CN117493661A

CN117493661A - Transaction recommendation method, device, storage medium and equipment

Info

Publication number: CN117493661A
Application number: CN202311299168.2A
Authority: CN
Inventors: 闵旭; 张晓露; 周俊
Original assignee: Alipay Hangzhou Information Technology Co Ltd
Current assignee: Alipay Hangzhou Information Technology Co Ltd
Priority date: 2023-10-08
Filing date: 2023-10-08
Publication date: 2024-02-02

Abstract

The specification discloses a transaction recommendation method, a device, a storage medium and equipment, wherein user data corresponding to a first user and a transaction to be recommended are obtained, the user data comprise first image features and first behavior features, then at least one similar user associated with the first user is determined according to the first image features in the user data, the first behavior features are subjected to feature reconstruction based on second behavior features corresponding to the similar users respectively, the reconstructed behavior features are obtained, finally the conversion rate of the first user for the transaction to be recommended is generated based on the reconstructed behavior features and the transaction to be recommended, and the transaction to be recommended is recommended to the first user based on the conversion rate.

Description

Transaction recommendation method, device, storage medium and equipment

Technical Field

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

Background

With the rapid development and popularization of internet technology, user data has shown explosive growth. Mining valuable behavioral information from a vast amount of user data, recommending interesting application transactions to users based on the user behavioral information is a focus of attention and research in recent years.

Disclosure of Invention

According to the transaction recommending method, the device, the storage medium and the equipment, the conversion rate of the user to the transaction to be recommended can be calculated according to the user data, so that the transaction to be recommended is recommended to the user according to the conversion rate. The technical scheme is as follows:

in a first aspect, embodiments of the present disclosure provide a transaction recommendation method, where the method includes:

acquiring user data and a transaction to be recommended corresponding to a first user, wherein the user data comprises a first image feature and a first behavior feature;

determining at least one similar user associated with the first user according to the first portrait characteristic, and acquiring second behavior characteristics corresponding to the similar users respectively;

performing feature reconstruction on the first behavior feature based on each second behavior feature to obtain a reconstructed behavior feature;

generating a conversion rate of the first user for the transaction to be recommended based on the reconfiguration behavior characteristic and the transaction to be recommended;

recommending the transaction to be recommended to the first user based on the conversion rate.

In a second aspect, embodiments of the present disclosure provide a method for training a conversion prediction model, the method comprising:

Constructing a training sample data set, wherein the training sample data comprises sample user portrait information, sample user behavior information, sample recommended transactions and sample conversion rate label values;

inputting the training sample data into a conversion rate prediction model, extracting sample user portrait features in the sample user portrait information, extracting sample user behavioral features in the sample user behavioral information and extracting sample transaction features in the sample recommended transaction based on the feature extraction network;

generating sample reconstruction behavior characteristics corresponding to the training sample data according to the sample user portrait characteristics by adopting a characteristic reconstruction network;

a conversion rate prediction network is adopted to conduct recommended prediction according to the sample reconstruction behavior characteristics and the sample transaction characteristics, and a sample conversion rate prediction value corresponding to the training sample data is obtained;

and performing supervision training on the conversion rate prediction model based on a preset loss function, the sample user behavior characteristic, the sample reconstruction behavior characteristic, the conversion rate label value and the sample conversion rate prediction value, and iteratively updating model parameters of the conversion rate prediction model until the conversion rate prediction model converges, so as to obtain the trained conversion rate prediction model.

In a third aspect, embodiments of the present disclosure provide a transaction recommendation device, including:

the user data acquisition module is used for acquiring user data and a transaction to be recommended corresponding to a first user, wherein the user data comprises a first image feature and a first behavior feature;

the similar feature determining module is used for determining at least one similar user associated with the first user according to the first portrait feature and acquiring second behavior features corresponding to the similar users respectively;

the behavior feature reconstruction module is used for carrying out feature reconstruction on the first behavior features based on the second behavior features to obtain reconstructed behavior features;

the conversion rate prediction module is used for generating the conversion rate of the first user for the transaction to be recommended based on the reconstruction behavior characteristics and the transaction to be recommended;

and the transaction recommending module is used for recommending the transaction to be recommended to the first user based on the conversion rate.

In a fourth aspect, embodiments of the present disclosure provide a conversion rate prediction model training apparatus, the apparatus including:

the sample data acquisition module is used for constructing a training sample data set, wherein the training sample data comprises sample user portrait information, sample user behavior information, sample recommended transactions and sample conversion rate label values;

The sample feature extraction module is used for inputting the training sample data into a conversion rate prediction model, extracting sample user portrait features in the sample user portrait information, extracting sample user behavioral features in the sample user behavioral information and extracting sample transaction features in the sample recommended transaction based on the feature extraction network;

the reconstruction feature generation module is used for generating sample reconstruction behavior features corresponding to the training sample data according to the sample user portrait features by adopting a feature reconstruction network;

the conversion rate prediction module is used for carrying out recommended prediction according to the sample reconstruction behavior characteristics and the sample transaction characteristics by adopting a conversion rate prediction network to obtain a sample conversion rate prediction value corresponding to the training sample data;

and the model supervision and training module is used for carrying out supervision and training on the conversion rate prediction model based on a preset loss function, the sample user behavior characteristic, the sample reconstruction behavior characteristic, the conversion rate label value and the sample conversion rate prediction value, and iteratively updating model parameters of the conversion rate prediction model until the conversion rate prediction model converges, so as to obtain the conversion rate prediction model after training.

In a fifth aspect, the present description embodiments provide a computer program product storing at least one instruction adapted to be loaded by a processor and to perform the above-described method steps.

In a sixth aspect, the present description provides a storage medium storing a computer program adapted to be loaded by a processor and to perform the above-mentioned method steps.

In a seventh aspect, embodiments of the present disclosure provide an electronic device, which may include: a processor and a memory; wherein the memory stores a computer program adapted to be loaded by the processor and to perform the above-mentioned method steps.

The technical scheme provided by some embodiments of the present specification has the following beneficial effects:

according to the transaction recommendation method provided by the embodiment of the specification, the user data and the transaction to be recommended corresponding to the first user are obtained, the user data comprise the first image features and the first behavior features, then at least one similar user associated with the first user is determined according to the first image features in the user data, the first behavior features are subjected to feature reconstruction based on the second behavior features corresponding to the similar users respectively, the reconstructed behavior features are obtained, finally the conversion rate of the first user for the transaction to be recommended is generated based on the reconstructed behavior features and the transaction to be recommended, and the transaction to be recommended is recommended to the first user based on the conversion rate, so that the problem that the transaction recommendation effect is poor due to the fact that the behavior sparse user has fewer user behavior features can be solved, the similar user is determined according to the first image features, the first behavior features of the first user are subjected to feature reconstruction according to the second behavior features of the similar user, the behavior features of the sparse user are enriched, and the transaction recommendation effect is further improved.

Drawings

In order to more clearly illustrate the embodiments of the present description or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present description, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

Fig. 1 is a flow chart of a transaction recommendation method according to an embodiment of the present disclosure;

fig. 2 is a flow chart of a transaction recommendation method according to an embodiment of the present disclosure;

fig. 3 is a flow chart of a transaction recommendation method according to an embodiment of the present disclosure;

FIG. 4 is a schematic flow chart of a training method for a conversion rate prediction model according to an embodiment of the present disclosure;

FIG. 5 is a schematic structural diagram of a conversion rate prediction model according to an embodiment of the present disclosure;

FIG. 6 is a training flow diagram of model training provided by embodiments of the present disclosure;

fig. 7 is a schematic structural diagram of a transaction recommendation device according to an embodiment of the present disclosure;

fig. 8 is a schematic structural diagram of a transaction recommendation device according to an embodiment of the present disclosure;

Fig. 9 is a schematic structural diagram of a conversion rate prediction model training device according to an embodiment of the present disclosure;

fig. 10 is a block diagram of an electronic device according to an embodiment of the present disclosure.

Detailed Description

The technical solutions of the embodiments of the present specification will be clearly and completely described below with reference to the drawings in the embodiments of the present specification, and it is apparent that the described embodiments are only some embodiments of the present specification, not all embodiments. All other embodiments, which can be made by one of ordinary skill in the art without undue burden from the present disclosure, are intended to be within the scope of the present disclosure.

In the description of the present specification, it should be understood that the terms "first," "second," and the like are used for descriptive purposes only and are not to be construed as indicating or implying relative importance. In the description of the present specification, it should be noted that, unless expressly specified and limited otherwise, "comprise" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, article, or apparatus that comprises a list of steps or elements is not limited to only those listed steps or elements but may include other steps or elements not listed or inherent to such process, method, article, or apparatus. The specific meaning of the terms in this specification will be understood by those of ordinary skill in the art in the light of the specific circumstances. In addition, in the description of the present specification, unless otherwise indicated, "a plurality" means two or more. "and/or", describes an association relationship of an association object, and indicates that there may be three relationships, for example, a and/or B, and may indicate: a exists alone, A and B exist together, and B exists alone. The character "/" generally indicates that the context-dependent object is an "or" relationship.

In addition, in the embodiment of the present specification, for example: the user data acquisition, storage, use, processing and the like involved in the transaction recommendation process all conform to the relevant regulations of national laws and regulations.

In the related art, in the process of recommending the transaction for the user, the effect of recommending the transaction is better for the user with more abundant user behavior information, and the effect of recommending the transaction is inferior to that of the user with less abundant user behavior information for the user with more sparse user behavior information.

Based on this, the present disclosure proposes a transaction recommendation method, by acquiring user data of a first user including a first image feature and a first behavior feature, finding at least one similar user associated with the first user according to the user image feature, wherein the similar user is a user with rich behavior information, then performing feature reconstruction on the first behavior feature of the first user according to a second behavior feature of the similar user to obtain a reconstructed behavior feature, finally generating a conversion rate of the first user for a transaction to be recommended according to the reconstructed behavior feature, recommending the transaction to be recommended to the first user according to the conversion rate, namely, when the first user is a user with sparse behavior information, reconstructing the first behavior feature of the first user according to the second behavior feature of the similar user with rich behavior information by finding the similar user with rich behavior information, so as to realize the enrichment of the behavior feature of the first user, and improve the transaction recommendation effect of the sparse user with respect to the behavior information.

The following is a detailed description of embodiments in connection with the examples of the present specification. The implementations described in the following exemplary examples do not represent all implementations consistent with the present specification. Rather, they are merely examples of apparatus and methods consistent with some aspects of the present description as detailed in the accompanying claims. The flow diagrams depicted in the figures are exemplary only and are not necessarily to be taken in the order shown. For example, some steps are juxtaposed and there is no strict order of logic, so the actual order of execution is variable.

Referring to fig. 1, a flow chart of a transaction recommendation method according to an embodiment of the present disclosure is shown. In the embodiments in the present specification, the transaction recommending method is applied to a transaction recommending apparatus or an electronic device configured with the transaction recommending apparatus. The following details about the flow shown in fig. 1, the transaction recommendation method specifically may include the following steps:

s102, user data and a transaction to be recommended corresponding to a first user are obtained, wherein the user data comprise first image features and first behavior features;

in this embodiment of the present disclosure, when performing a transaction recommendation method to perform transaction recommendation, a transaction recommendation device first obtains user data corresponding to a first user and a transaction to be recommended that is being recommended, where the user data includes a first image feature and a first behavior feature corresponding to the first user.

The first user is a user to be evaluated whether to conduct transaction recommendation or not. The first portrayal feature is a feature extracted from user portrayal information corresponding to the first user. The first behavioral characteristics are characteristics extracted from user behavior information corresponding to the first user.

Alternatively, the user portrayal information may include information related to personal information of the user such as hobbies, work places, work types, personal notes, economical conditions, and the like.

Alternatively, the user behavior information may include historical record information left by the user on the internet, such as a user history browse record, a user purchase record, a user view record, and a user comment record.

In one embodiment, the transaction recommendation device acquires user portrait information and user behavior information corresponding to a first user, and extracts a first portrait feature in the user portrait information and a first behavioral feature in the user behavior information.

S104, determining at least one similar user associated with the first user according to the first image characteristics, and acquiring second behavior characteristics corresponding to the similar users respectively;

in this embodiment of the present disclosure, the transaction recommendation device may determine at least one similar user associated with the first user according to the first image feature of the first user, and obtain second behavior features corresponding to the similar users respectively.

The user portrait of the similar user is similar or similar to the user portrait of the first user.

The second behavioral characteristics are behavioral characteristics extracted from user behavior information of similar users.

In one embodiment, the transaction recommendation device obtains at least one second portrait feature associated with the first portrait feature, the second portrait feature being a portrait feature of a similar user, and determines second behavior features corresponding to the second portrait features respectively.

It can be appreciated that the transaction recommendation device pre-stores the second image features and the second behavior features of the massive behavior-rich user. In the process of recommending the transaction, after the transaction recommending device acquires the first image feature and the first behavior feature of the first user, determining a second image feature similar to the first image feature according to the first image feature of the first user, wherein the user corresponding to the second image feature is the similar user of the first user, and then acquiring the second behavior feature corresponding to the similar user.

S106, carrying out feature reconstruction on the first behavior features based on the second behavior features to obtain reconstructed behavior features;

in the embodiment of the present disclosure, after obtaining the second behavior features of the similar users, feature reconstruction is performed on the first behavior features of the first users according to each second behavior feature, so as to obtain the reconstructed behavior features corresponding to the first users.

It will be appreciated that the second behavioral characteristics are behavioral characteristics of similar users associated with the first user, and that the similar users are behavior information rich users. When the first user is a user with sparse behavior information, the first behavior features corresponding to the user behavior information of the first user are fewer, and at the moment, the feature reconstruction is carried out on the first behavior features according to the second behavior features of the similar user with rich user behavior information, so that the behavior features of the first user are enriched, and the transaction recommendation effect on the first user can be improved.

S108, generating the conversion rate of the first user for the transaction to be recommended based on the reconstructed behavior characteristics and the transaction to be recommended;

in the embodiment of the present disclosure, after obtaining the reconfiguration behavior feature corresponding to the first user, the conversion rate of the first user for the transaction to be recommended is generated according to the reconfiguration behavior feature and the transaction to be recommended.

In one possible implementation, the first user's conversion rate for transactions to be recommended may be predicted based on a pre-trained conversion rate prediction model. The conversion rate prediction model is generated based on learning and training user portrait information and user behavior information of the behavior-rich user. The predicting the conversion rate of the first user for the transaction to be recommended based on the pre-trained conversion rate prediction model may specifically be: the method comprises the steps of obtaining user portraits information, user behavior information and transactions to be recommended of a first user, inputting the user portraits information, the user behavior information and the transactions to be recommended into a pre-trained conversion rate prediction model, extracting first portraits features in the user portraits information, extracting first behavior features in the user behavior information and transaction features corresponding to the transactions to be recommended, and then carrying out feature reconstruction on the first behavior features by a feature reconstruction network in the conversion rate prediction model according to the first portraits features to obtain reconstructed behavior features, and generating the conversion rate of the first user for the transactions to be recommended based on the reconstructed behavior features and the transaction features.

It should be noted that, because the conversion rate prediction model is generated after learning and training the user portrait information and the user behavior information of the behavior-enriched user, the feature reconstruction network fully learns the user portrait information and the user behavior information of the behavior-enriched user and the mapping relation between the user portrait information and the user behavior information of the behavior-enriched user, and after obtaining the first portrait feature of the first user, the feature reconstruction network can reconstruct the first behavioral feature according to the first portrait feature by using the learned user portrait information and the user behavior information of the behavior-enriched user, so as to obtain the reconstructed behavioral feature. The method and the device enrich the behavior characteristics of the first user, and further improve the prediction effect of the first user on the conversion rate of the transaction to be recommended.

S110, recommending the transaction to be recommended to the first user based on the conversion rate.

In the embodiment of the present disclosure, after obtaining the conversion rate of the first user for the transaction to be recommended, recommending the transaction to be recommended to the first user according to the conversion rate of the first user for the transaction to be recommended.

It should be noted that, in the embodiment of the present disclosure, the transaction to be recommended may be any form of recommended transaction such as advertisement recommendation, food recommendation, movie recommendation, travel recommendation, and the like.

The conversion rate refers to the probability of converting the transaction to be recommended for the user. For example, if the transaction to be recommended is a movie recommendation, the conversion rate of the user for the transaction to be recommended may be understood as the probability that the user accepts the movie recommendation and views the movie through the entry corresponding to the recommendation.

In one embodiment, the transaction to be recommended is recommended to the first user when the conversion rate of the first user for the transaction to be recommended is greater than a preset conversion rate threshold.

In one embodiment, a set of users to be recommended corresponding to a transaction to be recommended is obtained, the set of users to be recommended includes a first user and at least one second user, the first user and the at least one second user are ranked according to the conversion rate of the first user and the conversion rate of the second user, a conversion rate ranking sequence is obtained, and the transaction to be recommended is recommended to a preset number of target users before ranking in the conversion rate ranking sequence.

In the embodiment of the specification, the user data comprises the first image feature and the first behavior feature by acquiring the user data corresponding to the first user and the transaction to be recommended, then determining at least one similar user associated with the first user according to the first image feature in the user data, performing feature reconstruction on the first behavior feature based on the second behavior features corresponding to the similar users respectively to obtain the reconstructed behavior feature, generating the conversion rate of the first user for the transaction to be recommended based on the reconstructed behavior feature and the transaction to be recommended, recommending the transaction to be recommended to the first user based on the conversion rate, and solving the problem that the transaction recommendation effect is poor due to the fact that the behavior sparse user has fewer user behavior features.

Referring to fig. 2, for a schematic flow chart of a transaction recommendation method provided in an embodiment of the present disclosure, the transaction recommendation method may include the following steps:

s202, user portrait information and user behavior information corresponding to a first user are obtained;

in the embodiment of the present disclosure, the transaction recommendation device acquires user portrait information and user behavior information corresponding to the first user, so as to determine whether to recommend a transaction to be recommended to the first user according to the user portrait information and the user behavior information. The user portrait information may include information related to personal information of the user such as hobbies, work places, work types, personal records, economic conditions, and the like. The user behavior information may include historical record information left by the user on the internet, such as user history browsing records, user purchase records, user viewing records, user comment records, and the like.

S204, extracting a first image feature in the user portrait information and extracting a first behavior feature in the user behavior information;

in the embodiment of the present disclosure, after obtaining the user portrait information and the user behavior information corresponding to the first user, the first portrait feature in the user portrait information and the first behavior feature in the user behavior information are extracted.

S206, acquiring at least one second portrait characteristic associated with the first portrait characteristic, wherein the second portrait characteristic is a portrait characteristic of a similar user;

s208, determining second behavior features corresponding to the second portrait features respectively;

in this embodiment of the present disclosure, after acquiring a first image feature and a first behavior feature of a first user, the transaction recommendation device determines, according to the first image feature of the first user, a second image feature similar to the first image feature, where a user corresponding to the second image feature is a similar user of the first user, and then acquires a second behavior feature corresponding to the similar user.

Wherein the second behavioral characteristics are behavioral characteristics extracted from user behavioral information of similar users.

S210, carrying out feature reconstruction on the first behavior features based on the second behavior features to obtain reconstructed behavior features;

S212, generating the conversion rate of the first user for the transaction to be recommended based on the reconstruction behavior characteristics, the first behavior characteristics and the transaction to be recommended;

in the embodiment of the present disclosure, after obtaining the reconfiguration behavior feature corresponding to the first user, the conversion rate of the first user for the transaction to be recommended is generated according to the reconfiguration behavior feature, the first behavior feature and the transaction to be recommended.

It can be appreciated that the first behavior feature and the reconstructed behavior feature are both behavior features corresponding to the first user. The first behavior features are behavior features extracted from user behavior information of the first user, the reconstructed behavior features are behavior features reconstructed from second behavior features of similar users according to the first image features of the first user, and the conversion rate of the first user for the transaction to be recommended is generated by combining the first behavior features and the reconstructed behavior features, so that the generation accuracy of the conversion rate can be improved.

In one possible implementation, the first user's conversion rate for transactions to be recommended may be predicted based on a pre-trained conversion rate prediction model. The conversion rate prediction model is generated based on learning and training user portrait information and user behavior information of the behavior-rich user. The predicting the conversion rate of the first user for the transaction to be recommended based on the pre-trained conversion rate prediction model may specifically be: the method comprises the steps of obtaining user portraits information, user behavior information and transactions to be recommended of a first user, inputting the user portraits information, the user behavior information and the transactions to be recommended into a pre-trained conversion rate prediction model, extracting first portraits features in the user portraits information, extracting first behavior features in the user behavior information and transaction features corresponding to the transactions to be recommended, and then carrying out feature reconstruction on the first behavior features by a feature reconstruction network in the conversion rate prediction model according to the first portraits features to obtain reconstructed behavior features, and generating the conversion rate of the first user for the transactions to be recommended based on the reconstructed behavior features, the first behavior features and the transaction feature predictions.

And S214, recommending the transaction to be recommended to the first user based on the conversion rate.

In the embodiment of the present disclosure, step S214 is referred to the detailed description of step S110 in the embodiment of the present application, and will not be described herein.

In a feasible transaction recommendation application scenario, a transaction recommendation device is configured at a cloud end, the cloud end recommends transactions to be recommended to users who are actively online, the cloud end obtains conversion rates of the transactions to be recommended respectively for all the users who are actively online based on the transaction recommendation device, then sequences the users according to the conversion rates respectively corresponding to the users to obtain a conversion rate sequencing sequence, and the cloud end recommends the transactions to be recommended to the users with the front sequencing according to the conversion rate sequencing sequence.

In one embodiment, before acquiring the user data corresponding to the first user and the transaction to be recommended, the method may further include determining a type of the first user. Referring to fig. 3, for a schematic flow chart of a transaction recommendation method provided in the embodiment of the present disclosure, the transaction recommendation method may include the following steps:

S302, if the first user is a behavior rich user, acquiring a first behavior feature corresponding to the first user, and generating the conversion rate of the first user for the transaction to be recommended based on the first behavior feature and the transaction to be recommended;

in the embodiment of the present disclosure, user behavior information corresponding to a first user is obtained, whether the first user is a behavior-rich user is determined according to the user behavior information, if the first user is a user with rich user behavior information, a first behavior feature is extracted from the user behavior information of the first user, and a conversion rate of the first user for a transaction to be recommended is generated according to the first behavior feature and the transaction to be recommended.

S304, if the first user is a sparse behavior user, acquiring user data corresponding to the first user and a transaction to be recommended, wherein the user data comprises a first image feature and a first behavior feature;

in the embodiment of the present disclosure, user behavior information corresponding to a first user is obtained, whether the first user is a behavior-rich user is determined according to the user behavior information, if the first user is a user with sparse user behavior information, a step of obtaining user data corresponding to the first user and a transaction to be recommended is performed, rich reconstruction of a first behavior feature of the first user is implemented based on subsequent steps S306-S312, and a conversion rate of the first user for the transaction to be recommended is generated according to the reconstructed behavior feature after the feature reconstruction and the transaction to be recommended.

Wherein the user behavior information of the behavior sparse user is far less than the user behavior information of the behavior rich user.

S306, determining at least one similar user associated with the first user according to the first image characteristics, and acquiring second behavior characteristics corresponding to the similar users respectively;

S308, carrying out feature reconstruction on the first behavior features based on the second behavior features to obtain reconstructed behavior features;

S310, generating the conversion rate of the first user for the transaction to be recommended based on the reconstructed behavior characteristics and the transaction to be recommended;

And S312, recommending the transaction to be recommended to the first user based on the conversion rate.

In one embodiment, a set of users to be recommended corresponding to a transaction to be recommended is obtained, the set of users to be recommended includes a first user and at least one second user, the first user and the at least one second user are ranked according to the conversion rate of the first user and the conversion rate of the second user, a conversion rate ranking sequence is obtained, and the transaction to be recommended is recommended to a preset number of target users before ranking in the conversion rate ranking sequence. .

In the embodiment of the present disclosure, if the first user is a behavior enriching user with rich behavior information, the feature reconstruction is not required to be performed on the first behavior feature of the first user, the conversion rate of the first user for the transaction to be recommended is directly generated based on the first behavior feature of the first user, if the first user is a behavior sparse user with sparse behavior information, the problem that the transaction recommendation effect of the sparse behavior user due to the less behavior feature of the user is not good is solved by acquiring the user data corresponding to the first user and the transaction to be recommended, the user data includes the first image feature and the first behavior feature, then at least one similar user associated with the first user is determined according to the first image feature in the user data, the feature reconstruction is performed on the first behavior feature based on the second behavior feature corresponding to each similar user, the reconstructed behavior feature is obtained, finally the conversion rate of the first user for the transaction to be recommended is generated based on the reconstructed behavior feature and the transaction to be recommended, the problem that the transaction to be recommended is not good due to the less behavior feature of the user is solved, the sparse behavior feature of the similar user is determined according to the first image feature, the first behavior feature of the similar user is reconstructed, the sparse behavior feature is the first behavior feature of the similar user is improved, and the effect of the transaction recommendation is further improved.

Referring to fig. 4, a flow chart of a method for training a conversion rate prediction model according to an embodiment of the present disclosure is provided. In the embodiments in the present specification, the conversion rate prediction model training method is applied to a conversion rate prediction model training apparatus or an electronic device configured with the conversion rate prediction model training apparatus. The process shown in fig. 4 will be described in detail, and the method for training the conversion prediction model specifically may include the following steps:

s402, constructing a training sample data set, wherein the training sample data comprises sample user portrait information, sample user behavior information, sample recommended transactions and sample conversion rate label values;

in this illustrative embodiment, the training sample data set includes a plurality of training sample data including sample user profile information, sample user behavior information, sample recommendation transactions, and sample conversion label values. The sample users are behavior rich users with rich behavior information.

S404, inputting training sample data into a conversion rate prediction model, and extracting sample user portrait features in sample user portrait information, sample user behavioral features in sample user behavioral information and sample transaction features in sample recommended transactions based on a feature extraction network;

In the embodiment of the present disclosure, the conversion rate prediction model may include a feature extraction network, a feature reconstruction network, and a conversion rate prediction network, please refer to fig. 5, which is a schematic structural diagram of a conversion rate prediction model provided in the embodiment of the present disclosure. In the model training process, training sample data are input into a conversion rate prediction model, and sample user portrait features in sample user portrait information, sample user behavioral features in sample user behavioral information and sample transaction features in sample recommended transactions are extracted based on a feature extraction network.

S406, generating sample reconstruction behavior characteristics corresponding to training sample data according to the sample user portrait characteristics by adopting a characteristic reconstruction network;

in the embodiment of the present disclosure, after obtaining the sample user image feature corresponding to the training sample data, the sample user image feature is input into the feature reconstruction network, and the feature reconstruction network generates the sample reconstruction behavior feature corresponding to the sample user according to the sample user image feature prediction.

It can be understood that the feature reconstruction network generates a sample reconstruction behavior feature corresponding to the sample user according to the sample user image feature prediction, and when the model parameter is updated, a reconstruction loss value can be constructed according to the sample reconstruction behavior feature and the sample user behavior feature, and the model parameter is updated according to the reconstruction loss value, so that the sample reconstruction behavior feature predicted by the feature reconstruction network tends to the sample user behavior feature. After model training is completed, the reconstructed feature network can fully learn the mapping relation between the portrait information of each sample user and the behavior information of each sample user in the training sample data set, and can reconstruct the behavior features of the behavior sparse users aiming at the behavior sparse users in model application and realize the effect of enriching the behavior features.

S408, recommending and predicting according to the sample reconstruction behavior characteristics and the sample transaction characteristics by adopting a conversion rate prediction network to obtain a sample conversion rate prediction value corresponding to training sample data;

in the embodiment of the present disclosure, the conversion rate prediction network performs recommendation prediction according to the sample reconstruction behavior feature and the sample transaction feature corresponding to the sample recommendation transaction, so as to obtain a sample conversion rate prediction value of the sample user for the sample recommendation transaction.

And S410, performing supervision training on the conversion rate prediction model based on a preset loss function, sample user behavior characteristics, sample reconstruction behavior characteristics, a conversion rate label value and a sample conversion rate prediction value, and iteratively updating model parameters of the conversion rate prediction model until the conversion rate prediction model converges, so as to obtain the conversion rate prediction model after training.

In the embodiment of the specification, a model loss value is obtained by calculating a preset loss function according to the sample user behavior characteristic, the sample reconstruction behavior characteristic, the conversion rate tag value and the sample conversion rate predicted value, and model parameters of the conversion rate predicted model are updated according to the model loss value. And carrying out iterative training on the conversion rate prediction model according to each training sample data in the training sample data set until the model loss value is smaller than a preset threshold value, determining that the conversion rate prediction model converges, and obtaining the conversion rate prediction model after training.

In one embodiment, the preset loss function includes a reconstruction loss function and a prediction loss function, the reconstruction loss function is adopted to calculate a reconstruction loss value according to a sample user behavior feature and a sample reconstruction behavior feature, the prediction loss function is adopted to calculate a prediction loss value according to a conversion rate label value and a sample conversion rate prediction value, model parameters of the conversion rate prediction model are updated based on the reconstruction loss value and the prediction loss value, then whether the conversion rate prediction model updated by the parameters meets a preset convergence condition is judged, if yes, training is stopped, a conversion rate prediction model with completed training is obtained, if not, a step of inputting training sample data into the conversion rate prediction model is executed, and iterative training is performed based on training sample data in the training sample data set.

Referring to fig. 6, a training flowchart of model training is provided in an embodiment of the present disclosure. As shown in fig. 6, training sample data is input into a feature extraction network, the training sample data comprises sample user portrait information, sample user behavior information, sample recommended transactions and sample conversion rate tag values, the feature extraction network respectively extracts sample user portrait features in the sample user portrait information, sample user behavior features in the sample user behavior information and sample transaction features in the sample recommended transactions, the sample portrait features are input into a feature reconstruction network to obtain sample reconstruction behavior features output by the feature reconstruction network, and finally a conversion rate prediction network predicts sample conversion rate prediction values of sample users for the sample recommended transactions according to the sample reconstruction behavior features and the sample transaction features. In addition, the sample user behavior feature is used to calculate a reconstruction loss value from the sample reconstruction feature, and the sample conversion tag value is used to calculate a prediction loss value from the sample conversion prediction value.

In one embodiment, weight coefficients corresponding to the reconstructed loss value and the predicted loss value may be preset, an overall loss value corresponding to the reconstructed loss value and the predicted loss value is calculated according to the preset weight coefficients, and model parameters of the conversion rate prediction model are updated based on the overall loss value.

In the embodiment of the specification, the training sample data set corresponding to the behavior enrichment user is constructed, the conversion rate prediction model is trained according to the training sample data, and the user portrait information, the user behavior information and the mapping relation of the user portrait information and the user behavior information of the behavior enrichment user are fully learned, so that the trained conversion rate prediction model can conduct behavior enrichment reconstruction on behavior characteristics of the behavior sparse user, and the conversion rate prediction effect on the behavior sparse user is improved.

Fig. 7 is a schematic structural diagram of a transaction recommendation device according to an embodiment of the present disclosure. As shown in fig. 7, the transaction recommendation device 1 may be implemented as all or a part of an electronic apparatus by software, hardware or a combination of both. According to some embodiments, the transaction recommendation device 1 includes a user data acquisition module 11, a similar feature determination module 12, a behavior feature reconstruction module 13, a conversion rate prediction module 14, and a transaction recommendation module 15, and specifically includes:

The user data obtaining module 11 is configured to obtain user data corresponding to a first user and a transaction to be recommended, where the user data includes a first image feature and a first behavior feature;

a similar feature determining module 12, configured to determine at least one similar user associated with the first user according to the first portrait feature, and obtain second behavioral features corresponding to the similar users respectively;

a behavior feature reconstruction module 13, configured to perform feature reconstruction on the first behavior feature based on each of the second behavior features, to obtain a reconstructed behavior feature;

a conversion rate prediction module 14, configured to generate a conversion rate of the first user for the transaction to be recommended based on the reconstructed behavior feature and the transaction to be recommended;

and the transaction recommending module 15 is used for recommending the transaction to be recommended to the first user based on the conversion rate.

Optionally, the user data acquisition module 11 is specifically configured to:

acquiring user portrait information and user behavior information corresponding to the first user;

extracting a first image feature in the user portrait information and extracting a first behavioral feature in the user behavioral information.

Optionally, the similar feature determining module 12 is specifically configured to:

Acquiring at least one second portrait feature associated with the first portrait feature, the second portrait feature being a portrait feature of the similar user;

and determining second behavior characteristics corresponding to the second portrait characteristics respectively.

Optionally, the conversion prediction module 14 is specifically configured to:

and generating the conversion rate of the first user for the transaction to be recommended based on the reconstruction behavior characteristic, the first behavior characteristic and the transaction to be recommended.

Optionally, please refer to fig. 8, which is a schematic structural diagram of a transaction recommendation device according to an embodiment of the present disclosure. As shown in fig. 8, the transaction recommendation device further includes a user type determining module 16, specifically configured to:

if the first user is a behavior rich user, acquiring a first behavior characteristic corresponding to the first user, and generating the conversion rate of the first user for the transaction to be recommended based on the first behavior characteristic and the transaction to be recommended;

and if the first user is a sparse behavior user, executing the step of acquiring user data corresponding to the first user and the transaction to be recommended, wherein the sparse behavior user has less user behavior information than the rich behavior user.

Optionally, the transaction recommendation module 15 is specifically configured to:

and recommending the transaction to be recommended to the first user when the conversion rate is greater than a preset conversion rate threshold value.

Optionally, the transaction recommendation module 15 is further configured to:

acquiring a user set to be recommended corresponding to the transaction to be recommended, wherein the user set to be recommended comprises the first user and at least one second user;

sequencing the first user and the at least one second user according to the conversion rate of the first user and the conversion rate of the second user to obtain a conversion rate sequencing sequence;

recommending the transaction to be recommended to target users with preset quantity before sequencing in the conversion rate sequencing sequence.

It should be noted that, when the transaction recommendation device provided in the foregoing embodiment performs the transaction recommendation method, only the division of the foregoing functional modules is used as an example, and in practical application, the foregoing functional allocation may be performed by different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the transaction recommendation device and the transaction recommendation method embodiment provided in the foregoing embodiments belong to the same concept, which embody the detailed implementation process in the method embodiment, and are not described herein again.

The foregoing embodiment numbers of the present specification are merely for description, and do not represent advantages or disadvantages of the embodiments.

Fig. 9 is a schematic structural diagram of a training device for a conversion rate prediction model according to an embodiment of the present disclosure. As shown in fig. 9, the conversion rate prediction model training device 2 may be implemented as all or a part of the electronic apparatus by software, hardware, or a combination of both. According to some embodiments, the conversion rate prediction model training device 2 includes a sample data acquisition module 21, a sample feature extraction module 22, a reconstruction feature generation module 23, a conversion rate prediction module 24, and a model supervision training module 25, and specifically includes:

A sample data acquisition module 21, configured to construct a training sample data set, where the training sample data includes sample user profile information, sample user behavior information, sample recommendation transaction, and sample conversion rate tag values;

a sample feature extraction module 22, configured to input the training sample data into a conversion rate prediction model, extract sample user portrait features in the sample user portrait information, extract sample user behavioral features in the sample user behavioral information, and extract sample transaction features in the sample recommended transaction based on the feature extraction network;

a reconstruction feature generation module 23, configured to generate a sample reconstruction behavior feature corresponding to the training sample data according to the sample user portrait feature by using a feature reconstruction network;

the conversion rate prediction module 24 is configured to perform recommended prediction according to the sample reconstruction behavior feature and the sample transaction feature by using a conversion rate prediction network, so as to obtain a sample conversion rate predicted value corresponding to the training sample data;

the model supervision and training module 25 is configured to perform supervision and training on the conversion rate prediction model based on a preset loss function, the sample user behavior feature, the sample reconstruction behavior feature, the conversion rate tag value and the sample conversion rate predicted value, and iteratively update model parameters of the conversion rate prediction model until the conversion rate prediction model converges, so as to obtain a trained conversion rate prediction model.

Optionally, the preset loss function includes a reconstructed loss function and a predicted loss function, and the model supervision training module 25 is specifically configured to:

calculating a reconstruction loss value by adopting the reconstruction loss function according to the sample user behavior characteristics and the sample reconstruction behavior characteristics;

calculating a predicted loss value according to the conversion rate tag value and the sample conversion rate predicted value by adopting a predicted loss function;

updating model parameters of the conversion rate prediction model based on the reconstruction loss value and the prediction loss value;

judging whether the conversion rate prediction model updated by the parameters meets a preset convergence condition, if so, stopping training to obtain a conversion rate prediction model after training, and if not, executing the step of inputting the training sample data into the conversion rate prediction model.

Optionally, the model supervision training module 25 is specifically configured to, when executing the updating of the model parameters of the conversion rate prediction model based on the reconstructed loss value and the predicted loss value:

calculating the overall loss value corresponding to the reconstruction loss value and the predicted loss value according to a preset weight coefficient;

model parameters of the conversion rate prediction model are updated based on the overall loss value.

It should be noted that, when the conversion rate prediction model training apparatus provided in the foregoing embodiment executes the transaction recommendation method, only the division of the foregoing functional modules is used as an example, in practical application, the foregoing functional allocation may be completed by different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules, so as to complete all or part of the functions described above. In addition, the conversion rate prediction model training device and the conversion rate prediction model training method provided in the above embodiments belong to the same concept, which embody detailed implementation procedures and are not described herein.

The embodiment of the present disclosure further provides a computer storage medium, where the computer storage medium may store a plurality of instructions, where the instructions are adapted to be loaded by a processor and executed by the processor to perform the transaction recommendation method according to the embodiment shown in fig. 1 to fig. 6, and the specific execution process may refer to the specific description of the embodiment shown in fig. 1 to fig. 6, which is not repeated herein.

The present disclosure further provides a computer program product, where at least one instruction is stored, where the at least one instruction is loaded by the processor and executed by the processor to perform the transaction recommendation method according to the embodiment shown in fig. 1 to fig. 6, and the specific execution process may refer to the specific description of the embodiment shown in fig. 1 to fig. 6, which is not repeated herein.

Referring to fig. 10, a block diagram of an electronic device according to an embodiment of the present disclosure is provided. The electronic device in this specification may include one or more of the following: processor 110, memory 120, input device 130, output device 140, and bus 150. The processor 110, the memory 120, the input device 130, and the output device 140 may be connected by a bus 150.

Processor 110 may include one or more processing cores. The processor 110 connects various parts within the overall terminal using various interfaces and lines, performs various functions of the terminal 100 and processes data by executing or executing instructions, programs, code sets, or instruction sets stored in the memory 120, and invoking data stored in the memory 120. Alternatively, the processor 110 may be implemented in at least one hardware form of digital signal processing (digital signal processing, DSP), field-programmable gate array (field-programmable gate array, FPGA), programmable logic array (programmable logic Array, PLA). The processor 110 may integrate one or a combination of several of a central processor (central processing unit, CPU), an image processor (graphics processing unit, GPU), and a modem, etc. The CPU mainly processes an operating system, a user interface, an application program and the like; the GPU is used for being responsible for rendering and drawing of display content; the modem is used to handle wireless communications. It will be appreciated that the modem may not be integrated into the processor 110 and may be implemented solely by a single communication chip.

The memory 120 may include a random access memory (random Access Memory, RAM) or a read-only memory (ROM). Optionally, the memory 120 includes a non-transitory computer readable medium (non-transitory computer-readable storage medium). Memory 120 may be used to store instructions, programs, code, sets of codes, or sets of instructions.

The input device 130 is configured to receive input instructions or data, and the input device 130 includes, but is not limited to, a keyboard, a mouse, a camera, a microphone, or a touch device. The output device 140 is used to output instructions or data, and the output device 140 includes, but is not limited to, a display device, a speaker, and the like. In the embodiment of the present disclosure, the input device 130 may be a temperature sensor for acquiring an operation temperature of the terminal. The output device 140 may be a speaker for outputting audio signals.

In addition, those skilled in the art will appreciate that the configuration of the terminal illustrated in the above-described figures does not constitute a limitation of the terminal, and the terminal may include more or less components than illustrated, or may combine certain components, or may have a different arrangement of components. For example, the terminal further includes components such as a radio frequency circuit, an input unit, a sensor, an audio circuit, a wireless fidelity (wireless fidelity, WIFI) module, a power supply, a bluetooth module, and the like, which are not described herein again.

In the embodiment of the present specification, the execution subject of each step may be the terminal described above. Optionally, the execution subject of each step is an operating system of the terminal. The operating system may be an android system, an IOS system, or other operating systems, which embodiments of the present specification are not limited to.

In the electronic device of fig. 10, the processor 110 may be configured to invoke the transaction recommendation program stored in the memory 120 and execute to implement the transaction recommendation method as described in various method embodiments of the present description.

It will be clear to a person skilled in the art that the solution according to the present description can be implemented by means of software and/or hardware. "Unit" and "module" in this specification refer to software and/or hardware capable of performing a specific function, either alone or in combination with other components, such as Field programmable gate arrays (Field-Programmable Gate Array, FPGAs), integrated circuits (Integrated Circuit, ICs), etc.

It should be noted that, for simplicity of description, the foregoing method embodiments are all expressed as a series of action combinations, but it should be understood by those skilled in the art that the present description is not limited by the order of actions described, as some steps may be performed in other order or simultaneously in accordance with the present description. Further, those skilled in the art will also appreciate that the embodiments described in the specification are all preferred embodiments, and that the acts and modules referred to are not necessarily required in the specification.

In the foregoing embodiments, the descriptions of the embodiments are emphasized, and for parts of one embodiment that are not described in detail, reference may be made to related descriptions of other embodiments.

In the several embodiments provided in this specification, it should be understood that the disclosed apparatus may be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative, such as the division of the units, merely a logical function division, and there may be additional manners of dividing the actual implementation, such as multiple units or components may be combined or integrated into another system, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be through some service interface, device or unit indirect coupling or communication connection, electrical or otherwise.

The units described as separate units may or may not be physically separate, and units shown 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 may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

In addition, each functional unit in each embodiment of the present specification may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in software functional units.

Those of ordinary skill in the art will appreciate that all or a portion of the steps in the various methods of the above embodiments may be performed by hardware associated with a program that is stored in a computer readable memory, which may include: flash disk, read-Only Memory (ROM), random-access Memory (Random Access Memory, RAM), magnetic or optical disk, and the like.

The foregoing is merely exemplary embodiments of the present specification and is not intended to limit the scope of the present specification. It is intended that all equivalent variations and modifications as taught herein fall within the scope of the present disclosure. Other embodiments of the present disclosure will be apparent to those skilled in the art from consideration of the specification and practice of the disclosure herein. This specification is intended to cover any variations, uses, or adaptations of the specification following, in general, the principles of the specification and including such departures from the present disclosure as come within known or customary practice within the art to which the specification pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the specification being indicated by the following claims.

Claims

1. A transaction recommendation method, the method comprising:

2. The method of claim 1, the obtaining user data corresponding to the first user, comprising:

3. The method of claim 1, wherein the determining at least one similar user associated with the first user according to the first portrait feature, and obtaining second behavioral features respectively corresponding to the similar users, includes:

4. The method of claim 1, the generating a conversion rate of the first user for the transaction to be recommended based on the reconstructed behavioral characteristics and the transaction to be recommended, comprising:

5. The method of claim 1, the method further comprising:

6. The method of claim 1, the recommending the transaction to be recommended to the first user based on the conversion rate, comprising:

7. The method of claim 1, the recommending the transaction to be recommended to the first user based on the conversion rate, comprising:

8. A conversion rate prediction model training method, the conversion rate prediction model comprising a feature extraction network, a feature reconstruction network, and a conversion rate prediction network, the method comprising:

9. The method of claim 8, the preset loss function comprising a reconstruction loss function and a predictive loss function, the performing supervised training of the conversion prediction model based on the preset loss function, the sample user behavior feature, the sample reconstruction behavior feature, the conversion label value, and the sample conversion prediction value and iteratively updating model parameters of the conversion prediction model until the conversion prediction model converges, resulting in a trained conversion prediction model, comprising:

10. The method of claim 9, the updating model parameters of the conversion rate prediction model based on the reconstruction loss value and the prediction loss value, comprising:

11. A transaction recommendation device, comprising:

12. A conversion rate prediction model training device, comprising:

13. A storage medium having stored thereon a computer program, which when executed by a processor performs the steps of the method according to any of claims 1-7 or 8-10.

14. An electronic device, comprising: a processor and a memory; wherein the memory stores a computer program adapted to be loaded by the processor and to perform the steps of the method according to any of claims 1-7 or 8-10.

15. A computer program product having stored thereon at least one instruction which when executed by a processor implements the steps of the method of any of claims 1 to 7 or 8 to 10.