CN118229337A

CN118229337A - Training method of user behavior prediction model, user behavior prediction method and device

Info

Publication number: CN118229337A
Application number: CN202410231939.2A
Authority: CN
Inventors: 纪承; 王龙滔; 蔡振宇; 凌梓南; 吴榕; 刘柯可; 陈贻珈
Original assignee: Beijing Sohu New Power Information Technology Co ltd
Current assignee: Beijing Sohu New Power Information Technology Co ltd
Filing date: 2024-02-29
Publication date: 2024-06-21

Abstract

The disclosure provides a user behavior prediction model training method, a user behavior prediction method and a user behavior prediction device. The training method of the user behavior prediction model comprises the steps of obtaining user behavior data and obtaining a first characteristic of a user behavior prediction task according to the user behavior data; acquiring a second characteristic of the user behavior prediction task according to the first characteristic; and according to the second characteristic, carrying out joint training on the user behavior prediction task so as to obtain a multi-task model of user behavior prediction. The method and the device can greatly improve the prediction precision, the prediction efficiency and the prediction stability, and improve the user experience.

Description

Training method of user behavior prediction model, user behavior prediction method and device

Technical Field

The present disclosure relates to the field of artificial intelligence, and in particular, to a user behavior prediction model training method, a user behavior prediction method, a device, equipment, and a readable storage medium based on multitasking.

Background

Recommendation systems and advertising systems play a vital role in the fields of electronic commerce, social networks, search engines, etc., helping enterprises to achieve marketing and sales goals to a great extent, and have evolved tremendously over the past few years.

Currently, the primary goal of recommendation systems and advertising systems is to predict the behavior of a user and present the user with appropriate advertisements or merchandise. CVR (Conversion Rate) is an important indicator for measuring advertisement effectiveness, and CTR (Click-Through Rate) refers to the Rate at which a user completes a specific target action after clicking, such as submitting a form or purchasing a product. The CVR can reflect the marketing effect of the advertisement, helps advertisers to know the attraction and conversion effect of the advertisement to the target user, and is a key factor for predicting the CVR. Conventional CVR modeling methods typically employ deep learning methods, including DNN (Deep Neural Network ), FM (Factorization Machine, factorizer), and the like. These methods have some problems, namely, all samples are predicted by training and clicking the samples, so that the problem of sample selection deviation is caused, and the information in the whole sample space cannot be well utilized. In addition, the traditional method also has the problem of data sparsity, so that the fitting capability of the model becomes difficult. These problems directly affect the accuracy of the CVR predictions and thus affect the effectiveness and efficiency of the delivery of the recommendation system and the advertising system.

Accordingly, there is a need for better CVR modeling methods to address these issues, and for more accurate and reliable CVR prediction methods to improve the effectiveness and efficiency of delivery of recommendation systems and advertising systems.

Disclosure of Invention

The present disclosure has been made to solve the above-mentioned problems, and an object thereof is to provide a more accurate and reliable training method, a user behavior prediction method, a device, an apparatus, and a readable storage medium for a user behavior prediction model based on multitasking.

The present disclosure provides this summary section to introduce concepts in a simplified form that are further described below in the detailed description section. This summary is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used to limit the scope of the claimed subject matter.

In order to solve the above technical problems, an embodiment of the present disclosure provides a user behavior prediction model training method based on multitasking, which adopts the following technical scheme, including:

acquiring user behavior data, and acquiring a first characteristic of a user behavior prediction task according to the user behavior data;

Acquiring a second characteristic of the user behavior prediction task according to the first characteristic;

and according to the second characteristic, carrying out joint training on the user behavior prediction task so as to obtain a multi-task model of user behavior prediction.

In order to solve the above technical problems, an embodiment of the present disclosure provides a user behavior prediction method based on multitasking, which adopts the following technical solutions, including:

and predicting the user behavior by using the multi-task model which is trained as above, and obtaining a prediction result.

In order to solve the above technical problems, an embodiment of the present disclosure further provides a user behavior prediction model training device based on multitasking, which adopts the following technical scheme, including:

The first acquisition unit is used for acquiring user behavior data and acquiring a first characteristic of a user behavior prediction task according to the user behavior data;

the second obtaining unit is used for obtaining a second characteristic of the user behavior prediction task according to the first characteristic;

and the training unit is used for carrying out joint training on the user behavior prediction tasks according to the second characteristics so as to obtain a multi-task model of user behavior prediction.

In order to solve the above technical problems, an embodiment of the present disclosure provides a user behavior prediction apparatus based on multitasking, which adopts the following technical scheme, including:

The user behavior prediction model training device as described above;

And the prediction unit is used for predicting the user behavior prediction task according to the multitask model of the user behavior prediction and obtaining a prediction result.

In order to solve the above technical problems, an embodiment of the present disclosure further provides a computer device, which adopts the following technical solutions, including: a memory and a processor, the memory having stored therein a computer program, the processor implementing the method of any of the preceding claims when the computer program is executed.

In order to solve the above technical problems, an embodiment of the present disclosure further provides a computer readable storage medium, which adopts the following technical solutions, including: the computer readable storage medium has stored thereon a computer program which, when executed by a processor, implements a method as claimed in any of the preceding claims.

According to the technical scheme disclosed by the disclosure, compared with the prior art, the problems of sample selection deviation and data sparsity can be effectively solved, the prediction precision, the prediction efficiency and the stability are greatly improved, and the user experience is improved.

Drawings

FIG. 1 is a schematic diagram of one embodiment of a system architecture according to the present disclosure;

FIG. 2 is a flow chart of one embodiment of a multitasking based user behavior prediction model training method according to the present disclosure;

FIG. 3 is a schematic diagram of one embodiment of a multitasking based user behavior prediction model training apparatus according to the present disclosure;

FIG. 4 is a schematic diagram of one embodiment of a multitasking based user behavior prediction apparatus according to the present disclosure;

Fig. 5 is a schematic diagram of one embodiment of a computer device according to the present disclosure.

The above and other features, advantages, and aspects of embodiments of the present disclosure will become more apparent by reference to the following detailed description when taken in conjunction with the accompanying drawings. The same or similar reference numbers will be used throughout the drawings to refer to the same or like elements. It should be understood that the figures are schematic and that elements and components are not necessarily drawn to scale.

Detailed Description

Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure belongs; the terminology used in the description of the applications herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure; the terms "comprising" and "having" and any variations thereof in the description and claims of the present disclosure and in the description of the figures above are intended to cover a non-exclusive inclusion. The terms first, second and the like in the description and in the claims or in the above-described figures, are used for distinguishing between different objects and not necessarily for describing a sequential or chronological order.

Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment may be included in at least one embodiment of the present disclosure. The appearances of such phrases in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. Those of skill in the art will explicitly and implicitly appreciate that the embodiments described herein may be combined with other embodiments.

In order to better understand the present disclosure, a technical solution in the embodiments of the present disclosure will be clearly and completely described below with reference to the accompanying drawings.

[ System Structure ]

First, a structure of a system of one embodiment of the present disclosure is explained. As shown in fig. 1, the system architecture 100 may include terminal devices 101, 102, 103, 104, a network 105, and a server 106. The network 105 serves as a medium for providing communication links between the terminal devices 101, 102, 103, 104 and the server 106.

In this embodiment, an electronic device (for example, the terminal device 101, 102, 103, or 104 shown in fig. 1) on which the method operates may perform transmission of various information through the network 105. The network 105 may include various connection types, such as wired, wireless communication links, or fiber optic cables, among others. It should be noted that the wireless connection may include, but is not limited to, 3G/4G/5G connections, wi-Fi connections, bluetooth connections, wiMAX connections, zigbee connections, UWB connections, local area networks ("LANs"), wide area networks ("WANs"), the internet (e.g., the internet), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as other now known or later developed network connections. The network 105 may communicate using any currently known or future developed network protocol, such as HTTP (Hyper Text Transfer Protocol ), and may be interconnected with digital data communications (e.g., communication networks) in any form or medium.

The user may interact with the server 106 via the network 105 using the terminal devices 101, 102, 103, 104 to receive or send messages or the like. Various client applications, such as a video live and play class application, a web browser application, a shopping class application, a search class application, an instant messaging tool, a mailbox client, social platform software, etc., may be installed on the terminal device 101, 102, 103, or 104.

The terminal device 101, 102, 103 or 104 may be various electronic devices having a touch display screen and/or supporting web browsing, including, but not limited to, a smart phone, a tablet computer, an electronic book reader, an MP3 (moving picture experts compression standard audio layer 3) player, an MP4 (moving picture experts compression standard audio layer 4) player, a head mounted display device, a notebook computer, a digital broadcast receiver, a PDA (personal digital assistant), a PMP (portable multimedia player), a mobile terminal such as a car navigation terminal, etc., a digital TV, a desktop computer, etc.

The server 106 may be a server providing various services, such as a background server providing support for pages displayed or data transmitted on the terminal device 101, 102, 103, or 104.

It should be understood that the number of terminal devices, networks and servers in fig. 1 is merely illustrative. There may be any number of terminal devices, networks, and servers, as desired for implementation.

Here, the terminal device may implement the embodiment method of the present disclosure independently or by running applications in various operating systems in cooperation with other electronic terminal devices, and may also implement the embodiment method of the present disclosure by running applications in other operating systems.

[ User behavior prediction model training method based on multitasking ]

First, to illustrate the technical concept of the present disclosure, the present disclosure proposes a new CTCVR (Click-Through & Conversion Rate) modeling method aiming at the defects existing in the prior art, CTCVR refers to the ratio of the target behavior of the user after exposure by clicking the advertisement, and is the product of the Click Rate after exposure and the Conversion Rate after clicking. CTCVR comprehensively considers the influences of click rate and conversion rate, and can reflect the marketing effect of the advertisement more accurately. The limitations of the traditional method are solved and the prediction precision of CTR and CVR is improved by using a xDeepFM (Extreme Deep Factorization Machine, ultra-deep factorizer) algorithm-enhanced ESMM (ENTIRE SPACE Multi-Task Model, full-sample space Multi-Task Model) Model. Specifically, the present disclosure addresses the problem of bias selection and data sparsity of conventional CVR model training samples, employing sequential patterns and feature representation transfer learning strategies to directly perform CVR modeling over full sample space. Meanwhile, the characteristic expression capacity of the ESMM model is enhanced by introducing the xDeepFM model, the prediction precision of CTR and CVR is improved, and the effect and efficiency of putting of a recommendation system and an advertisement system are further optimized. Here, the xDeepFM model is used for learning the higher-order interaction between features, so as to improve the expression capability of the model, and of course, other algorithm models, such as a Wide & Deep, deep fm, etc., may be used, which also has similar functions, and is not limited thereto.

Specifically, the present disclosure solves the sample selection bias problem existing in the conventional method by fully utilizing the sequential mode (exposure-click-conversion) of the user behavior based on ESMM model, and simultaneously adopts the feature representation transfer learning strategy to directly perform CVR modeling on the whole sample space.

The ESMM model of the present disclosure is built in the framework of multitasking learning, where CTR and CVR can be predicted simultaneously. In the ESMM model, the CTR and CVR tasks improve modeling through a shared feature embedding layer. The ESMM model can avoid the sample selection bias problem by directly performing CVR modeling over the full sample space. In addition, the ESMM model also adopts a feature transfer learning strategy, so that the influence of the data sparsity problem can be reduced. Here, the multitasking learning may be implemented by a MMoE (Multi-gate Mixture-of-expertise) algorithm model or the like, and is not limited thereto.

In order to further and effectively express the intersection of high-order features, the DNN layer in the ESMM model is replaced by xDeepFM, so that modeling capacity of the model on the feature intersection relationship is enhanced, high-order interaction relationships of vector levels among features are processed, expression capacity of the model is improved, prediction precision of CTR and CVR is further improved, and the problem of data sparsity is further solved.

Finally, the experimental result of the present disclosure on the data set shows that the ESMM model reinforced by xDeepFM is significantly better than the competition method, and can effectively improve the advertisement recommendation effect and efficiency.

Referring to fig. 2, a flow chart of one embodiment of a multitasking based user behavior prediction model training method according to the present disclosure is shown.

The user behavior prediction model training method based on the multitasking comprises the following steps:

S21, acquiring user behavior data, and acquiring a first characteristic of a user behavior prediction task according to the user behavior data.

In one or more embodiments, obtaining user behavior data includes, for example: acquiring at least one data set corresponding to a user behavior prediction task as original data, wherein the original data may comprise user behavior data, commodity information, advertisement information and the like; and carrying out data preprocessing on the data set to acquire user behavior data.

In one or more embodiments, data preprocessing primarily includes operations such as data cleansing, missing value processing, outlier processing, and normalization to ensure that the model is able to receive high quality data. Here, this step may not be performed during each data processing, but may be performed before model training, and is not limited thereto. The specific operation is as follows:

(1) Data cleaning: extraneous, redundant, or erroneous data is removed to ensure the quality of the data.

(2) Missing value processing: the missing values in the data are processed by interpolation, deletion or prediction and other methods so as to avoid calculation errors or model training deviation caused by the missing values.

(3) Outlier processing: through the box line graph, 3The abnormal value is detected by methods such as principle and the like, and the abnormal value is processed by methods such as deletion or correction and the like so as to improve the stability of the data.

(4) Standardization: and the data is subjected to standardized processing, so that the features with different scales have the same dimension, and model training is facilitated.

In one or more embodiments, after data preprocessing, feature extraction is performed according to user behavior data to obtain first features of a user behavior prediction task, where feature extraction mainly includes extracting useful features from the preprocessed data, i.e., obtaining the first features, and converting the first features into an input format acceptable to a model.

For example, the method comprises the steps of performing feature engineering on user behavior data to extract target features of a user behavior prediction task; and acquiring a first characteristic of the user behavior prediction task according to the target characteristic. The specific operation is as follows:

(1) Characteristic engineering: useful features such as statistical features, temporal features, etc. are extracted from the raw data, constructed.

(2) Feature coding: and performing processes such as single-heat coding, label coding and the like on the category characteristics so that the category characteristics can be used as input of a model.

(3) Feature selection and feature dimension reduction: features with the highest relevance to CTR and CVR predictions are screened out through methods such as relevance analysis and Principal Component Analysis (PCA), and dimension reduction processing is carried out to reduce the computational complexity of the model, so that the prediction accuracy of the model is improved. Here, this step may not be performed during each data processing, but may be performed before model training, and is not limited thereto.

S22, according to the first characteristics, obtaining second characteristics of the user behavior prediction task.

In one or more embodiments, based on the first feature, obtaining a second feature of the user behavior prediction task includes:

(1) Initializing a first preset model, namely xDeepFM models, including the structure, parameters and the like of the models;

(2) Processing the first features according to a first preset model, namely xDeepFM model, inputting the first features into a xDeepFM model, and performing high-order cross representation learning of the features by the model through a Deep Neural Network (DNN) and CIN (Compressed Interaction Network) component to acquire second features of a user behavior prediction task;

(3) A high-order cross-feature representation of the second feature is output for use in the input of the ESMM model training unit.

S23, carrying out joint training on the user behavior prediction tasks according to the second characteristics so as to obtain a multi-task model of the user behavior prediction.

In one or more embodiments, the step S23 includes:

(1) The high-order cross-feature representation of the second feature of the xDeepFM model output is used as input to the ESMM model training unit.

(2) And initializing a second preset model, namely ESMM models, including the structures, parameters and the like of the models. The model adopts a multi-task learning method; the user behavior prediction task includes at least two user behavior prediction subtasks, such as a CTR prediction task and a CVR prediction task, respectively.

(3) And carrying out joint training on the second characteristics of at least two user behavior prediction tasks through a second preset model to obtain a multi-task model of user behavior prediction.

In the training process, model parameters are updated by adopting a loss function optimization technology, for example, so as to improve the learning and characterization capacity of the model; and/or regularization technology is adopted to prevent the parameter from being updated excessively, so that the problem of fitting excessively occurs, and the generalization capability of the model is ensured. Here, the loss function optimization technique may perform optimization of the model parameters using an optimization algorithm such as a gradient descent method, a random gradient descent method (SGD), or the like.

(4) And evaluating the performances of the model in different training iterations by setting a verification set or a cross verification method and the like, so as to select optimal model parameters. Here, this step may not be performed during each data processing, but may be performed during a model training phase, and is not limited thereto.

In the method, the CTR and CVR tasks are jointly trained through a multi-task learning mode of the ESMM model, but in order to further increase the expression capacity of the model, the cross relation among the features is fully utilized, the xDeepFM model is further introduced, and the strong high-order feature cross capacity is utilized to enhance the feature expression capacity of the ESMM model. The xDeepFM model can automatically learn and cross the features, and effectively capture the interaction relation between different features, so that the prediction accuracy of CTR and CVR is improved.

It should be understood that, although the steps in the flowcharts of the figures are shown in order as indicated by the arrows, these steps are not necessarily performed in order as indicated by the arrows. The steps are not strictly limited in order and may be performed in other orders, unless explicitly stated herein. Moreover, at least some of the steps in the flowcharts of the figures may include a plurality of sub-steps or stages that are not necessarily performed at the same time, but may be performed at different times, the order of their execution not necessarily being sequential, but may be performed in turn or alternately with other steps or at least a portion of the other steps or stages.

[ User behavior prediction method based on multitasking ]

The disclosure also provides a user behavior prediction method based on multitasking, which is mainly used for applying the trained ESMM model to an actual scene to output the prediction results of CTR and CVR. After model training is completed, the model needs to be deployed into an actual application scene. Over time, the model may need to be updated and optimized according to new data and requirements. Here, this step may not need to be performed during each data processing, and is not limited.

In one or more embodiments, the method for predicting user behavior based on multitasking is specifically implemented as follows:

(1) Acquiring data to be predicted of a user behavior prediction task, for example, at least one data set as original data;

(2) Inputting the data to be predicted into a multitasking model, namely a trained ESMM model, and carrying out model reasoning;

(3) And calculating corresponding CTR and CVR predicted values according to model reasoning so as to obtain a predicted result of the user behavior prediction task. And according to the prediction result, personalized recommendation and optimization can be performed in the fields of advertisement recommendation, electronic commerce and the like. For example, in advertisement recommendation, advertisements with higher CTR and CVR may be preferentially displayed according to the prediction results of CTR and CVR, thereby improving the overall advertisement effect. The results of the CTR and CVR double-target prediction are shown in the following Table 1:

Model name	CTR	CVR
			MMOE	0.8305583	0.7356386
PLE	0.8338284	0.74134326
			ESMM	0.8270741	0.7415657
ESMM+xDeepFM	0.8380734	0.7531605

TABLE 1

As can be seen from table 1, in the CTR and CVR joint prediction problem, the model proposed by the present disclosure performs better than other models in both indexes.

Through the steps, the method for modeling and predicting the CTCVR of the ESMM reinforced structure by using xDeepFM is realized. On the basis of keeping the advantages of the original ESMM model, the representation capability of the model is further enhanced by introducing xDeepFM, and the joint prediction of CTR and CVR is realized, so that the prediction performance is remarkably improved. In addition, the technical scheme disclosed by the invention has better generalization capability and practical application value, and is easy to deploy and expand in a practical system. By means of the technical scheme, more accurate CTR and CVR prediction results can be provided for the fields of advertisement recommendation, electronic commerce and the like, and therefore more efficient personalized recommendation and optimization are achieved.

According to the multitasking-based user behavior prediction method disclosed by the invention, the following technical effects can be achieved:

High-order feature interaction learning: by using xDeepFM model algorithms, higher-order interactions between features can be learned, capturing more abundant information and potential laws. This will help to improve the accuracy of CTR and CVR predictions.

Joint prediction: by combining the xDeepFM model with the ESMM model, a joint prediction of CTR and CVR is achieved. This helps optimize the advertisement delivery strategy, reduces the risk of misdelivery, thereby improving advertising effectiveness and delivery benefits.

High prediction accuracy: because xDeepFM is adopted to replace the original part DNN structure of ESMM, the expression capacity of the model can be improved, and the prediction accuracy is further improved. This will help provide more accurate recommendations for advertisers and users, thereby improving user experience and advertising effectiveness.

[ User behavior prediction model training device based on multitasking ]

In order to implement the technical solution in the embodiments of the present disclosure, an embodiment of the present disclosure provides a device for training a user behavior prediction model based on multiple tasks, as shown in fig. 3, where the device for training a user behavior prediction model based on multiple tasks includes, for example, a first acquiring unit 301, a second acquiring unit 302, and a training unit 303, and of course, the device for training a user behavior prediction model based on multiple tasks may also include other functional operation modules, which is not limited.

A first obtaining unit 301, configured to obtain user behavior data and obtain a first feature of a user behavior prediction task according to the user behavior data; here, the first obtaining unit 301 may implement, for example, the corresponding function or technical solution in step S21 or other steps in the above method, which is not described herein.

A second obtaining unit 302, configured to obtain, according to the first feature, a second feature of the user behavior prediction task; here, the second obtaining unit 302 may implement the corresponding function or technical solution in step S22 or other steps in the above method, which is not described herein.

And the training unit 303 is configured to perform joint training on the user behavior prediction task according to the second feature, so as to obtain a multitasking model of user behavior prediction. Here, the training unit 303 may implement the corresponding function or technical scheme in step S23 or other steps in the above method, which is not described herein.

It should be understood that while each block in the block diagrams of the figures may represent a module, a portion of the module contains one or more executable instructions for implementing the specified logical function(s), the modules are not necessarily sequentially executed in order. The modules and functional units in the embodiments of the apparatus in the present disclosure may be integrated into one processing module, or each unit may exist alone physically, or two or more modules or functional units may be integrated into one module. The integrated modules may be implemented in hardware or in software functional modules. The integrated modules may also be stored in a computer readable storage medium if implemented in the form of software functional modules and sold or used as a stand-alone product. The above-mentioned storage medium may be a read-only memory, a magnetic disk or an optical disk, or the like.

[ User behavior prediction apparatus based on multitasking ]

The disclosure also provides a user behavior prediction device based on multitasking, which is mainly used for applying the trained ESMM model to an actual scene to output prediction results of CTR and CVR. After model training is completed, the model needs to be deployed into an actual application scene. Over time, the model may need to be updated and optimized according to new data and requirements. For example, the training device includes the first acquiring unit 301 (401), the second acquiring unit 302 (402), and the training unit 303 (403) in the foregoing training device for a user behavior prediction model, and corresponding functions are not described herein again.

The user behavior prediction device based on the multitasking further comprises a prediction unit 404, configured to predict the user behavior prediction task according to the multitasking model of the user behavior prediction, and obtain a prediction result. Here, the prediction unit 404 may implement, for example, the corresponding functions or technical solutions in the above-mentioned multitasking user behavior prediction method or other steps, which are not described herein.

[ Computer Equipment ]

Referring now to FIG. 5, there is illustrated a schematic diagram of a computer device suitable for use in implementing embodiments of the present disclosure. The computer device in the embodiments of the present disclosure is only one example, and should not impose any limitation on the functionality and scope of use of the embodiments of the present disclosure.

As shown in fig. 5, the electronic device 500 may include a processing means (e.g., a central processor, a graphics processor, etc.) 501 for controlling the overall operation of the electronic device. The processing means may comprise one or more processors to execute instructions to perform all or part of the steps of the methods described above. The processing device 501 may also include one or more modules for processing interactions with other devices.

The storage 502 is used to store various types of data, and the storage 502 may be a system, apparatus, or device that includes various types of computer readable storage media, or a combination thereof, such as electrical, magnetic, optical, electromagnetic, infrared, or semiconductor, or a combination of any of the foregoing. More specific examples of the computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this disclosure, a computer-readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.

The sensor means 503 for sensing the prescribed measured information and converting it into a usable output signal according to a certain rule may comprise one or more sensors. For example, it may include an acceleration sensor, a gyro sensor, a magnetic sensor, a pressure sensor, a temperature sensor, or the like for detecting changes in the on/off state, relative positioning, acceleration/deceleration, temperature, humidity, light, or the like of the electronic apparatus.

The processing means 501, the memory means 502 and the sensor means 503 are connected to each other by a bus 504. An input/output (I/O) interface 505 is also connected to bus 504.

The multimedia device 506 may include an input device such as a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, etc. for receiving input signals from a user, where various input devices may cooperate with various sensors of the sensor device 503 to perform gesture operation input, image recognition input, distance detection input, etc.; the multimedia device 506 may also include an output device such as a Liquid Crystal Display (LCD), speaker, vibrator, etc.

The power supply device 507, which is used to provide power to various devices in the electronic apparatus, may include a power management system, one or more power supplies, and components to distribute power to other devices.

The communication means 508 may allow the electronic device 500 to communicate with other devices wirelessly or by wire to exchange data.

Each of the above-described devices may also be connected to the I/O interface 505 to enable the application of the electronic apparatus 500.

While an electronic device having various means is shown in the figures, it is to be understood that not all of the illustrated means are required to be implemented or provided. More or fewer devices may be implemented or provided instead.

In particular, according to embodiments of the present disclosure, the processes described above with reference to flowcharts may be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program product comprising a computer program embodied on a non-transitory computer readable medium, the computer program comprising program code for performing the method shown in the flow chart. In such embodiments, the computer program may be downloaded and installed from a network via a communications device, or from a storage device. The above-described functions defined in the methods of the embodiments of the present disclosure are performed when the computer program is executed by a processing device.

In the context of this disclosure, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.

It is noted that the computer readable medium described above in the present disclosure may be a computer readable signal medium or a computer readable storage medium or any combination of the two. In the present disclosure, however, the computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, with the computer-readable program code embodied therein. Such a propagated data signal may take any of a variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination of the foregoing. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to: electrical wires, fiber optic cables, RF (radio frequency), and the like, or any suitable combination of the foregoing.

The computer readable medium may be contained in the electronic device; or may exist alone without being incorporated into the electronic device.

Computer program code for carrying out operations of the present disclosure may be written in one or more programming languages, including, but not limited to, an object oriented programming language such as Java, python, C ++ and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of remote computers, the remote computer may be connected to the user computer through any kind of network or may be connected to an external computer (e.g., connected through the internet using an internet service provider).

The flowcharts and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

The units involved in the embodiments of the present disclosure may be implemented by means of software, or may be implemented by means of hardware. Wherein the names of the units do not constitute a limitation of the units themselves in some cases.

The functions described above herein may be performed, at least in part, by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: a Field Programmable Gate Array (FPGA), an Application Specific Integrated Circuit (ASIC), an Application Specific Standard Product (ASSP), a system on a chip (SOC), a Complex Programmable Logic Device (CPLD), and the like.

According to one or more embodiments of the present disclosure, a method for training a user behavior prediction model based on multitasking is provided, which adopts the following technical scheme, including:

According to one or more embodiments of the present disclosure, a method for training a user behavior prediction model based on multitasking is provided, and the following technical solutions are adopted, where the obtaining user behavior data includes:

Acquiring at least one data set corresponding to the user behavior prediction task;

And carrying out data preprocessing on the data set to acquire the user behavior data.

According to one or more embodiments of the present disclosure, a method for training a user behavior prediction model based on multitasking is provided, which adopts the following technical scheme, where the obtaining, according to the user behavior data, a first feature of a user behavior prediction task includes:

performing feature engineering on the user behavior data to extract and obtain target features of the user behavior prediction task;

and acquiring a first characteristic of the user behavior prediction task according to the target characteristic.

According to one or more embodiments of the present disclosure, a method for training a user behavior prediction model based on multitasking is provided, which adopts a technical scheme that, according to the first feature, acquiring a second feature of the user behavior prediction task includes:

initializing a first preset model;

and processing the first characteristics according to the first preset model, and acquiring second characteristics of the user behavior prediction task.

According to one or more embodiments of the present disclosure, a method for training a user behavior prediction model based on multitasking is provided, which adopts a technical scheme that, according to the second feature, jointly training the user behavior prediction task to obtain a multitasking model of user behavior prediction includes:

the user behavior prediction tasks comprise at least two user behavior prediction tasks;

Initializing a second preset model;

And carrying out joint training on the second characteristics of the at least two user behavior prediction tasks through the second preset model to obtain a multitask model of the user behavior prediction.

According to one or more embodiments of the present disclosure, a method for predicting user behavior based on multitasking is provided, which adopts the following technical scheme, including:

acquiring data to be predicted of the user behavior prediction task;

inputting the data to be predicted into the multitasking model for model reasoning;

And obtaining a prediction result of the user behavior prediction task according to the model reasoning.

According to one or more embodiments of the present disclosure, there is provided a user behavior prediction model training apparatus based on multitasking, which adopts the following technical scheme, including:

According to one or more embodiments of the present disclosure, there is provided a user behavior prediction apparatus based on multitasking, which adopts the following technical solution, including:

The user behavior prediction model training device as described above;

According to one or more embodiments of the present disclosure, there is provided a computer device comprising a memory having a computer program stored therein and a processor, which when executing the computer program implements the method of any of the preceding claims.

According to one or more embodiments of the present disclosure, there is provided a computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements a method as in any of the preceding claims.

The foregoing description is only of the preferred embodiments of the present disclosure and description of the principles of the technology being employed. It will be appreciated by persons skilled in the art that the scope of the disclosure referred to in this disclosure is not limited to the specific combinations of features described above, but also covers other embodiments which may be formed by any combination of features described above or equivalents thereof without departing from the spirit of the disclosure. Such as those described above, are mutually substituted with the technical features having similar functions disclosed in the present disclosure (but not limited thereto).

Moreover, although operations are depicted in a particular order, this should not be understood as requiring that such operations be performed in the particular order shown or in sequential order. In certain circumstances, multitasking and parallel processing may be advantageous. Likewise, while several specific implementation details are included in the above discussion, these should not be construed as limiting the scope of the present disclosure. Certain features that are described in the context of separate embodiments can also be implemented in combination in a single embodiment. Conversely, various features that are described in the context of a single embodiment can also be implemented in multiple embodiments separately or in any suitable subcombination.

Although the subject matter has been described in language specific to structural features and/or methodological acts, it is to be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are example forms of implementing the claims.

Claims

1. A multitasking-based user behavior prediction model training method, comprising:

2. The method of claim 1, wherein the obtaining user behavior data comprises:

3. The method of claim 1, wherein the obtaining a first characteristic of a user behavior prediction task from the user behavior data comprises:

4. The method of claim 1, wherein the obtaining a second characteristic of the user behavior prediction task from the first characteristic comprises:

initializing a first preset model;

5. The method of any of claims 1-4, wherein the jointly training the user behavior prediction tasks to obtain a multitasking model of user behavior prediction based on the second features comprises:

Initializing a second preset model;

6. A method for predicting user behavior based on multitasking, comprising:

Predicting said user behavior and obtaining a prediction result by using said multitasking model trained by a multitasking-based user behavior prediction model training method as claimed in any one of claims 1-5, said method comprising:

acquiring data to be predicted of the user behavior prediction task;

7. A multitasking based user behavior prediction model training apparatus, comprising:

8. A multitasking-based user behavior prediction apparatus, comprising:

The user behavior prediction model training apparatus of claim 7;

9. A computer device comprising a memory and a processor, the memory having stored therein a computer program, the processor implementing the method of any of claims 1-6 when the computer program is executed.

10. A computer readable storage medium, characterized in that the computer readable storage medium has stored thereon a computer program which, when executed by a processor, implements the method according to any of claims 1-6.