A behavior guidance method, device, computer equipment and storage medium

Abstract

The present disclosure provides a behavior guidance method, apparatus, computer equipment and storage medium, the method comprising: obtaining behavior feature information of at least one user to be screened; wherein the user to be screened is a user who has not performed a target behavior; based on the behavior feature information of the at least one user to be screened, screening out target users that meet the target behavior execution conditions from the at least one user to be screened; obtaining the marked first personalized feature information for the target user and the marked second personalized feature information for the user 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 a first estimation model, determining a matching user corresponding to each target user from the users to be matched, the matching user being used to guide the target user to perform the target behavior.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, and in particular to a behavior guidance method, device, computer equipment and storage medium. Background Art

[0002] With the rapid development of Internet technology, online education has been rapidly popularized. In order to encourage users to perform certain preset behaviors and increase user stickiness with online education platforms, such as encouraging users to consume, users are generally guided by the staff of online education platforms.

[0003] However, this method is relatively inefficient and has poor guidance effect. Summary of the invention

[0004] The embodiments of the present disclosure at least provide a behavior guidance method, apparatus, computer equipment, and storage medium.

[0005] In a first aspect, an embodiment of the present disclosure provides a behavior guidance method, including:

[0006] Obtaining behavioral characteristic information of at least one user to be screened; wherein the user to be screened is a user who has not performed the target behavior;

[0007] Based on the behavior characteristic information of the at least one to-be-screened user, a target user meeting the target behavior execution condition is screened out from the at least one to-be-screened user;

[0008] Acquire the first personalized feature information marked for the target user and the second personalized feature information marked for the user to be matched;

[0009] Based on the acquired first personalized feature information of the target user, the second personalized feature information of the user to be matched, the behavioral feature information of the target user and the first estimation model, a matching user corresponding to each target user is determined from the users to be matched, and the matching user is used to guide the target user to perform the target behavior.

[0010] In a possible implementation, the step of selecting a target user who meets the target behavior execution condition from the at least one user to be screened based on the behavior characteristic information of the at least one user to be screened includes:

[0011] Based on the behavior feature information of the to-be-screened user and the second estimation model, determining an initial estimation result for characterizing the behavior probability of the to-be-screened user performing the target behavior;

[0012] Based on the initial estimation result, target users whose corresponding behavior probabilities meet preset screening conditions are screened out from the users to be screened.

[0013] In a possible implementation manner, the behavior characteristic information of the user to be screened includes at least one of the following information:

[0014] Click-through behavior sequence, historical participation information for target behavior, user attribute information, and interaction information.

[0015] In a possible implementation manner, the first personalized characteristic information of the target user includes: the reason why the target user does not perform the target behavior and / or personality characteristic information of the target user.

[0016] In a possible implementation manner, the second personalized characteristic information of the user to be matched includes: attribute information of the user to be matched and / or behavior guidance characteristic information of the user to be matched.

[0017] In a possible implementation manner, determining an initial estimation result for characterizing the probability of the to-be-screened user performing the target behavior based on the to-be-screened user's behavior feature information and the second estimation model includes:

[0018] Determine a first representation matrix corresponding to the behavior characteristic information of the user to be screened;

[0019] The first characterization matrix is ​​input into the second estimation model to determine the initial estimation result.

[0020] In a possible implementation manner, when the behavior characteristic information includes a plurality of behavior characteristic information, determining the first representation matrix corresponding to the behavior characteristic information of the to-be-screened user includes:

[0021] Determine a representation vector corresponding to each behavior feature information, and a concatenation order corresponding to the plurality of behavior feature information;

[0022] The representation vectors corresponding to the plurality of behavior feature information are concatenated in the concatenation order to obtain the first representation matrix.

[0023] In a possible implementation, based on the initial estimation result, screening out target users whose corresponding behavior probabilities meet preset screening conditions from the users to be screened includes:

[0024] Determine the to-be-screened user whose corresponding behavior probability is greater than a preset probability as the target user; or,

[0025] The users to be screened are sorted from large to small according to the corresponding behavior probabilities, and the top N users to be screened are determined as the target users, where N is a preset positive integer.

[0026] In a possible implementation, 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 behavioral feature information of the target user, and the first estimation model includes:

[0027] Determine a second representation matrix corresponding to the first personalized feature information of the target user, the second personalized feature information of the to-be-matched user, and the behavior feature information of the target user;

[0028] Inputting the second characterization matrix into the first estimation model to determine a target estimation result, wherein the target estimation result is used to represent the probability that the target user performs the target behavior under the guidance of each to-be-matched user;

[0029] Based on the target estimation result, a corresponding matching user is determined for each target user.

[0030] In a possible implementation manner, the method further includes training the second estimation model according to the following method:

[0031] Acquire behavior feature information and first annotation information of the sample user, wherein the first annotation information is used to indicate whether the sample user performs the target behavior within a target time period;

[0032] Determining the behavior probability of the sample user performing the target behavior based on the behavior feature information of the sample user and the second estimation model to be trained;

[0033] The second estimation model is trained based on the first annotation information and the behavior probability.

[0034] In a possible implementation manner, the method further includes training the first estimation model according to the following method:

[0035] Acquire behavior feature information and second annotation information of the sample user, wherein the second annotation information is used to indicate whether the sample user performs the target behavior after the target time period;

[0036] Based on the trained second estimation model and the behavior characteristic information of the sample users, target sample users meeting the target behavior execution condition are screened out from the sample users;

[0037] Acquire the first personalized feature information marked for the target sample user and the second personalized feature information marked for the user to be matched;

[0038] Determine the probability that the target sample user performs the target behavior under the guidance of each user to be matched based on the first estimation model to be trained, the first personalized feature information of the target sample user, the second personalized feature information of the user to be matched, and the behavior feature information of the sample user;

[0039] The second estimation model is trained based on the probability that the target sample user performs the target behavior under the guidance of each to-be-matched user and the first annotation information.

[0040] In a second aspect, the present disclosure also provides a behavior guidance device, including:

[0041] A first acquisition module is used to acquire behavior characteristic information of at least one user to be screened; wherein the user to be screened is a user who has not performed a target behavior;

[0042] A screening module, configured to screen out target users that meet the 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;

[0043] A second acquisition module, used to acquire the marked first personalized feature information for the target user and the marked second personalized feature information for the user to be matched;

[0044] A determination module is used to determine 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 behavioral feature information of the target user and the first estimation model, wherein the matching user is used to guide the target user to perform the target behavior.

[0045] In a possible implementation, the screening module, when screening out a target user that meets the target behavior execution condition from the at least one user to be screened based on the behavior characteristic information of the at least one user to be screened, is configured to:

[0046] Based on the behavior feature information of the to-be-screened user and the second estimation model, determining an initial estimation result for characterizing the behavior probability of the to-be-screened user performing the target behavior;

[0047] Based on the initial estimation result, target users whose corresponding behavior probabilities meet preset screening conditions are screened out from the users to be screened.

[0048] In a possible implementation manner, the behavior characteristic information of the user to be screened includes at least one of the following information:

[0049] Click-through behavior sequence, historical participation information for target behavior, user attribute information, and interaction information.

[0050] In a possible implementation manner, the first personalized characteristic information of the target user includes: the reason why the target user does not perform the target behavior and / or personality characteristic information of the target user.

[0051] In a possible implementation manner, the second personalized characteristic information of the user to be matched includes: attribute information of the user to be matched and / or behavior guidance characteristic information of the user to be matched.

[0052] In a possible implementation, the screening module, when determining an initial estimation result for characterizing the probability of the to-be-screened user performing the target behavior based on the behavior feature information of the to-be-screened user and the second estimation model, is used to:

[0053] Determine a first representation matrix corresponding to the behavior characteristic information of the user to be screened;

[0054] The first characterization matrix is ​​input into the second estimation model to determine the initial estimation result.

[0055] In a possible implementation, when the behavior characteristic information includes a plurality of behavior characteristic information, the screening module, when determining the first representation matrix corresponding to the behavior characteristic information of the to-be-screened user, is used to:

[0056] Determine a representation vector corresponding to each behavior feature information, and a concatenation order corresponding to the plurality of behavior feature information;

[0057] The representation vectors corresponding to the plurality of behavior feature information are concatenated in the concatenation order to obtain the first representation matrix.

[0058] In a possible implementation manner, when the screening module screens out target users whose corresponding behavior probabilities meet preset screening conditions from the users to be screened based on the initial estimation result, it is configured to:

[0059] Determine the to-be-screened user whose corresponding behavior probability is greater than a preset probability as the target user; or,

[0060] The users to be screened are sorted from large to small according to the corresponding behavior probabilities, and the top N users to be screened are determined as the target users, where N is a preset positive integer.

[0061] In a possible implementation manner, the screening module, 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 behavioral feature information of the target user, and the first estimation model, is used to:

[0062] Determine a second representation matrix corresponding to the first personalized feature information of the target user, the second personalized feature information of the to-be-matched user, and the behavior feature information of the target user;

[0063] Inputting the second characterization matrix into the first estimation model to determine a target estimation result, wherein the target estimation result is used to represent the probability that the target user performs the target behavior under the guidance of each to-be-matched user;

[0064] Based on the target estimation result, a corresponding matching user is determined for each target user.

[0065] In a possible implementation manner, the behavior guidance device further includes a training module, and the training module is used to:

[0066] Acquire behavior feature information and first annotation information of the sample user, wherein the first annotation information is used to indicate whether the sample user performs the target behavior;

[0067] Determining the behavior probability of the sample user performing the target behavior based on the behavior feature information of the sample user and the second estimation model to be trained;

[0068] The second estimation model is trained based on the first annotation information and the behavior probability.

[0069] In a possible implementation manner, the training module is further used to:

[0070] Acquire behavior feature information and second annotation information of the sample user, wherein the second annotation information is used to indicate whether the sample user performs the target behavior;

[0071] Based on the trained second estimation model and the behavior characteristic information of the sample users, target sample users meeting the target behavior execution condition are screened out from the sample users;

[0072] Acquire the first personalized feature information marked for the target sample user and the second personalized feature information marked for the user to be matched;

[0073] Determine the probability that the target sample user performs the target behavior under the guidance of each user to be matched based on the first estimation model to be trained, the first personalized feature information of the target sample user, the second personalized feature information of the user to be matched, and the behavior feature information of the sample user;

[0074] The second estimation model is trained based on the probability that the target sample user performs the target behavior under the guidance of each to-be-matched user and the first annotation information.

[0075] In a third aspect, an optional implementation of the present disclosure further provides a computer device, a processor, and a memory, wherein the memory stores machine-readable instructions executable by the processor, and the processor is used 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 perform the steps of the above-mentioned first aspect, or any possible implementation of the first aspect.

[0076] In a fourth aspect, an optional implementation of the present disclosure further provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed, the steps of the above-mentioned first aspect, or any possible implementation of the first aspect are executed.

[0077] The behavior guidance method, device, computer equipment and storage medium provided by the embodiments of the present disclosure first select target users that meet the target behavior execution conditions from the users to be screened based on the obtained behavior feature information of the users to be screened; then determine the corresponding matching users for each target user based on the obtained first personalized feature information of the target user, the second personalized feature information of the users to be matched, the behavior feature information of the target user and the first estimation model. In this way, a step of screening can be performed based on the behavior feature information of the users to be screened, and then matching users can be determined for the selected target users, which can improve the guidance efficiency of the target behavior while improving the guidance effect of the target behavior.

[0078] In order to make the above-mentioned objectives, features and advantages of the present disclosure more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the following is a brief introduction to the drawings required for use in the embodiments. The drawings herein are incorporated into the specification and constitute a part of the specification. These drawings illustrate embodiments consistent with the present disclosure and are used together with the specification to illustrate the technical solutions of the present disclosure. It should be understood that the following drawings only illustrate certain embodiments of the present disclosure and should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can also be obtained based on these drawings without creative work.

[0080] Figure 1 A flow chart of a behavior guidance method provided by an embodiment of the present disclosure is shown;

[0081] Figure 2 A flowchart of a training process of a second estimation model in a behavior guidance method provided by an embodiment of the present disclosure is shown;

[0082] Figure 3 A flowchart of a training process of a first estimation model in a behavior guidance method provided by an embodiment of the present disclosure is shown;

[0083] Figure 4 A schematic diagram of a behavior guiding device 400 provided in an embodiment of the present disclosure is shown;

[0084] Figure 5 A schematic diagram of a computer device 500 provided by an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0085] In order to make the purpose, technical scheme and advantages of the embodiments of the present disclosure clearer, the technical scheme in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only part of the embodiments of the present disclosure, rather than all of the embodiments. The components of the embodiments of the present disclosure generally described and shown here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present disclosure is not intended to limit the scope of the present disclosure claimed for protection, but merely represents the selected embodiments of the present disclosure. Based on the embodiments of the present disclosure, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present disclosure.

[0086] According to research, in order to encourage users to perform certain preset behaviors and increase user stickiness with online education platforms, such as encouraging users to consume, users are generally guided by the staff of online education platforms. However, in related technologies, all users who interact with online education platforms are generally guided by staff, which is inefficient and has poor guidance effects.

[0087] Based on the above research, the behavior guidance method, device, computer equipment and storage medium provided by the embodiment of the present disclosure first selects target users that meet the target behavior execution conditions from the users to be screened based on the obtained behavior feature information of the users to be screened; then determines the corresponding matching users for each target user based on the obtained first personalized feature information of the target user, the second personalized feature information of the users to be matched, the behavior feature information of the target user and the first estimation model. In this way, a step of screening can be performed based on the behavior feature information of the users to be screened, and then matching users can be determined for the screened target users, which can improve the guidance efficiency of the target behavior while improving the guidance effect of the target behavior.

[0088] The defects existing in the above solutions are the results obtained by the inventor after practice and careful research. Therefore, the discovery process of the above problems and the solutions proposed by the present disclosure for the above problems below should be the contributions made by the inventor to the present disclosure during the disclosure process.

[0089] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0090] To facilitate understanding of this embodiment, a behavior guidance method disclosed in the embodiment of the present disclosure is first introduced in detail. The execution subject of the behavior guidance method provided in the embodiment of the present disclosure is generally a computer device with certain computing capabilities, and the computer device includes, for example: a terminal device or a server or other processing device, and the terminal device can be a user equipment (User Equipment, UE), a mobile device, a user terminal, a terminal, a cellular phone, a cordless phone, a personal digital assistant (Personal Digital Assistant, PDA), a handheld device, a computing device, a vehicle-mounted device, a wearable device, etc. In some possible implementations, the behavior guidance method can be implemented by a processor calling a computer-readable instruction stored in a memory.

[0091] The behavior guidance method provided by the embodiment of the present disclosure is described below by taking the execution subject as a terminal device as an example.

[0092] See also Figure 1 FIG. 1 is a flow chart of a behavior guidance method provided by an embodiment of the present disclosure, wherein the method includes steps S101 to S104, wherein:

[0093] S101: Obtaining behavioral feature information of at least one user to be screened; wherein the user to be screened is a user who has not performed a target behavior.

[0094] S102: Based on the behavior feature information of the at least one user to be screened, a target user that meets the target behavior execution condition is screened out from the at least one user to be screened.

[0095] S103: Acquire the first personalized feature information marked for the target user and the second personalized feature information marked for the user to be matched.

[0096] S104: Based on the acquired first personalized feature information of the target user, the second personalized feature information of the user to be matched, the behavioral feature information of the target user and the first estimation model, a matching user corresponding to each target user is determined from the users to be matched, and the matching user is used to guide the target user to perform the target behavior.

[0097] Here are detailed instructions for the above steps:

[0098] For S101,

[0099] In a possible application scenario, the to-be-screened users may be students or trainees who have participated in an online class, and the online class may be a recorded video class or a live online video class.

[0100] In a possible implementation manner, the behavior characteristic information of the user to be screened includes at least one of the following information:

[0101] Click-through behavior sequence, historical participation information for target behavior, user attribute information, and interaction information.

[0102] The click and browse behavior sequence may be a sequence formed by arranging the click and browse behaviors of the to-be-screened user using the terminal device on the preset website / application in a certain order;

[0103] The historical participation information for the target behavior represents the participation of the to-be-screened user in the target behavior in the past preset time period. For example, if the target behavior is to purchase an online course, the historical participation information for the target behavior may be the user's trial listening time, number of trial listening times, trial listening feedback, etc.;

[0104] The user attribute information may include basic attributes such as age, gender, and region of the user to be screened. If the target behavior is to purchase an online course, the user attribute information may also include the user's learning stage;

[0105] The interaction information represents the interaction between the to-be-screened user and other persons based on the target behavior, and may include, for example, the number of interactions with other persons, the content of the interactions, and other information.

[0106] When the obtained behavior characteristic information of the user to be screened includes user attribute information, in a possible implementation manner, the behavior characteristic information corresponding to the user to be screened may be obtained from a stored user characteristic information library.

[0107] Here, the user can create an account on the client of the terminal device. When creating an account, the user terminal can ask questions corresponding to the characteristic information, and reserve space for the user to fill in. After the user fills in, the user characteristic information is sent to the user characteristic information library for storage. Exemplarily, the list of questions in the display page when filling in the questions corresponding to the user characteristic information can be shown in Table 1:

[0108] Table 1

[0109]

[0110] When the behavior characteristic information includes a click-browsing behavior sequence, when obtaining the behavior characteristic information of the user to be screened, it is possible to obtain pre-stored historical operation data of the user to be screened, and determine the corresponding click-browsing behavior sequence based on the historical operation data. Here, the historical operation data includes historical browsing data and historical click data.

[0111] Taking obtaining a click-through behavior sequence as an example, the user's various click behaviors can be sorted in advance, and then the user's historical operation data can be arranged according to the sorting results of the click behaviors to obtain a click-through behavior sequence. For example, the four operations of whether the user clicked "Advertisement 1", "Link 2", "Image 3", and "Video 4" can be arranged in sequence, and then based on the user's historical operation data, the corresponding value of each item is determined to obtain a click-through behavior sequence corresponding to the historical operation data.

[0112] For example, take the binary classification (i.e., "1" represents "yes" and "0" represents "no") of the click-through behavior sequence as an example. When the user clicks "advertisement 1" and "link 2" but does not click "image 3" and "video 4", the corresponding click-through behavior sequence generated is: [1 1 0 0]

[0114] For S102,

[0115] In a possible implementation, when selecting a target user who meets the target behavior execution conditions from the at least one user to be screened, some screening conditions can be preset for the behavior characteristic information, and then the user whose corresponding behavior characteristic information among the users to be screened meets the preset conditions is determined as the target user.

[0116] In another possible implementation, when a target user that meets the target behavior execution condition is screened out from the at least one user to be screened based on the behavior characteristic information of the at least one user to be screened, an initial prediction result for characterizing the behavior probability of the user to be screened performing the target behavior can be determined based on the behavior characteristic information of the user to be screened and a second prediction model; then, based on the initial prediction result, target users whose corresponding behavior probability meets the preset screening condition are screened out from the users to be screened. Exemplarily, the second prediction model can be a binary classification random forest (RF) model with screening capability.

[0117] In one possible implementation, when determining an initial estimation result for characterizing the behavioral probability of the user to be screened performing a target behavior based on the behavioral characteristic information of the user to be screened and a second estimation model, a first characterization matrix corresponding to the behavioral characteristic information of the user to be screened can be first determined; then, the first characterization matrix can be input into the second estimation model to determine the initial estimation result.

[0118] In the case where the behavior characteristic information includes multiple behavior characteristic information, when determining the first characterization matrix corresponding to the behavior characteristic information of the user to be screened, for example, a corresponding condition can be set for each characteristic information, and then based on whether this condition is met, a binary classification (that is, "1" represents "yes" and "0" represents "no") processing is performed to determine the characterization vector corresponding to each behavior characteristic information, and determine the splicing order corresponding to the multiple behavior characteristic information; then, the characterization vectors corresponding to the multiple behavior characteristic information are spliced ​​according to the splicing order to obtain the first characterization matrix.

[0119] In the case where the behavior characteristic information includes multiple behavior characteristic information, when determining the first characterization matrix corresponding to the behavior characteristic information of the user to be screened, for example, the behavior characteristic information of the user to be screened can be arranged in a preset order to obtain a feature sequence of each user to be screened; then these feature sequences are transformed based on feature coding to obtain corresponding feature vectors; finally, the feature sequences of each user to be screened are merged to obtain a first characterization matrix for characterizing the behavior characteristic information. Wherein, the feature coding method can be mult-hot coding.

[0120] For example, when user 1 clicks "Advertisement 1", "Link 2", "Picture 3", and "Video 4"; user 2 does not click "Advertisement 1" and "Link 2", but clicks "Picture 3" and "Video 4"; user 3 clicks "Advertisement 1" and "Link 2", but does not click "Picture 3" and "Video 4"; user 4 does not click "Advertisement 1", "Link 2", "Picture 3", and "Video 4", the first representation matrix is ​​obtained as follows:

[0121]

[0122] Among them, the value of the first column indicates whether "Advertisement 1" is clicked, the value of the second column indicates whether "Link 2" is clicked, the value of the third column indicates whether "Image 3" is clicked, and the value of the fourth column indicates whether "Video 4" is clicked.

[0123] Then, the first characterization matrix is ​​input into the second estimation model to obtain an initial estimation result for characterizing the behavior probability of the to-be-screened user performing the target behavior.

[0124] Exemplarily, still taking the above users 1 to 4 as an example, after inputting the first characterization matrix corresponding to each user into the second estimation model, the initial estimation result output by the second estimation model can be obtained as follows: [0.8 0.75 0.75 0.5]

[0126] Among them, "0.8", "0.75", "0.75" and "0.5" respectively represent the prediction results of the second prediction model for users 1 to 4, that is, user 1 has an 80% probability of performing the target behavior; user 2 has a 75% probability of performing the target behavior; user 3 has a 75% probability of performing the target behavior; and user 4 has a 50% probability of performing the target behavior.

[0127] Specifically, in the same time period, there may be many users who perform the target behavior, while the time, energy and number of people who can be guided by the users to be matched are also limited. In order to improve the efficiency of guiding the target behavior, the initial estimation result obtained by the second estimation model can be used as the basis for screening, so as to determine the corresponding behavior probability among the users to be screened that meets the preset screening conditions, that is, the target users with higher guidance value.

[0128] When selecting target users whose corresponding behavior probabilities meet the preset screening conditions from the users to be screened based on the initial estimation results, any one of the following methods may be used:

[0129] Method A: Sort the users to be screened from large to small according to the corresponding behavior probabilities, and determine the top N users to be screened as the target users, where N is a preset positive integer.

[0130] Specifically, the behavior probabilities of the to-be-screened users can 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 sorted sequence numbers are "1", "2", "2", and "4", the estimated value of the sequence number "1" and any one of the two behavior probabilities of the sequence number "2" can be selected; or, the behavior probabilities of the sequence number "1" and the two sequence numbers "2" can be selected at the same time.

[0131] Exemplarily, still taking the above-mentioned users 1 to 4 as an example, the four behavior probabilities of "0.8", "0.75", "0.75", and "0.5" can be ranked from high to low, and the three users to be screened corresponding to the first two estimated values ​​"0.8" and "0.75" can be determined as the target users, or the two users to be screened corresponding to the estimated value "0.8" and the first or second "0.75" can be determined as the target users.

[0132] Method B: Determine the to-be-screened users whose corresponding behavior probabilities are greater than a preset probability as the target users.

[0133] Specifically, a preset probability M can be set within the range of 0 to 1 based on actual screening needs, such as 60 points (60%) for qualified and 90 points (90%) for excellent.

[0134] Exemplarily, still taking the above users 1 to 4 as an example, when the preset probability M is 0.6, the three to-be-screened users corresponding to the behavior probabilities of "0.8" and "0.75" can be determined as the target users.

[0135] For S103,

[0136] In a possible implementation, in order to make the feature information of the target user and the user to be matched more abundant and real, the first personalized feature of the target user may be manually labeled, and the labeling method may be binary classification, that is, "1" represents "yes" and "0" represents "no".

[0137] The first personalized characteristic information of the target user may include, for example, the reason why the target user does not perform the target behavior and / or the personality characteristic information of the target user.

[0138] Here, the reason for not executing the target behavior can be various possible reasons that have been counted in advance and numbered, and then for each target user, the corresponding number of not executing the target behavior can be marked, and after identifying the corresponding number, the terminal device can perform the corresponding coding. For example, 10 reasons for not executing the target behavior are preset, namely reasons 1 to 10, and the reason why the target user 1 did not execute the target behavior is reason 5, then the corresponding code generated is: [0 0 0 0 1 0 0 0 0 0]

[0139] The optional reason why the target user does not perform the target behavior may be a summary of reasons why multiple users do not perform the target behavior.

[0140] In a possible implementation manner, the second personalized characteristic information of the user to be matched includes: attribute information of the user to be matched and / or behavior guidance characteristic information of the user to be matched.

[0141] Here, the user to be matched may be a shopping guide in a shopping mall, a live streamer selling goods, a teacher teaching a class, or other user with behavior guidance capabilities.

[0142] Among them, taking the user to be matched as a teacher as an example, the attribute information of the user to be matched may be basic information such as the historical conversion rate, homework correction rate, and teaching time of the teacher to be matched; the behavior guidance feature information of the user to be matched may be the personal business expertise of the teacher to be matched, the logic and methodology for guiding students to perform target behaviors, and other feature information.

[0143] For S104,

[0144] When determining the 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 behavioral feature information of the target user, and the first estimation model, a second characterization matrix corresponding to the first personalized feature information of the target user, the second personalized feature information of the user to be matched, and the behavioral feature information of the target user can be first determined; then, the second characterization matrix is ​​input into the first estimation model to determine a target estimation result, wherein the target estimation result is used to represent the probability that the target user performs the target behavior under the guidance of each user to be matched; finally, based on the target estimation result, a corresponding matching user is determined for each target user.

[0145] Specifically, the first personalized feature information of the target user, the second personalized feature information of the user to be matched, and the behavior feature information of the target user can be transformed based on feature coding, and then arranged in a preset order to obtain a second representation matrix. The feature coding can be mult-hot coding.

[0146] For example, target user 1 is marked as aggressive, and clicked "Advertisement 1", "Link 2", "Picture 3", and "Video 4"; target user 2 is marked as not aggressive, and did not click "Advertisement 1" and "Link 2", but clicked "Picture 3" and "Video 4"; target user 3 is marked as not aggressive, and clicked "Advertisement 1" and "Link 2", but did not click "Picture 3" and "Video 4"; the historical conversion rate of user 1 to be matched is high; the historical conversion rate of user 2 to be matched is low. By arranging and combining the three target users and the two users to be matched, six matching methods can be obtained, and the second characterization matrix obtained is, for example:

[0147]

[0148] Among them, the first column indicates whether the corresponding target user is marked as aggressive; the second column indicates whether the corresponding target user clicked "Advertisement 1"; the third column indicates whether the corresponding target user clicked "Link 2"; the fourth column indicates whether the corresponding target user clicked "Image 3"; the fifth column indicates whether the corresponding target user clicked "Video 4"; the sixth column indicates the historical conversion rate of the corresponding to-be-matched user. The first and second rows represent target user 1; the third and fourth rows represent target user 2; the fifth and sixth rows represent target user 3. The sixth element of the first, third, and fifth rows represents to-be-matched user 1, and the sixth element of the second, fourth, and sixth rows represents to-be-matched user 2.

[0149] Then, the second representation matrix is ​​input into the first estimation model to generate the target estimation result. The first estimation model may be an extreme gradient boosting (xgb) model of binary classification with screening and matching capabilities.

[0150] Exemplarily, still taking the above target users 1 to 3 and to-be-matched users 1 and 2 as an example, after the second representation matrix is ​​input into the first estimation model, the target estimation result output by the first estimation model can be obtained as follows:

[0151]

[0152] Among them, "0.9" and "0.6" in the first row represent that the probability of target user 1 performing the target behavior under the guidance of to-be-matched user 1 is 90%, and the probability of performing the target behavior under the guidance of to-be-matched user 2 is 60%; the second row and the third row represent the probability of target user 2 and target user 3 performing the target behavior under the guidance of to-be-matched user 1 and to-be-matched user 2, respectively. Furthermore, the probability of a target user performing the target behavior under the guidance of a to-be-matched user can represent the degree of matching with the to-be-matched user.

[0153] Furthermore, the target estimation result can be converted into an estimation result table as shown in Table 2:

[0154] Table 2

[0155] User to be matched 1 User 2 to be matched Target User 1 90％ 60％ Target User 2 50％ 70％ Target User 3 60％ 65％

[0156] After obtaining the estimation result table, it can be found that the user to be matched 1 has the highest matching degree with target user 1; the user to be matched 2 has the highest matching degree with target user 2; and the user to be matched 2 has the highest matching degree with target user 3.

[0157] After determining the user to be matched who has the highest matching degree with the target user, the guidance experience, specialties, contact information and other information of the user to be matched who has the highest matching degree can be sent to the corresponding target user; at the same time, the specialties, contact information and other information of the target user can also be sent to the corresponding user to be matched. In this way, a connection can be established between the target user and the user to be matched, which is convenient for subsequent behavior guidance.

[0158] The present disclosure also provides a method for training a second estimation model, see Figure 2 As shown, it is a flowchart of a training method of a second estimation model provided by an embodiment of the present disclosure, comprising the following steps:

[0159] S201. Obtain behavior feature information and first annotation information of a sample user, wherein the first annotation information is used to indicate whether the sample user performs the target behavior.

[0160] S202: Determine the behavior probability of the sample user performing the target behavior based on the behavior feature information of the sample user and the second estimation model to be trained.

[0161] S203: Training the second estimation model based on the first annotation information and the behavior probability.

[0162] Here, when obtaining the behavioral characteristic information and first annotation information of sample users used to train the second estimation model, the period of large-scale execution of the target behavior can be determined as the window period based on historical data. For example, 3-7 days after the end of the free trial class / low-priced trial class is the window period. The behavioral characteristic information and first annotation information of sample users obtained for training the second estimation model are the data corresponding to the window period.

[0163] In a possible implementation, after obtaining the behavioral feature information and the first annotation information of the sample user, the information can be cleaned and screened to extract the important parts to improve the training efficiency of the second prediction model. For example, based on the understanding of the business, browsing and clicking behaviors with important business significance can be screened out from numerous events; and meaningless information such as meaningless icons, pictures, user IDs, duplicate information, and network abnormality logs in the interactive information of the sample user can be removed.

[0164] When training the second prediction model, the second prediction model can output the probability of the target user performing the target behavior. Based on the probability of the target user performing the target behavior and a pre-set probability threshold, the prediction result of the second prediction model can be determined, and the prediction result can indicate whether the target user will perform the target behavior. Based on the prediction result of the second prediction model and the first labeled data, the loss value in this training process can be determined, and then based on the determined loss value, the model parameters of the second prediction model can be adjusted.

[0165] In practical applications, after the training of the second prediction model is completed, it can also be optimized multiple times during the application process. Specifically, the latest result of whether the user performs the target behavior and the predicted result can be matched and stored to improve the sample data; then based on the improved sample data, the second prediction model is retrained to achieve continuous optimization of the second prediction model.

[0166] The present disclosure also provides a method for training a first estimation model, see Figure 3 As shown, it is a flowchart of a training method of a first estimation model provided by an embodiment of the present disclosure, comprising the following steps:

[0167] S301. Obtain behavior feature information and second annotation information of a sample user, wherein the second annotation information is used to indicate whether the sample user performs the target behavior.

[0168] S302: based on the trained second estimation model and the behavior characteristic information of the sample users, select target sample users that meet the target behavior execution conditions from the sample users.

[0169] S303: Obtain the first personalized feature information marked for the target sample user and the second personalized feature information marked for the to-be-matched user.

[0170] S304: Based on the first estimation model to be trained, the first personalized feature information of the target sample user, the second personalized feature information of the user to be matched, and the behavior feature information of the sample user, determine the probability that the target sample user performs the target behavior under the guidance of each user to be matched.

[0171] S305: Training the second estimation model based on the probability that the target sample user performs the target behavior under the guidance of each to-be-matched user and the first annotation information.

[0172] Here, the behavior feature information and second annotation information of the sample users obtained for training the first estimation model may be data obtained after the window period.

[0173] When training the first prediction model, the first prediction model can output the probability that the target user performs the target behavior under the guidance of each user to be matched. Based on the probability that the target user performs the target behavior under the guidance of each user to be matched and the pre-set probability threshold, the prediction result of the second prediction model can be determined. The prediction result can indicate whether the target user will perform the target behavior under the guidance of the user to be matched. Based on the prediction result of the second prediction model and the first labeled data, the loss value in this training process can be determined, and then based on the determined loss value, the model parameters of the first prediction model can be adjusted.

[0174] In practical applications, after the training of the second prediction model is completed, it can also be optimized multiple times during the application process. Specifically, the result of whether the user performs the target behavior and the result of the model prediction can be matched and stored to improve the sample data; then based on the improved sample data, the first prediction model is retrained to achieve continuous optimization of the first prediction model.

[0175] The behavior guidance method provided by the embodiment of the present disclosure first selects target users that meet the target behavior execution conditions from the users to be screened based on the obtained behavior feature information of the users to be screened; then determines the corresponding matching users for each target user based on the obtained first personalized feature information of the target user, the second personalized feature information of the users to be matched, the behavior feature information of the target user, and the first estimation model. In this way, a step of screening can be performed based on the behavior feature information of the users to be screened, and then matching users can be determined for the screened target users, which can improve the guidance efficiency of the target behavior while improving the guidance effect of the target behavior.

[0176] Those skilled in the art will appreciate that, in the above method of specific implementation, the order in which the steps are written does not imply a strict execution order and does not constitute any limitation on the implementation process. The specific execution order of the steps should be determined by their functions and possible internal logic.

[0177] Based on the same inventive concept, a behavior guidance device corresponding to the behavior guidance method is also provided in the embodiment of the present disclosure. Since the principle of solving the problem by the device in the embodiment of the present disclosure is similar to the above-mentioned behavior guidance method in the embodiment of the present disclosure, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be repeated.

[0178] Reference Figure 4 , which is a schematic diagram of a behavior guidance device 400 provided in an embodiment of the present disclosure, the device comprises: a first acquisition module 401, a screening module 402, a second acquisition module 403, and a determination module 404; wherein,

[0179] The first acquisition module 401 is used to acquire behavior characteristic information of at least one user to be screened; wherein the user to be screened is a user who has not performed the target behavior;

[0180] A screening module 402 is used to screen out target users that meet the target behavior execution condition from the at least one user to be screened based on the behavior characteristic information of the at least one user to be screened;

[0181] A second acquisition module 403 is used to acquire the first personalized feature information marked for the target user and the second personalized feature information marked for the user to be matched;

[0182] The determination module 404 is used to determine the 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 behavioral feature information of the target user and the first estimation model, and the matching user is used to guide the target user to perform the target behavior.

[0183] In a possible implementation, the screening module 402, when screening out a target user that meets the target behavior execution condition from the at least one user to be screened based on the behavior characteristic information of the at least one user to be screened, is configured to:

[0184] Based on the behavior feature information of the to-be-screened user and the second estimation model, determining an initial estimation result for characterizing the behavior probability of the to-be-screened user performing the target behavior;

[0185] Based on the initial estimation result, target users whose corresponding behavior probabilities meet preset screening conditions are screened out from the users to be screened.

[0186] In a possible implementation manner, the behavior characteristic information of the user to be screened includes at least one of the following information:

[0187] Click-through behavior sequence, historical participation information for target behavior, user attribute information, and interaction information.

[0188] In a possible implementation manner, the first personalized characteristic information of the target user includes: the reason why the target user does not perform the target behavior and / or personality characteristic information of the target user.

[0189] In a possible implementation manner, the second personalized characteristic information of the user to be matched includes: attribute information of the user to be matched and / or behavior guidance characteristic information of the user to be matched.

[0190] In a possible implementation, the screening module 402, when determining the initial estimation result for characterizing the probability of the to-be-screened user performing the target behavior based on the behavior feature information of the to-be-screened user and the second estimation model, is used to:

[0191] Determine a first representation matrix corresponding to the behavior characteristic information of the user to be screened;

[0192] The first characterization matrix is ​​input into the second estimation model to determine the initial estimation result.

[0193] In a possible implementation, when the behavior characteristic information includes a plurality of behavior characteristic information, the screening module 402, when determining the first representation matrix corresponding to the behavior characteristic information of the to-be-screened user, is configured to:

[0194] Determine a representation vector corresponding to each behavior feature information, and a concatenation order corresponding to the plurality of behavior feature information;

[0195] The representation vectors corresponding to the plurality of behavior feature information are concatenated in the concatenation order to obtain the first representation matrix.

[0196] In a possible implementation, the screening module 402, when screening out target users whose corresponding behavior probabilities meet preset screening conditions from the users to be screened based on the initial estimation result, is configured to:

[0197] Determine the to-be-screened user whose corresponding behavior probability is greater than a preset probability as the target user; or,

[0198] The users to be screened are sorted from large to small according to the corresponding behavior probabilities, and the top N users to be screened are determined as the target users, where N is a preset positive integer.

[0199] In a possible implementation manner, the screening module 402, 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 behavioral feature information of the target user, and the first estimation model, is used to:

[0200] Determine a second representation matrix corresponding to the first personalized feature information of the target user, the second personalized feature information of the to-be-matched user, and the behavior feature information of the target user;

[0201] Inputting the second characterization matrix into the first estimation model to determine a target estimation result, wherein the target estimation result is used to represent the probability that the target user performs the target behavior under the guidance of each to-be-matched user;

[0202] Based on the target estimation result, a corresponding matching user is determined for each target user.

[0203] In a possible implementation manner, the behavior guidance device further includes a training module 405, and the training module is used to:

[0204] Acquire behavior feature information and first annotation information of the sample user, wherein the first annotation information is used to indicate whether the sample user performs the target behavior;

[0205] Determining the behavior probability of the sample user performing the target behavior based on the behavior feature information of the sample user and the second estimation model to be trained;

[0206] The second estimation model is trained based on the first annotation information and the behavior probability.

[0207] In a possible implementation manner, the training module 405 is further used to:

[0208] Acquire behavior feature information and second annotation information of the sample user, wherein the second annotation information is used to indicate whether the sample user performs the target behavior;

[0209] Based on the trained second estimation model and the behavior characteristic information of the sample users, target sample users meeting the target behavior execution condition are screened out from the sample users;

[0210] Acquire the first personalized feature information marked for the target sample user and the second personalized feature information marked for the user to be matched;

[0211] Determine the probability that the target sample user performs the target behavior under the guidance of each user to be matched based on the first estimation model to be trained, the first personalized feature information of the target sample user, the second personalized feature information of the user to be matched, and the behavior feature information of the sample user;

[0212] The second estimation model is trained based on the probability that the target sample user performs the target behavior under the guidance of each to-be-matched user and the first annotation information.

[0213] The behavior guidance device provided by the embodiment of the present disclosure first selects target users that meet the target behavior execution conditions from the users to be screened based on the obtained behavior feature information of the users to be screened; and then determines the corresponding matching users for each target user based on the obtained first personalized feature information of the target user, the second personalized feature information of the users to be matched, the behavior feature information of the target user, and the first estimation model. In this way, a step of screening can be performed based on the behavior feature information of the users to be screened, and then matching users can be determined for the screened target users, which can improve the guidance efficiency of the target behavior while improving the guidance effect of the target behavior.

[0214] For descriptions of the processing flow of each module in the device and the interaction flow between each module, reference may be made to the relevant descriptions in the above method embodiment, which will not be described in detail here.

[0215] The present disclosure also provides a computer device, such as Figure 5 FIG. 5 is a schematic diagram of a computer device 500 provided in an embodiment of the present disclosure, including:

[0216] Based on the same technical concept, the embodiment of the present disclosure also provides a computer device 500. Figure 5 As shown, it is a schematic diagram of the structure of a computer device 500 provided in an embodiment of the present disclosure, including a processor 501, a memory 502, and a bus 503. Among them, the memory 502 is used to store execution instructions, including a memory 5021 and an external memory 5022; the memory 5021 here is also called an internal memory, which is used to temporarily store the operation data in the processor 501, and the data exchanged with the external memory 5022 such as a hard disk. The processor 501 exchanges data with the external memory 5022 through the memory 5021. When the computer device 500 is running, the processor 501 communicates with the memory 502 through the bus 503, so that the processor 501 executes the following instructions:

[0217] Obtaining behavioral characteristic information of at least one user to be screened; wherein the user to be screened is a user who has not performed the target behavior;

[0218] Based on the behavior characteristic information of the at least one to-be-screened user, a target user meeting the target behavior execution condition is screened out from the at least one to-be-screened user;

[0219] Acquire the first personalized feature information marked for the target user and the second personalized feature information marked for the user to be matched;

[0220] Based on the acquired first personalized feature information of the target user, the second personalized feature information of the user to be matched, the behavioral feature information of the target user and the first estimation model, a matching user corresponding to each target user is determined from the users to be matched, and the matching user is used to guide the target user to perform the target behavior.

[0221] In a possible implementation, in the instructions executed by the processor 501, the step of screening out a target user that meets the target behavior execution condition from the at least one user to be screened based on the behavior characteristic information of the at least one user to be screened includes:

[0222] Based on the behavior feature information of the to-be-screened user and the second estimation model, determining an initial estimation result for characterizing the behavior probability of the to-be-screened user performing the target behavior;

[0223] Based on the initial estimation result, target users whose corresponding behavior probabilities meet preset screening conditions are screened out from the users to be screened.

[0224] In a possible implementation, in the instruction executed by the processor 501, the behavior characteristic information of the user to be screened includes at least one of the following information:

[0225] Click-through behavior sequence, historical participation information for target behavior, user attribute information, and interaction information.

[0226] In a possible implementation, in the instruction executed by the processor 501, the first personalized characteristic information of the target user includes: the reason why the target user does not perform the target behavior and / or the personality characteristic information of the target user.

[0227] In a possible implementation, in the instruction executed by the processor 501, the second personalized feature information of the user to be matched includes: attribute information of the user to be matched and / or behavior guidance feature information of the user to be matched.

[0228] In a possible implementation manner, in the instructions executed by the processor 501, determining, based on the behavior feature information of the to-be-screened user and the second estimation model, an initial estimation result for characterizing the behavior probability of the to-be-screened user performing the target behavior includes:

[0229] Determine a first representation matrix corresponding to the behavior characteristic information of the user to be screened;

[0230] The first characterization matrix is ​​input into the second estimation model to determine the initial estimation result.

[0231] In a possible implementation manner, in the instruction executed by the processor 501, when the behavior characteristic information includes multiple behavior characteristic information, determining the first representation matrix corresponding to the behavior characteristic information of the to-be-screened user includes:

[0232] Determine a representation vector corresponding to each behavior feature information, and a concatenation order corresponding to the plurality of behavior feature information;

[0233] The representation vectors corresponding to the plurality of behavior feature information are concatenated in the concatenation order to obtain the first representation matrix.

[0234] In a possible implementation manner, in the instructions executed by the processor 501, the step of screening out target users whose corresponding behavior probabilities meet preset screening conditions from the users to be screened based on the initial estimation result includes:

[0235] Determine the to-be-screened user whose corresponding behavior probability is greater than a preset probability as the target user; or,

[0236] The users to be screened are sorted from large to small according to the corresponding behavior probabilities, and the top N users to be screened are determined as the target users, where N is a preset positive integer.

[0237] In a possible implementation manner, in the instructions executed by the processor 501, 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 behavioral feature information of the target user, and the first estimation model includes:

[0238] Determine a second representation matrix corresponding to the first personalized feature information of the target user, the second personalized feature information of the to-be-matched user, and the behavior feature information of the target user;

[0239] Inputting the second characterization matrix into the first estimation model to determine a target estimation result, wherein the target estimation result is used to represent the probability that the target user performs the target behavior under the guidance of each to-be-matched user;

[0240] Based on the target estimation result, a corresponding matching user is determined for each target user.

[0241] In a possible implementation manner, the instructions executed by the processor 501 further include training the second estimation model according to the following method:

[0242] Acquire behavior feature information and first annotation information of the sample user, wherein the first annotation information is used to indicate whether the sample user performs the target behavior within a target time period;

[0243] Determining the behavior probability of the sample user performing the target behavior based on the behavior feature information of the sample user and the second estimation model to be trained;

[0244] The second estimation model is trained based on the first annotation information and the behavior probability.

[0245] In a possible implementation manner, the instructions executed by the processor 501 further include training the first estimation model according to the following method:

[0246] Acquire behavior feature information and second annotation information of the sample user, wherein the second annotation information is used to indicate whether the sample user performs the target behavior after the target time period;

[0247] Based on the trained second estimation model and the behavior characteristic information of the sample users, target sample users meeting the target behavior execution condition are screened out from the sample users;

[0248] Acquire the first personalized feature information marked for the target sample user and the second personalized feature information marked for the user to be matched;

[0249] Determine the probability that the target sample user performs the target behavior under the guidance of each user to be matched based on the first estimation model to be trained, the first personalized feature information of the target sample user, the second personalized feature information of the user to be matched, and the behavior feature information of the sample user;

[0250] The second estimation model is trained based on the probability that the target sample user performs the target behavior under the guidance of each to-be-matched user and the first annotation information.

[0251] The embodiment of the present disclosure also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the behavior guidance method described in the above method embodiment are executed. The storage medium can be a volatile or non-volatile computer-readable storage medium.

[0252] The computer program product of the behavior guidance method provided in the embodiments of the present disclosure includes a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the steps of the behavior guidance method described in the above method embodiments. Please refer to the above method embodiments for details, which will not be repeated here.

[0253] The present disclosure also provides a computer program, which implements any one of the methods of the aforementioned embodiments when executed by a processor. The computer program product can be implemented in hardware, software, or a combination thereof. In an optional embodiment, the computer program product is embodied as a computer storage medium, and in another optional embodiment, the computer program product is embodied as a software product, such as a software development kit (SDK), etc.

[0254] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, the specific working process of the system and device described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here. In the several embodiments provided in the present disclosure, it should be understood that the disclosed system, device and method can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of the device or unit can be electrical, mechanical or other forms.

[0255] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0256] In addition, each functional unit in each embodiment of the present disclosure may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0257] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium that is executable by a processor. Based on this understanding, the technical solution of the present disclosure, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present disclosure. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0258] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present disclosure, which are used to illustrate the technical solutions of the present disclosure, rather than to limit them. The protection scope of the present disclosure is not limited thereto. Although the present disclosure is described in detail with reference to the above-described embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-described embodiments within the technical scope disclosed in the present disclosure, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present disclosure, and should be included in the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be based on the protection scope of the claims.

Claims

1. A behavior guidance method, characterized in that: include: Obtaining behavioral characteristic information of at least one user to be screened; wherein the user to be screened is a user who has not performed the target behavior; Based on the behavior characteristic information of the at least one to-be-screened user, a target user that meets the target behavior execution condition is screened out from the at least one to-be-screened user; wherein the target user is a user with guidance value; Acquire the first personalized feature information marked for the target user and the second personalized feature information marked for the user to be matched; wherein the first personalized feature information includes the reason why the target user does not perform the target behavior, and the second personalized feature information includes the behavior guidance feature information of the user 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 estimation model, determining a matching user corresponding to each target user from the users to be matched, the matching user being used to guide the target user to perform the target behavior; The step of selecting a target user that meets the target behavior execution condition from the at least one user to be screened based on the behavior characteristic information of the at least one user to be screened includes: Based on the behavior feature information of the user to be screened and a second prediction model, determining an initial prediction result for characterizing the probability of the user to be screened performing a target behavior; the first prediction model and the second prediction model are different; Based on the initial estimation result, target users whose corresponding behavior probabilities meet preset screening conditions are screened out from the users to be screened.

2. The method according to claim 1, characterized in that The behavior characteristic information of the user to be screened includes at least one of the following information: Click-through behavior sequence, historical participation information for target behavior, user attribute information, and interaction information.

3. The method according to claim 1, characterized in that The first personalized characteristic information of the target user also includes: personality characteristic information of the target user.

4. The method according to claim 1, characterized in that The second personalized feature information of the user to be matched also includes: attribute information of the user to be matched.

5. The method according to claim 1, characterized in that The determining, based on the behavior feature information of the to-be-screened user and the second estimation model, an initial estimation result for characterizing the probability of the to-be-screened user performing the target behavior comprises: Determine a first representation matrix corresponding to the behavior characteristic information of the user to be screened; The first characterization matrix is ​​input into the second estimation model to determine the initial estimation result.

6. The method according to claim 5, characterized in that In the case where the behavior characteristic information includes a plurality of behavior characteristic information, determining the first representation matrix corresponding to the behavior characteristic information of the to-be-screened user includes: Determine a representation vector corresponding to each behavior feature information, and a concatenation order corresponding to the plurality of behavior feature information; The representation vectors corresponding to the plurality of behavior feature information are concatenated in the concatenation order to obtain the first representation matrix.

7. The method according to claim 1, characterized in that The step of screening out target users whose corresponding behavior probabilities meet preset screening conditions from the users to be screened based on the initial estimation result includes: Determine the to-be-screened user whose corresponding behavior probability is greater than a preset probability as the target user; or, The users to be screened are sorted from large to small according to the corresponding behavior probabilities, and the top N users to be screened are determined as the target users, where N is a preset positive integer.

8. The method according to claim 1, characterized in that 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 second personalized feature information of the user to be matched, the behavioral feature information of the target user, and the first estimation model includes: Determine a second representation matrix corresponding to the first personalized feature information of the target user, the second personalized feature information of the to-be-matched user, and the behavior feature information of the target user; Inputting the second characterization matrix into the first estimation model to determine a target estimation result, wherein the target estimation result is used to represent the probability that the target user performs the target behavior under the guidance of each to-be-matched user; Based on the target estimation result, a corresponding matching user is determined for each target user.

9. The method according to claim 1, characterized in that: The method further comprises training the second estimation model according to the following method: Acquire behavior feature information and first annotation information of the sample user, wherein the first annotation information is used to indicate whether the sample user performs the target behavior; Determining the behavior probability of the sample user performing the target behavior based on the behavior feature information of the sample user and the second estimation model to be trained; The second estimation model is trained based on the first annotation information and the behavior probability.

10. The method according to claim 9, characterized in that The method further comprises training the first prediction model according to the following method: Acquire behavior feature information and second annotation information of the sample user, wherein the second annotation information is used to indicate whether the sample user performs the target behavior; Based on the trained second estimation model and the behavior characteristic information of the sample users, target sample users meeting the target behavior execution condition are screened out from the sample users; Acquire the first personalized feature information marked for the target sample user and the second personalized feature information marked for the user to be matched; Determine the probability that the target sample user performs the target behavior under the guidance of each user to be matched based on the first estimation model to be trained, the first personalized feature information of the target sample user, the second personalized feature information of the user to be matched, and the behavior feature information of the sample user; The second estimation model is trained based on the probability that the target sample user performs the target behavior under the guidance of each to-be-matched user and the first annotation information.

11. A behavior guidance device, characterized in that: include: A first acquisition module is used to acquire behavior characteristic information of at least one user to be screened; wherein the user to be screened is a user who has not performed a target behavior; A screening module, configured to screen out a target user that meets the target behavior execution condition from the at least one user to be screened based on the behavior characteristic information of the at least one user to be screened; wherein the target user is a user with guidance value; A second acquisition module is used to acquire the first personalized feature information marked for the target user and the second personalized feature information marked for the user to be matched; wherein the first personalized feature information includes the reason why the target user does not perform the target behavior, and the second personalized feature information includes the behavior guidance feature information of the user to be matched; a determination module, configured to determine 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 estimation model, wherein the matching user is used to guide the target user to perform the target behavior; The screening module, when screening out a target user that meets the target behavior execution condition from the at least one user to be screened based on the behavior characteristic information of the at least one user to be screened, is used to: Based on the behavior feature information of the user to be screened and a second prediction model, determining an initial prediction result for characterizing the probability of the user to be screened performing a target behavior; the first prediction model and the second prediction model are different; Based on the initial estimation result, target users whose corresponding behavior probabilities meet preset screening conditions are screened out from the users to be screened.

12. A computer device, characterized in that: include: A processor and a memory, wherein the memory stores machine-readable instructions executable by the processor, and the processor is used to execute the machine-readable instructions stored in the memory. When the machine-readable instructions are executed by the processor, the processor executes the steps of the behavior guidance method as described in any one of claims 1 to 10.

13. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program. When the computer program is executed by a computer device, the computer device executes the steps of the behavior guidance method according to any one of claims 1 to 10.

Images