre-estimation model, and determining the initial pre-estimation result.

Under the condition that the behavior feature information includes a plurality of behavior feature information, when determining the first characterization matrix corresponding to the behavior feature information of the user to be screened, for example, a corresponding condition may be set for each feature information, and then based on whether the condition is met, two kinds of classification (that is, "1" represents "yes" and "0" represents "no") processing are performed, so as to determine a characterization vector corresponding to each behavior feature information, and determine a splicing order corresponding to the plurality of behavior feature information; and then, splicing the characterization vectors corresponding to the behavior characteristic information according to the splicing sequence to obtain the first characterization matrix.

When the behavior feature information includes a plurality of behavior feature information, when determining the first characterization matrix corresponding to the behavior feature information of the user to be screened, for example, the behavior feature information of the user to be screened may be arranged according to a preset sequence to obtain a feature sequence of each user to be screened; then, converting the characteristic sequences based on the characteristic codes to obtain corresponding characteristic vectors; and finally, combining the characteristic sequences of the users to be screened to obtain a first characterization matrix for characterizing the behavior characteristic information. Wherein, the characteristic coding mode can be mult-hot coding.

Illustratively, user 1 clicks "Ad 1", "Link 2", "Picture 3", "video 4"; user 2 did not click on "Ad 1" and "Link 2", but clicked on "Picture 3" and "video 4"; user 3 clicked "Ad 1" and "Link 2", but not "Picture 3" and "video 4"; the user 4 does not click on "advertisement 1", "link 2", "picture 3", or "video 4". The obtained first characterization matrix is:

wherein, the value of the first column indicates whether to click on "advertisement 1", the value of the second column indicates whether to click on "Link 2", the value of the third column indicates whether to click on "Picture 3", and the value of the fourth column indicates whether to click on "video 4".

And then, inputting the first representation matrix into the second pre-estimation model to obtain an initial pre-estimation result of behavior probability for representing the user to be screened to execute the target behavior.

For example, still taking the users 1 to 4 as examples, after the first characterization matrix corresponding to each user is input to the second prediction model, the initial prediction result output by the second prediction model may be obtained as follows:

[0.8 0.75 0.75 0.5]

wherein, "0.8", "0.75" and "0.5" respectively represent the estimation results of the second estimation model for users 1 to 4, that is, user 1 has a probability of 80% to execute the target behavior; user 2 has a 75% probability of performing the target action; user 3 has a 75% probability of performing the target action; user 4 has a 50% probability of performing the target action.

Specifically, in the same time period, there may be many users who execute the target behavior, and the time, energy, and number of guided persons of the users to be matched are also limited, and in order to improve the efficiency of the target behavior guidance, the initial estimation result obtained by the second estimation model may be used as a basis for screening, so as to determine that the behavior probability corresponding to the user to be screened meets a preset screening condition, that is, the target user with higher guidance value.

When a target user with a corresponding behavior probability meeting a preset screening condition is screened from the users to be screened based on the initial estimation result, any one of the following methods can be adopted:

the method A comprises the following steps: and sequencing the users to be screened from large to small according to the corresponding behavior probability, and determining the top N users to be screened as the target users, wherein N is a preset positive integer.

Specifically, the behavior probabilities of the users to be screened may be sorted from high to low. When the nth behavior probability is the same as the surrounding behavior probabilities, for example, when N is 2 and the ranked sequence numbers are "1", "2", and "4", any one of the pre-estimated value with sequence number "1" and the behavior probabilities with two sequence numbers "2" may be selected; or, the behavior probabilities with sequence number "1" and two sequence numbers "2" are selected simultaneously.

For example, still taking the above users 1 to 4 as an example, the four behavior probabilities of "0.8", "0.75" and "0.5" may be ranked from high to low, and 3 users to be filtered corresponding to the first 2 estimated values of "0.8" and "0.75" may be determined as the target users, or 2 users to be filtered corresponding to the estimated values of "0.8" and the 1 st or 2 nd "0.75" may be determined as the target users.

The method B comprises the following steps: and determining the user to be screened, of which the corresponding behavior probability is greater than the preset probability, as the target user.

Specifically, a preset probability M can be set within a range of 0 to 1 based on the requirement of realistic screening, such as that qualification corresponds to 60 points (60%) and excellence corresponds to 90 points (90%).

For example, still taking the users 1 to 4 as examples, when the preset probability M is 0.6, it may be determined that 3 users to be screened, corresponding to behavior probabilities of "0.8" and "0.75", are the target users.

For S103,

In a possible implementation manner, in order to make the feature information of the target user and the user to be matched richer and more realistic, the first personalized feature of the target user may be labeled manually, and the labeling manner may be classification, that is, "1" represents "yes" and "0" represents "no".

The first personalized feature information of the target user may include, for example: reasons why the target user did not perform the target behavior and/or personality characteristic information of the target user.

Here, the reason why the target behavior is not executed may be various possible reasons counted in advance and numbered, then, for each target user, the number of the target behavior that is not executed corresponding to the target user may be marked, and after the corresponding number is identified, the terminal device may perform corresponding encoding. For example, 10 reasons for not executing the target behavior are preset, which are reasons 1 to 10, respectively, and the reason for not executing the target behavior by the target user 1 is reason 5, the generated corresponding codes are: [0000100000]

Wherein the selectable reason why the target user did not perform the target behavior may be an aggregation of reasons why multiple users did not perform the target behavior.

Here, the user to be matched may be a user having a behavior guidance capability such as a shopping guide, a tape delivery anchor, a teacher with class, or the like.

Taking the user to be matched as a teacher with a lesson as an example, the attribute information of the user to be matched can be basic information such as historical conversion rate, job correction rate, teaching duration and the like of the teacher needing to be matched; the behavior guidance characteristic information of the user to be matched can be characteristic information of personal business expertise of a teacher to be matched, logic and methodology for guiding students to execute target behaviors and the like.

For S104,

When a matching user corresponding to each target user is determined from the users to be matched based on the acquired first characterization feature information of the target user, the second characterization feature information of the user to be matched, the behavior feature information of the target user and the first pre-estimation model, the first characterization feature information of the target user, the second characterization feature information of the user to be matched and the second characterization matrix corresponding to the behavior feature information of the target user can be determined firstly; then, inputting the second characterization matrix into the first pre-estimation model, and determining a target pre-estimation result, wherein the target pre-estimation result is used for representing the probability of the target user executing the target behavior under the guidance of each user to be matched; and finally, determining a corresponding matched user for each target user based on the target estimation result.

Specifically, the first characterization feature information of the target user, the second characterization feature information of the user to be matched, and the behavior feature information of the target user may be converted based on feature codes, and then arranged according to a preset sequence to obtain a second characterization matrix. Wherein the signature codes may be mult-hot codes.

Illustratively, target user 1 is marked with a coring click on "Ad 1", "Link 2", "Picture 3", "video 4"; target user 2 is marked as not having a coring, has not clicked on "ad 1" and "link 2", but has clicked on "picture 3" and "video 4"; target user 3 is marked as not having a coring, having clicked on "ad 1" and "link 2", but not having clicked on "picture 3" and "video 4"; the historical conversion rate of the user 1 to be matched is high; the historical conversion rate of the user 2 to be matched is low. The 3 target users and 2 users to be matched are arranged and combined to obtain 6 matching modes, and the obtained second characterization matrix is, for example:

wherein the first column indicates whether the corresponding target user is marked with a coring; the second column indicates whether the corresponding target user clicked on "ad 1"; the third column indicates whether the corresponding target user clicked "Link 2"; the fourth column indicates whether the corresponding target user clicked "Picture 3"; the fifth column indicates whether the corresponding target user clicked on "video 4"; the sixth column indicates the historical conversion rate of the corresponding user to be matched. The first and second rows represent target user 1; the third and fourth rows represent target user 2; the fifth and sixth rows represent the target users 3. The sixth elements of the first, third and fifth rows represent users 1 to be matched, and the sixth elements of the second, fourth and sixth rows represent users 2 to be matched.

And then, inputting the second characterization matrix into the first pre-estimation model to generate the target pre-estimation result. Wherein the first predictive model may be a two-class eXtreme Gradient Boosting (xgb) model with screening and matching capabilities.

For example, still taking the target users 1 to 3 and the users 1 and 2 to be matched as examples, after the second characterization matrix is input to the first pre-estimation model, the target pre-estimation result output by the first pre-estimation model can be obtained as follows:

wherein, the "0.9" and "0.6" in the 1 st row represent that the probability of the target user 1 performing the target behavior under the guidance of the user 1 to be matched is 90%, and the probability of the target behavior under the guidance of the user 2 to be matched is 60%; the 2 nd row and the 3 rd row respectively represent the probability that the target user 2 and the target user 3 execute the target behavior under the guidance of the user 1 to be matched and the user 2 to be matched respectively. Further, the probability that the target user performs the target behavior under the guidance of the user to be matched may characterize the matching degree with the user to be matched.

Further, the target predictors may be converted into a predictor table as shown in table 2:

TABLE 2

	User 1 to be matched	User 2 to be matched
			Target user 1	90％	60％
Target user 2	50％	70％
			Target user 3	60％	65％

After the estimation result table is obtained, the user 1 to be matched with the target user 1 with the highest matching degree can be obtained; the user 2 to be matched is the user with the highest matching degree with the target user 2; the user 2 to be matched is the user with the highest matching degree with the target user 3.

After determining the user to be matched with the target user with the highest matching degree, the information of the user to be matched with the highest matching degree, such as guide experience, speciality, contact way and the like, can be sent to the corresponding target user; meanwhile, the information such as the speciality, the contact information and the like of the target user can be sent to the corresponding user to be matched. By the method, the relation between the target user and the user to be matched can be established, and subsequent behavior guidance is facilitated.

The embodiment of the present disclosure further provides a training method of a second predictive model, as shown in fig. 2, which is a flowchart of the training method of the second predictive model provided in the embodiment of the present disclosure, and the method includes the following steps:

s201, behavior feature information and first marking information of a sample user are obtained, wherein the first marking information is used for indicating whether the sample user executes the target behavior.

S202, determining the behavior probability of the sample user for executing the target behavior based on the behavior feature information of the sample user and the second pre-estimation model to be trained.

S203, training the second pre-estimation model based on the first marking information and the behavior probability.

Here, when acquiring the behavior feature information and the first label information of the sample user for training the second estimation model, the period for executing the target behavior in a large scale may be determined as a window period based on the historical data, for example, the window period may be 3 to 7 days after the end of the free trial listening/low-price trial listening class, and the behavior feature information and the first label information of the sample user for training the second estimation model, which are acquired, may be data corresponding to the window period.

In a possible implementation manner, after the behavior feature information and the first labeling information of the sample user are obtained, the information can be cleaned and screened, and important parts of the information can be extracted, so that the training efficiency of the second estimation model is improved. For example, browsing click behaviors with important business meanings can be screened from a plurality of events based on understanding of business; meaningless information such as meaningless icons, pictures, user ids, repeated information and abnormal network logs in the interactive information of the sample users can be eliminated.

When the second pre-estimation model is trained, the second pre-estimation model can output the probability of the target user executing the target behavior, the pre-estimation result of the second pre-estimation model can be determined based on the probability of the target user executing the target behavior and a preset probability threshold, the pre-estimation result can represent whether the target user executes the target behavior, the loss value in the training process can be determined based on the pre-estimation result of the second pre-estimation model and the first marking data, and then the model parameters of the second pre-estimation model are adjusted based on the determined loss value.

In practical application, after the training of the second pre-estimation model is completed, multiple optimizations may be performed during the application process. Specifically, the result of whether the latest user executes the target behavior and the estimated result can be matched and stored to perfect sample data; and then based on the sample data after the completion, training the second pre-estimation model again to realize continuous optimization of the second pre-estimation model.

The embodiment of the present disclosure further provides a training method of a first predictive model, and as shown in fig. 3, the flowchart of the training method of the first predictive model provided in the embodiment of the present disclosure includes the following steps:

s301, behavior feature information and second marking information of the sample user are obtained, wherein the second marking information is used for indicating whether the sample user executes the target behavior.

S302, screening target sample users meeting target behavior execution conditions from the sample users based on the trained second estimation model and the behavior characteristic information of the sample users.

S303, obtaining labeled first characterization feature information aiming at the target sample user and labeled second characterization feature information aiming at the user to be matched.

S304, determining the probability of the target sample user executing the target behavior under the guidance of each user to be matched based on the first pre-estimation model to be trained, the first characterization feature information of the target sample user, the second characterization feature information of the user to be matched and the behavior feature information of the sample user.

S305, training the second pre-estimation model based on the probability that the target sample user executes the target behavior under the guidance of each user to be matched and the first marking information.

Here, the behavior feature information and the second label information of the sample user for training the first pre-estimation model may be acquired data after the window period.

When the first pre-estimation model is trained, the first pre-estimation model can output the probability that a target user executes a target behavior under the guidance of each user to be matched, the pre-estimation result of the second pre-estimation model can be determined based on the probability that the target user executes the target behavior under the guidance of each user to be matched and a preset probability threshold, the pre-estimation result can represent whether the target user executes the target behavior under the guidance of the user to be matched, the loss value in the training process can be determined based on the pre-estimation result of the second pre-estimation model and the first marking data, and then the model parameters of the first pre-estimation model are adjusted based on the determined pre-estimation loss value.

In practical application, after the training of the second pre-estimation model is completed, multiple optimizations may be performed during the application process. Specifically, the result of whether the user executes the target behavior and the result of model estimation can be matched and stored to perfect sample data; and then based on the sample data after the completion, training the first pre-estimation model again to realize continuous optimization of the first pre-estimation model.

According to the behavior guidance method provided by the embodiment of the disclosure, firstly, target users meeting target behavior execution conditions are screened from users to be screened based on the acquired behavior characteristic information of the users to be screened; and determining a corresponding matched user for each target user based on the acquired first characterization feature information of the target user, the acquired second characterization feature information of the user to be matched, the acquired behavior feature information of the target user and the first estimation model. By the method, one-step screening can be performed on the basis of the behavior characteristic information of the user to be screened, and then the matched user is determined for the screened target user, so that the guidance efficiency of the target behavior can be improved, and the guidance effect of the target behavior can be improved.

It will be understood by those skilled in the art that in the method of the present invention, the order of writing the steps does not imply a strict order of execution and any limitations on the implementation, and the specific order of execution of the steps should be determined by their function and possible inherent logic.

Based on the same inventive concept, the embodiment of the present disclosure further provides a behavior guidance device corresponding to the behavior guidance method, and as the principle of the device in the embodiment of the present disclosure for solving the problem is similar to the behavior guidance method described above in the embodiment of the present disclosure, the implementation of the device may refer to the implementation of the method, and repeated details are not described again.

Referring to fig. 4, there is shown a schematic diagram of a behavior guidance device 400 according to an embodiment of the present disclosure, the device including: a first obtaining module 401, a screening module 402, a second obtaining module 403, and a determining module 404; wherein the content of the first and second substances,

a first obtaining module 401, configured to obtain behavior feature information of at least one user to be screened; the user to be screened is a user who does not execute the target behavior;

a screening module 402, configured to screen, based on the behavior feature information of the at least one user to be screened, a target user that meets a target behavior execution condition from the at least one user to be screened;

a second obtaining module 403, configured to obtain labeled first personalized feature information for the target user and labeled second personalized feature information for a user to be matched;

a determining module 404, configured to determine, based on the acquired first personalized feature information of the target user, the second personalized feature information of the user to be matched, the behavior feature information of the target user, and the first pre-estimation model, a matching user corresponding to each target user from the users to be matched, where the matching user is used to guide the target user to execute the target behavior.

In a possible implementation manner, the screening module 402, when screening, based on the behavior feature information of the at least one user to be screened, a target user meeting a target behavior execution condition from the at least one user to be screened, is configured to:

In a possible implementation manner, the screening module 402, when determining an initial prediction result for characterizing a behavior probability of the user to be screened performing a target behavior based on the behavior feature information of the user to be screened and a second prediction model, is configured to:

In a possible implementation manner, when the behavior feature information includes a plurality of behavior feature information, the screening module 402, when determining the first characterization matrix corresponding to the behavior feature information of the user to be screened, is configured to:

In a possible implementation manner, the screening module 402, when the target user whose corresponding behavior probability meets a preset screening condition is screened from the users to be screened based on the initial estimation result, is configured to:

In a possible implementation manner, when the matching user corresponding to each target user is determined from the users to be matched based on the obtained first personalized feature information of the target user, the second personalized feature information of the user to be matched, the behavior feature information of the target user, and the first pre-estimation model, the screening module 402 is configured to:

In a possible implementation, the behavior guidance device further includes a training module 405, which is configured to:

In a possible implementation, the training module 405 is further configured to:

The behavior guiding device provided by the embodiment of the disclosure screens target users meeting target behavior execution conditions from users to be screened based on acquired behavior characteristic information of the users to be screened; and determining a corresponding matched user for each target user based on the acquired first characterization feature information of the target user, the acquired second characterization feature information of the user to be matched, the acquired behavior feature information of the target user and the first estimation model. By the method, one-step screening can be performed on the basis of the behavior characteristic information of the user to be screened, and then the matched user is determined for the screened target user, so that the guidance efficiency of the target behavior can be improved, and the guidance effect of the target behavior can be improved.

The description of the processing flow of each module in the device and the interaction flow between the modules may refer to the related description in the above method embodiments, and will not be described in detail here.

An embodiment of the present disclosure further provides a computer device, as shown in fig. 5, which is a schematic diagram of a computer device 500 provided in an embodiment of the present disclosure, including:

based on the same technical concept, the embodiment of the present disclosure also provides a computer device 500. Referring to fig. 5, a schematic structural diagram of a computer device 500 provided in the embodiment of the present disclosure includes a processor 501, a memory 502, and a bus 503. The memory 502 is used for storing execution instructions and includes a memory 5021 and an external memory 5022; the memory 5021 is also referred to as an internal memory, and is used for temporarily storing operation data in the processor 501 and data exchanged with an external storage 5022 such as a hard disk, the processor 501 exchanges data with the external storage 5022 through the memory 5021, and when the computer device 500 operates, the processor 501 communicates with the storage 502 through the bus 503, so that the processor 501 executes the following instructions:

In a possible implementation manner, the instructions executed by the processor 501, for screening, from the at least one user to be screened, a target user meeting a target behavior execution condition based on the behavior feature information of the at least one user to be screened, include:

In a possible implementation manner, in the instructions executed by the processor 501, the behavior feature information of the user to be filtered includes at least one of the following information:

In a possible implementation manner, the processor 501 executes the instructions, where the first personalized feature information of the target user includes: reasons why the target user did not perform the target behavior and/or personality characteristic information of the target user.

In a possible implementation manner, the instructions executed by the processor 501 include the second personalized feature information of the user to be matched: and the attribute information of the user to be matched and/or the behavior guide characteristic information of the user to be matched.

In a possible implementation manner, in the instructions executed by the processor 501, the determining an initial prediction result used for characterizing a probability of a behavior of the user to be screened performing a target behavior based on the behavior feature information of the user to be screened and the second prediction model includes:

In a possible implementation manner, in the instructions executed by the processor 501, in a case that the behavior feature information includes a plurality of behavior feature information, the determining a first characterization matrix corresponding to the behavior feature information of the user to be filtered includes:

In a possible implementation manner, in the instructions executed by the processor 501, the screening, based on the initial estimation result, a target user whose corresponding behavior probability meets a preset screening condition from the users to be screened includes:

In a possible implementation manner, in an instruction executed by the processor 501, the determining, based on the acquired first personalized feature information of the target user, the second personalized feature information of the user to be matched, the behavior feature information of the target user, and the first pre-estimation model, a matching user corresponding to each target user from the users to be matched includes:

In a possible implementation, the instructions executed by the processor 501 further include training the second prediction model according to the following method:

In a possible implementation, the instructions executed by the processor 501 further include training the first predictive model according to the following method:

The embodiments of the present disclosure also provide a computer-readable storage medium, on which a computer program is stored, where the computer program is executed by a processor to perform the steps of the behavior guidance method in the above method embodiments. The storage medium may be a volatile or non-volatile computer-readable storage medium.

The computer program product of the behavior guidance method provided in the embodiments of the present disclosure includes a computer-readable storage medium storing a program code, where instructions included in the program code may be used to execute the steps of the behavior guidance method in the above method embodiments, which may be referred to specifically in the above method embodiments, and are not described herein again.

The embodiments of the present disclosure also provide a computer program, which when executed by a processor implements any one of the methods of the foregoing embodiments. The computer program product may be embodied in hardware, software or a combination thereof. In an alternative embodiment, the computer program product is embodied in a computer storage medium, and in another alternative embodiment, the computer program product is embodied in a Software product, such as a Software Development Kit (SDK), or the like.

It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the system and the apparatus described above may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again. In the several embodiments provided in the present disclosure, it should be understood that the disclosed system, apparatus, and method may be implemented in other ways. The above-described embodiments of the apparatus are merely illustrative, and for example, the division of the units is only one logical division, and there may be other divisions when actually implemented, and for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection of devices or units through some communication interfaces, and may be in an electrical, mechanical or other form.

The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, functional units in the embodiments of the present disclosure may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit.

The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a non-volatile computer-readable storage medium executable by a processor. Based on such understanding, the technical solution of the present disclosure may be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present disclosure. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.

Finally, it should be noted that: the above-mentioned embodiments are merely specific embodiments of the present disclosure, which are used for illustrating the technical solutions of the present disclosure and not for limiting the same, and the scope of the present disclosure is not limited thereto, and although the present disclosure is described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: any person skilled in the art can modify or easily conceive of the technical solutions described in the foregoing embodiments or equivalent technical features thereof within the technical scope of the present disclosure; such modifications, changes or substitutions do not depart from the spirit and scope of the embodiments of the present disclosure, and should be construed as being included therein. Therefore, the protection scope of the present disclosure shall be subject to the protection scope of the claims.

Claims

1. A behavior guidance method, comprising:

2. The method according to claim 1, wherein the screening, based on the behavior feature information of the at least one user to be screened, a target user meeting target behavior execution conditions from the at least one user to be screened includes:

3. The method according to claim 1, wherein the behavior feature information of the user to be filtered comprises at least one of the following information:

4. The method of claim 1, wherein the first personalized feature information of the target user comprises: reasons why the target user did not perform the target behavior and/or personality characteristic information of the target user.

5. The method according to claim 1, wherein the second personalized feature information of the user to be matched comprises: and the attribute information of the user to be matched and/or the behavior guide characteristic information of the user to be matched.

6. The method according to claim 2, wherein the determining an initial prediction result for characterizing a behavior probability of the user to be screened for performing a target behavior based on the behavior feature information of the user to be screened and a second prediction model comprises:

7. The method according to claim 6, wherein in a case that the behavior feature information includes a plurality of behavior feature information, the determining a first characterization matrix corresponding to the behavior feature information of the user to be filtered includes:

8. The method according to claim 2, wherein the screening, based on the initial estimation result, a target user whose corresponding behavior probability meets a preset screening condition from the users to be screened comprises:

9. The method according to claim 1, wherein the determining a matching user corresponding to each target user from the users to be matched based on the acquired first personalized feature information of the target user, the acquired second personalized feature information of the user to be matched, the acquired behavior feature information of the target user, and the first pre-estimation model includes:

10. The method of claim 1, further comprising training the second predictive model according to the following method:

11. The method of claim 10, further comprising training the first predictive model according to the following method:

12. A behavior guidance device, comprising:

13. A computer device, comprising: a processor, a memory storing machine-readable instructions executable by the processor, the processor to execute machine-readable instructions stored in the memory, the processor to perform the steps of the behavior guidance method of any of claims 1 to 11 when the machine-readable instructions are executed by the processor.

14. A computer-readable storage medium, characterized in that a computer program is stored on the computer-readable storage medium, which computer program, when executed by a computer device, performs the steps of the behavior guidance method according to any one of claims 1 to 11.