CN114610996A - Information pushing method and device - Google Patents

Information pushing method and device Download PDF

Info

Publication number
CN114610996A
CN114610996A CN202210235884.3A CN202210235884A CN114610996A CN 114610996 A CN114610996 A CN 114610996A CN 202210235884 A CN202210235884 A CN 202210235884A CN 114610996 A CN114610996 A CN 114610996A
Authority
CN
China
Prior art keywords
information
training
network model
task
model
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202210235884.3A
Other languages
Chinese (zh)
Inventor
李天浩
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Jingdong Century Trading Co Ltd
Beijing Wodong Tianjun Information Technology Co Ltd
Original Assignee
Beijing Jingdong Century Trading Co Ltd
Beijing Wodong Tianjun Information Technology Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Beijing Jingdong Century Trading Co Ltd, Beijing Wodong Tianjun Information Technology Co Ltd filed Critical Beijing Jingdong Century Trading Co Ltd
Priority to CN202210235884.3A priority Critical patent/CN114610996A/en
Publication of CN114610996A publication Critical patent/CN114610996A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9535—Search customisation based on user profiles and personalisation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/06—Buying, selling or leasing transactions
    • G06Q30/0601—Electronic shopping [e-shopping]
    • G06Q30/0631—Item recommendations

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Business, Economics & Management (AREA)
  • Databases & Information Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Computational Linguistics (AREA)
  • Artificial Intelligence (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Biophysics (AREA)
  • Biomedical Technology (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Accounting & Taxation (AREA)
  • Finance (AREA)
  • Development Economics (AREA)
  • Economics (AREA)
  • Marketing (AREA)
  • Strategic Management (AREA)
  • General Business, Economics & Management (AREA)
  • User Interface Of Digital Computer (AREA)

Abstract

The invention discloses an information pushing method and device, and relates to the technical field of artificial intelligence. One embodiment of the method comprises: inputting the user behavior characteristics and the information characteristics corresponding to the user behavior characteristics into the trained multi-task network model; predicting a plurality of behavior categories of the user corresponding to the operation information according to the output of the multitask network model; screening target display information according to the behavior category and pushing the target display information to a terminal; a plurality of behavior categories of the user are predicted through the trained multi-task network model, corresponding target display information is screened out, accuracy of information pushing is improved, and then conversion rate of the information pushing is improved.

Description

Information pushing method and device
Technical Field
The invention relates to the field of artificial intelligence, in particular to a method and a device for pushing information.
Background
With the development of information technology and internet industry, information overload becomes a challenge for users to process information, so in internet application, a method for pushing personalized information for users is generally adopted to reduce the influence of information overload on users, and meanwhile, the information pushing accuracy can be improved.
Particularly, for the information that the push information is the goods sold by the e-commerce, various push targets of the goods sold by the e-commerce need to be considered, for example, when pushing, besides predicting the click rate of the user on the goods sold by the e-commerce, the purchase rate of the user on the goods needs to be predicted, so as to increase the conversion rate of the push information.
At present, a machine learning method is mainly used for constructing a model aiming at a single pushing target, information is pushed through the model subsequently, and the problems of low pushing efficiency and low accuracy rate of the pushed information are mainly caused aiming at the single pushing target.
Disclosure of Invention
In view of this, embodiments of the present invention provide a method and an apparatus for pushing information, which can input user behavior characteristics and information characteristics corresponding to the user behavior characteristics into a trained multitask network model; predicting a plurality of behavior categories of the user to operation corresponding information according to the output of the multitask network model; screening target display information according to the behavior category and pushing the target display information to a terminal; a plurality of behavior categories of the user are predicted through the trained multi-task network model, corresponding target display information is screened out, accuracy of information pushing is improved, and then conversion rate of the information pushing is improved.
In order to achieve the above object, according to an aspect of the embodiments of the present invention, there is provided an information pushing method, including: acquiring user behavior characteristics of a terminal; inputting the user behavior characteristics and the information characteristics of a plurality of pieces of information in an information pool corresponding to the user behavior characteristics into a trained multi-task network model; the multitask network model comprises a regression task model and a classification task model, and is trained by a plurality of push targets and a sample with the plurality of push targets; predicting a plurality of behavior categories of one or more pieces of information corresponding to the operation by the user according to the output of the multitask network model; and screening target display information from the plurality of pieces of information according to the behavior category, and pushing the target display information to the terminal.
Optionally, the information pushing method further includes:
acquiring historical data of a plurality of behavior categories aiming at a plurality of types of training information; generating a first user behavior feature for training based on the historical data; acquiring a user portrait characteristic corresponding to the first user behavior characteristic for training; adding the first user behavior feature for training and the corresponding user portrait feature to a training sample set; and training the multitask network model by utilizing the training sample set.
Optionally, the information pushing method further includes:
inputting the first user behavior characteristics for training and the corresponding portrait characteristics into a trained generative confrontation network model, and outputting a plurality of simulation samples by using the generative confrontation network model; adding the simulation sample to the set of training samples.
Optionally, in the information pushing method, the generating a confrontation network model includes a discrimination network and multiple generation networks; determining the categories of the simulation samples output by the plurality of generating networks by utilizing the judging network; the category of the simulation sample comprises a simulation positive sample or a simulation negative sample; training the multitask network model; the method comprises the following steps: selecting the simulation positive sample and/or the simulation negative sample to be added to a training sample set corresponding to the service type according to the service type; and training the multi-task network model by utilizing the training sample set corresponding to the service type.
Optionally, the information pushing method includes:
the multitasking network model comprises a multilayer perceptron; training the multitask network model; the method comprises the following steps: inputting the first user behavior characteristic for training and information characteristics of a plurality of types of information for training into the multilayer perceptron, and outputting the association information of the first user behavior characteristic for training and the information characteristics by using the multilayer perceptron; and inputting the associated information into the multitask network model.
Optionally, the information pushing method includes: a loss function corresponding to the multitask network model is constructed on the basis of a regression task function of the regression task model and a classification task function of the classification task model; training the multitask network model comprises: and training the regression task model and the classification task model, and controlling training iteration through the loss function, the regression task function and the classification task function respectively.
Optionally, the information pushing method includes:
classifying behavior categories for the training first user behavior features; determining a task type to which the behavior category belongs; the task type is any one of a regression task or a classification task;
training the multitask network model, including: training a regression task model contained in the multi-task network model by using training samples contained in the training sample set and belonging to the regression task; training a classification task model contained in the multi-task network model by using training samples contained in the training sample set and belonging to the classification task; and taking the output of the regression task model and/or the output of the classification task model as the output of the multitask network model.
In order to achieve the above object, according to a second aspect of the embodiments of the present invention, there is provided an information pushing apparatus, including: the system comprises a user information acquisition module, a user behavior prediction module and a push information acquisition module; wherein the content of the first and second substances,
the user information acquiring module is used for acquiring the user behavior characteristics of the terminal; inputting the user behavior characteristics and information characteristics of a plurality of pieces of information in an information pool corresponding to the user behavior characteristics into a trained multi-task network model; the multitask network model comprises a regression task model and a classification task model, and is trained by a plurality of push targets and a sample with the plurality of push targets;
the user behavior predicting module is used for predicting a plurality of behavior categories of one or more pieces of information corresponding to the operation by the user according to the output of the multitask network model;
and the information acquisition and pushing module is used for screening target display information from the plurality of pieces of information according to the behavior category and pushing the target display information to the terminal.
Optionally, the information pushing apparatus further includes:
acquiring historical data of a plurality of behavior categories aiming at a plurality of types of training information; generating a first user behavior feature for training based on the historical data; acquiring a user portrait characteristic corresponding to the first user behavior characteristic for training; adding the first user behavior feature for training and the corresponding user portrait feature to a training sample set; and training the multitask network model by utilizing the training sample set.
Optionally, the information pushing apparatus further includes:
inputting the first user behavior characteristics for training and the corresponding portrait characteristics into a trained generative confrontation network model, and outputting a plurality of simulation samples by using the generative confrontation network model; adding the simulation sample to the set of training samples.
Optionally, the information pushing apparatus is configured to generate the confrontation network model including a discrimination network and a plurality of generation networks; determining the categories of the simulation samples output by the plurality of generating networks by utilizing the judging network; the category of the simulation sample comprises a simulation positive sample or a simulation negative sample; training the multitask network model; the method comprises the following steps: selecting the simulation positive sample and/or the simulation negative sample to be added to a training sample set corresponding to the service type according to the service type; and training the multi-task network model by utilizing the training sample set corresponding to the service type.
Optionally, the information pushing apparatus includes:
the multitasking network model comprises a multilayer perceptron; training the multitask network model; the method comprises the following steps: inputting the first user behavior characteristics for training and information characteristics of a plurality of types of information for training into the multilayer perceptron, and outputting association information of the first user behavior characteristics for training and the information characteristics by using the multilayer perceptron; and inputting the associated information into the multitask network model.
Optionally, the information pushing apparatus includes:
a loss function corresponding to the multitask network model is constructed on the basis of a regression task function of the regression task model and a classification task function of the classification task model; training the multitask network model comprises: and training the regression task model and the classification task model, and controlling training iteration through the loss function, the regression task function and the classification task function respectively.
Optionally, the information pushing apparatus includes: classifying behavior categories for the training first user behavior features; determining a task type to which the behavior category belongs; the task type is any one of a regression task or a classification task; training the multitask network model, including: training a regression task model contained in the multi-task network model by using training samples contained in the training sample set and belonging to the regression task; training a classification task model contained in the multi-task network model by using training samples contained in the training sample set and belonging to the classification task; and taking the output of the regression task model and/or the output of the classification task model as the output of the multitask network model.
To achieve the above object, according to a third aspect of the embodiments of the present invention, there is provided an electronic device for pushing information, including: one or more processors; the storage device is used for storing one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of the information push methods.
To achieve the above object, according to a fourth aspect of the embodiments of the present invention, there is provided a computer readable medium having a computer program stored thereon, wherein the computer program is configured to implement, when executed by a processor, any one of the above methods for pushing information.
One embodiment of the above invention has the following advantages or benefits: the user behavior characteristics and the information characteristics corresponding to the user behavior characteristics can be input into the trained multi-task network model; predicting a plurality of behavior categories of the user corresponding to the operation information according to the output of the multitask network model; screening target display information according to the behavior category and pushing the target display information to a terminal; a plurality of behavior categories of the user are predicted through the trained multi-task network model, corresponding target display information is screened out, accuracy of information pushing is improved, and then conversion rate of the information pushing is improved.
Further effects of the above-mentioned non-conventional alternatives will be described below in connection with the embodiments.
Drawings
The drawings are included to provide a better understanding of the invention and are not to be construed as unduly limiting the invention. Wherein:
fig. 1 is a schematic flowchart of a method for pushing information according to an embodiment of the present invention;
FIG. 2 is a schematic flow chart of training a multitasking network model according to an embodiment of the present invention;
fig. 3 is a schematic structural diagram of an information pushing apparatus according to an embodiment of the present invention;
FIG. 4 is an exemplary system architecture diagram in which embodiments of the present invention may be employed;
fig. 5 is a schematic block diagram of a computer system suitable for use in implementing a terminal device or server according to an embodiment of the present invention.
Detailed Description
Exemplary embodiments of the present invention are described below with reference to the accompanying drawings, in which various details of embodiments of the invention are included to assist understanding, and which are to be considered as merely exemplary. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the invention. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.
As shown in fig. 1, an embodiment of the present invention provides an information pushing method, which may include the following steps:
step S101: acquiring user behavior characteristics of a terminal; inputting the user behavior characteristics and information characteristics of a plurality of pieces of information in an information pool corresponding to the user behavior characteristics into a trained multi-task network model; the multitask network model comprises a regression task model and a classification task model, and is trained by a plurality of push targets and a sample with the plurality of push targets.
Specifically, the user behavior characteristics are indicated as different user behaviors for different application scenarios, for example: aiming at an application scene that a user uses the electronic mall, user behavior characteristics comprise browsing, clicking, purchasing, evaluating and the like of the user on the application containing information of the electronic mall; aiming at an application scene that a user uses the video service, the user behavior characteristics comprise browsing, evaluation, praise, collection and the like of the user on video information. Further, the information pool contains a plurality of information (such as article information, video information and the like) to be pushed; the information characteristics include characteristic information for describing information, such as: when the information is article information, the information characteristics include characteristics such as a category, a model, a color, and a historical sales amount of the article, and when the information is video information, the information characteristics include characteristics such as a category, a content characteristic, and a click amount of the video.
And further, inputting the user behavior characteristics and information characteristics of a plurality of pieces of information in an information pool corresponding to the user behavior characteristics into a trained multi-task network model. The trained multi-task network model comprises a regression task model and a classification task model, and is trained by a plurality of push targets and samples with the push targets. Specifically, the multitasking network model may be an optimized MMOE model (Modeling Task Relationships in Multi-Task Learning with Multi-gate texture-of-Exters model); further, the multitask network model comprises a regression task model and a classification task model, and is trained by a plurality of pushing targets (such as click rate, browsing duration, purchase rate and the like) and a sample with the plurality of pushing targets. The description of training the multitask network model is consistent with the description of step S201-step S204, and is not repeated here.
In various application scenarios of information push, a plurality of push targets need to be considered, for example, for an application scenario of an electronic mall, information push needs to combine a user click rate and a purchase rate to improve a conversion rate of item information; in the application scenario of information (e.g. video) recommendation, information push needs to combine the user browsing rate and evaluation rate to obtain the conversion rate of information; therefore, the user behavior characteristics can be used as input by utilizing the well-trained multi-task network model (comprising a regression task model and a classification task model), and one or more information can be pushed for the user by integrating various user behavior characteristics and information characteristics, so that the accuracy and the pushing effect of information pushing are improved, and the conversion rate of pushed information is further improved.
Step S102: and predicting a plurality of behavior categories of the user on one or more pieces of information corresponding to the operation according to the output of the multitask network model.
Specifically, the output of the multitask network model is generated by the output of the regression task model and/or the classification task model, and a corresponding task type may be selected for the category of the push target, for example: if the predicted pushing target is the click rate, the output of the classification task model can be used for prediction; if the predicted pushing target is the browsing duration, the probability corresponding to the browsing duration range can be predicted by utilizing the output of the regression task model; further, combining the output of the classification task model and/or the output of the regression task model as the output of the multitask network model, thereby predicting a plurality of behavior categories of one or more pieces of information corresponding to the operation by the user; the behavior category is set according to different application scenarios, for example: aiming at the application scene of the electronic market, the behavior category of the user can be click, purchase, the time length for adding a shopping cart and the like; for the video service application scene, the user behavior category can be click, browsing duration, collection and the like.
Step S103: and screening target display information from the plurality of pieces of information according to the behavior category, and pushing the target display information to the terminal.
Specifically, according to the numerical value of the behavior category in the application scene, the numerical value may be (0, 1) and the like for the classification task model; aiming at the regression task model, the numerical value can be one or more probability values, and the multitask network model screens target display information from the multiple pieces of information for a user by combining the output numerical value of the classification task model and the output numerical value of the regression task model; when the target display information is screened, a set number of display information can be selected as the target display information according to a set sequence (for example, a sequence from high sales to low sales), and the target display information is pushed to the terminal so as to display corresponding push information on the terminal.
As shown in fig. 2, an embodiment of the present invention provides a method for training a multitask network model, which may include the following steps:
step S201: acquiring historical data of a plurality of behavior categories aiming at a plurality of types of training information; generating a first user behavior feature for training based on the historical data; acquiring a user portrait characteristic corresponding to the first user behavior characteristic for training; adding the first user behavior feature for training and the corresponding user portrait feature to a training sample set.
Specifically, the first user behavior feature sample data for training is acquired, the historical data of a plurality of behavior categories for a plurality of kinds of information for training is acquired, preferably, the historical data can be preprocessed by using a training library generation module, for example, browsing information, clicking information and user portrait features of a user are acquired, the historical data can be combined with a user equipment number, the historical data indicated as invalid data can be removed, and the efficiency and the effect of training the multitask network model are improved; further, a sampling module can be used for sampling positive and negative samples based on the original training samples obtained in the training library generating module, and the proportion of the positive and negative samples is determined according to an application scene; further utilizing a characteristic generation module to further extract characteristics of the training samples obtained by the sampling module, and generating a training sample set; generating a first user behavior characteristic for training based on the historical data; acquiring a user portrait characteristic corresponding to the first user behavior characteristic for training; adding the first user behavior feature for training and the corresponding user portrait feature to a training sample set.
Step S202: inputting the first user behavior characteristics for training and the corresponding portrait characteristics into a trained generative confrontation network model, and outputting a plurality of simulation samples by using the generative confrontation network model; adding the simulation sample to the set of training samples.
Further, on the basis of obtaining a training sample based on historical data, a simulation sample is generated by utilizing a generation confrontation network model; therefore, the number and the information capacity of the training samples are increased, the effect of training the multi-task network model is further improved, and the prediction accuracy of the multi-task network model is improved. The method for unsupervised learning is to generate a confrontation network (GAN), and to learn the confrontation network by means of mutual game of two neural networks; the generation countermeasure network is composed of a generation network and a discrimination network. The invention adopts a plurality of generation networks and a single judgment network model, namely, the generation countermeasure network model comprises a judgment network and a plurality of generation networks; to improve the simulation degree of the simulation sample, the example of the confrontation network model is generated as shown in formula (1).
Figure BDA0003539991450000091
Wherein G represents a function corresponding to the generator, and D represents a function corresponding to the discriminator; x, z represent the input characteristics and E represents the expected value. Inputting the first user behavior characteristics for training and the corresponding portrait characteristics into a trained generative confrontation network model, and outputting a plurality of simulation samples by using the generative confrontation network model; further, the simulation sample is added to the set of training samples.
Preferably, the generation countermeasure network model includes a discriminant network and a plurality of generation networks; determining the categories of the simulation samples output by the plurality of generating networks by utilizing the judging network; the category of the simulation sample comprises a simulation positive sample or a simulation negative sample; training the multitask network model; the method comprises the following steps: selecting the simulation positive sample and/or the simulation negative sample to be added to a training sample set corresponding to the service type according to the service type; and training the multi-task network model by utilizing the training sample set corresponding to the service type. Specifically, according to application scenarios corresponding to different service types, a simulation true sample and/or a simulation negative sample are/is selected from simulation samples and used as the simulation samples corresponding to the service types, and the simulation samples are added into a training sample set, so that the feature discrimination and the sample information capacity of the training samples are further improved.
Before a simulation sample is obtained by using a generated confrontation network model, the generated confrontation network model needs to be trained, specifically, in the training process, a generator parameter can be firstly fixed to train a discriminator, and then a discriminator parameter is fixed to train the generator, and iteration is carried out; and selecting a simulation positive sample and/or a simulation negative sample according to the service type, and adjusting the proportion of the real training sample and the simulation sample according to the service type by combining the characteristic information of the real training sample, thereby improving the information capacity of the sample.
Step S203: inputting the first user behavior characteristic for training and information characteristics of a plurality of types of information for training into the multilayer perceptron, and outputting the association information of the first user behavior characteristic for training and the information characteristics by using the multilayer perceptron; and inputting the associated information into the multitask network model.
Specifically, the multitask network model includes a Multilayer perceptron (MLP), and preferably, the first user behavior feature for training and information features of the training information are input into the Multilayer perceptron, the Multilayer perceptron is used to output associated information between the first user behavior feature for training and the information features, and understanding of the associated information of the sample input features is enhanced, so that information capacity of training samples is further expanded.
Further, the correlation information is input into the multitask network model to train the multitask network model.
Step S204: and training the multitask network model by utilizing the training sample set.
Specifically, a training sample set for training is obtained, and the multitask network model is trained.
Further, a loss function corresponding to the multitask network model is constructed based on a regression task function of the regression task model and a classification task function of the classification task model; training the multitask network model comprises: and training the regression task model and the classification task model, and controlling training iteration through the loss function, the regression task function and the classification task function respectively.
Specifically, a loss function corresponding to the multitask network model is shown in formula (2), where ω represents a model parameter required by the multitask network model, and σ represents a variance; Σ represents the sum of k tasks (classification tasks and/or regression tasks).
Figure BDA0003539991450000111
Illustratively, the classification task function of the classification task model is shown in equation (3).
-logSoftmax(y，fW(x)) (3)
The variance associated with the regression task function of the regression task model is shown in formula (4)
||y-fW(x)||2 (4)
Namely, a loss function corresponding to the multitask network model is constructed on the basis of a regression task function of the regression task model and a classification task function of the classification task model; training the multitask network model comprises: and training the regression task model and the classification task model, and controlling training iteration through the loss function, the regression task function and the classification task function respectively. And training a loss function, the regression task function and the classification task function corresponding to the multi-task network model through multiple iterations to generate an optimized multi-task network model through model training.
Further, the first user behavior characteristics for training are divided into behavior categories; determining a task type to which the behavior category belongs; the task type is any one of a regression task or a classification task; training the multitask network model, including: training a regression task model contained in the multi-task network model by using training samples contained in the training sample set and belonging to the regression task; training a classification task model contained in the multi-task network model by using training samples contained in the training sample set and belonging to the classification task; and taking the output of the regression task model and/or the output of the classification task model as the output of the multitask network model. Specifically, the behavior classes are divided by the first user behavior characteristics (such as clicking, purchasing, browsing and the like) according to training, and the pushing target can be generated based on the user behaviors (such as clicking, purchasing, browsing and the like), for example: click rate, purchase rate, browsing duration range probability and the like; further, determining a task type for the behavior category; for example, the behavior category is "click" and the corresponding task type is a classification task type; determining the behavior type 'browsing duration' as a regression task type and the like; the method includes the steps of training corresponding task models by using different types of training samples respectively, namely training regression task models included in the multitask network model by using training samples included in the training sample set and belonging to the regression tasks, training classification task models included in the multitask network model by using the training samples included in the training sample set and belonging to the classification tasks, and further using the output of the regression task models and/or the output of the classification task models as the output of the multitask network model to improve the training effect of the multitask network model. That is, the multitask network model, and the regression task model or the classification task model included in the multitask network model are trained by a plurality of push targets and a sample having the plurality of push targets.
As shown in fig. 3, an embodiment of the present invention provides an apparatus 300 for pushing information, including: a user information obtaining module 301, a user behavior predicting module 302 and a push information obtaining module 303; wherein the content of the first and second substances,
the user information acquiring module 301 is configured to acquire user behavior characteristics of a terminal; inputting the user behavior characteristics and information characteristics of a plurality of pieces of information in an information pool corresponding to the user behavior characteristics into a trained multi-task network model; the multitask network model comprises a regression task model and a classification task model, and is trained by a plurality of push targets and a sample with the plurality of push targets;
the predicted user behavior module 302 is configured to predict, according to the output of the multitasking network model, a plurality of behavior categories of one or more pieces of information corresponding to the operation by the user;
the information obtaining and pushing module 303 is configured to screen target display information from the multiple pieces of information according to the behavior category, and push the target display information to the terminal.
An embodiment of the present invention further provides an electronic device for information push, including: one or more processors; the storage device is used for storing one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors are enabled to realize the method provided by any one of the above embodiments.
Embodiments of the present invention further provide a computer-readable medium, on which a computer program is stored, where the computer program, when executed by a processor, implements the method provided in any of the above embodiments.
Fig. 4 shows an exemplary system architecture 400 of an information pushing method or an information pushing apparatus to which an embodiment of the present invention may be applied.
As shown in fig. 4, the system architecture 400 may include terminal devices 401, 402, 403, a network 404, and a server 405. The network 404 serves as a medium for providing communication links between the terminal devices 401, 402, 403 and the server 405. Network 404 may include various types of connections, such as wire, wireless communication links, or fiber optic cables, to name a few.
A user may use terminal devices 401, 402, 403 to interact with a server 405 over a network 404 to receive or send messages or the like. The terminal devices 401, 402, 403 may have various client applications installed thereon, such as an e-mall client application, a web browser application, an information service application, and the like.
The terminal devices 401, 402, 403 may be various electronic devices having display screens and supporting various client applications, including but not limited to smart phones, tablet computers, laptop portable computers, desktop computers, and the like.
The server 405 may be a server providing various services, such as a background management server providing support for client applications used by users with the terminal devices 401, 402, 403. The background management server can process the received push information acquisition request and feed back the selected target display information to the terminal equipment.
It should be noted that the method for pushing information provided by the embodiment of the present invention is generally executed by the server 405, and accordingly, a device for pushing information is generally disposed in the server 405.
It should be understood that the number of terminal devices, networks, and servers in fig. 4 is merely illustrative. There may be any number of terminal devices, networks, and servers, as desired for implementation.
Referring now to FIG. 5, shown is a block diagram of a computer system 500 suitable for use with a terminal device implementing an embodiment of the present invention. The terminal device shown in fig. 5 is only an example, and should not bring any limitation to the functions and the scope of use of the embodiments of the present invention.
As shown in fig. 5, the computer system 500 includes a Central Processing Unit (CPU)501 that can perform various appropriate actions and processes according to a program stored in a Read Only Memory (ROM)502 or a program loaded from a storage section 508 into a Random Access Memory (RAM) 503. In the RAM 503, various programs and data necessary for the operation of the system 500 are also stored. The CPU 501, ROM 502, and RAM 503 are connected to each other via a bus 504. An input/output (I/O) interface 505 is also connected to bus 504.
The following components are connected to the I/O interface 505: an input portion 506 including a keyboard, a mouse, and the like; an output portion 507 including a display such as a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), and the like, and a speaker; a storage portion 508 including a hard disk and the like; and a communication section 509 including a network interface card such as a LAN card, a modem, or the like. The communication section 509 performs communication processing via a network such as the internet. The driver 510 is also connected to the I/O interface 505 as necessary. A removable medium 511 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like is mounted on the drive 510 as necessary, so that a computer program read out therefrom is mounted into the storage section 508 as necessary.
In particular, according to the embodiments of the present disclosure, the processes described above with reference to the 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 computer readable medium, the computer program comprising program code for performing the method illustrated in the flow chart. In such an embodiment, the computer program may be downloaded and installed from a network through the communication section 509, and/or installed from the removable medium 511. The computer program performs the above-described functions defined in the system of the present invention when executed by the Central Processing Unit (CPU) 501.
It should be noted that the computer readable medium shown in the present invention can be a computer readable signal medium or a computer readable storage medium or any combination of the two. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination 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 present invention, 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. In the present invention, however, a computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. 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 thereof. A computer readable signal medium may 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: wireless, wire, fiber optic cable, RF, etc., or any suitable combination of the foregoing.
The flowchart 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 invention. 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 or flowchart illustration, and combinations of blocks in the block diagrams 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 modules and/or units described in the embodiments of the present invention may be implemented by software, and may also be implemented by hardware. The described modules and/or units may also be provided in a processor, and may be described as: a processor includes an acquire user information module, a predict user behavior module, and an acquire push information module. The names of the modules do not limit the module itself in some cases, for example, the push information acquiring module may be further described as a "module for screening target display information from the pieces of information according to the behavior category".
As another aspect, the present invention also provides a computer-readable medium that may be contained in the apparatus described in the above embodiments; or may be separate and not incorporated into the device. The computer readable medium carries one or more programs which, when executed by a device, cause the device to comprise: acquiring user behavior characteristics of a terminal; inputting the user behavior characteristics and information characteristics of a plurality of pieces of information in an information pool corresponding to the user behavior characteristics into a trained multi-task network model; the multitask network model comprises a regression task model and a classification task model, and is trained by a plurality of push targets and a sample with the plurality of push targets; predicting a plurality of behavior categories of the user for one or more pieces of information corresponding to the operation according to the output of the multitask network model; and screening target display information from the plurality of pieces of information according to the behavior category, and pushing the target display information to the terminal.
According to the embodiment of the invention, the user behavior characteristics and the information characteristics corresponding to the user behavior characteristics can be input into the trained multi-task network model; predicting a plurality of behavior categories of the user corresponding to the operation information according to the output of the multitask network model; screening target display information according to the behavior category and pushing the target display information to a terminal; a plurality of behavior categories of the user are predicted through the trained multi-task network model, corresponding target display information is screened out, accuracy of information pushing is improved, and then conversion rate of the information pushing is improved.
The above-described embodiments should not be construed as limiting the scope of the invention. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions can occur, depending on design requirements and other factors. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (10)

1. A method of information push, comprising:
acquiring user behavior characteristics of a terminal;
inputting the user behavior characteristics and information characteristics of a plurality of pieces of information in an information pool corresponding to the user behavior characteristics into a trained multi-task network model; the multitask network model comprises a regression task model and a classification task model, and is trained by a plurality of push targets and a sample with the plurality of push targets;
predicting a plurality of behavior categories of one or more pieces of information corresponding to the operation by the user according to the output of the multitask network model;
and screening target display information from the plurality of pieces of information according to the behavior category, and pushing the target display information to the terminal.
2. The method of claim 1, further comprising:
acquiring historical data of a plurality of behavior categories aiming at a plurality of types of training information; generating a first user behavior feature for training based on the historical data;
acquiring a user portrait characteristic corresponding to the first user behavior characteristic for training;
adding the first user behavior feature for training and the corresponding user portrait feature to a training sample set;
and training the multitask network model by utilizing the training sample set.
3. The method of claim 2, further comprising:
inputting the first user behavior characteristics for training and the corresponding portrait characteristics into a trained generative confrontation network model, and outputting a plurality of simulation samples by using the generative confrontation network model;
adding the simulation sample to the set of training samples.
4. The method of claim 3,
the generation confrontation network model comprises a discrimination network and a plurality of generation networks;
determining the categories of the simulation samples output by the plurality of generating networks by utilizing the judging network; the category of the simulation sample comprises a simulation positive sample or a simulation negative sample;
training the multitask network model; the method comprises the following steps:
selecting the simulation positive sample and/or the simulation negative sample to be added to a training sample set corresponding to the service type according to the service type;
and training the multi-task network model by utilizing the training sample set corresponding to the service type.
5. The method of claim 2,
the multitasking network model comprises a multilayer perceptron;
training the multitask network model; the method comprises the following steps:
inputting the first user behavior characteristic for training and information characteristics of a plurality of types of information for training into the multilayer perceptron, and outputting the association information of the first user behavior characteristic for training and the information characteristics by using the multilayer perceptron;
and inputting the associated information into the multitask network model.
6. The method according to any one of claims 2 to 5,
a loss function corresponding to the multitask network model is constructed on the basis of a regression task function of the regression task model and a classification task function of the classification task model;
training the multitask network model comprises:
training the regression task model and the classification task model,
and controlling training iteration through the loss function, the regression task function and the classification task function respectively.
7. The method according to any one of claims 2 to 5,
classifying behavior categories for the training first user behavior features;
determining a task type to which the behavior category belongs; the task type is any one of a regression task or a classification task;
training the multitask network model, including:
training a regression task model contained in the multi-task network model by using training samples contained in the training sample set and belonging to the regression task;
training a classification task model contained in the multi-task network model by using training samples contained in the training sample set and belonging to the classification task;
and taking the output of the regression task model and/or the output of the classification task model as the output of the multitask network model.
8. An information pushing apparatus, comprising: the system comprises a user information acquisition module, a user behavior prediction module and a push information acquisition module; wherein the content of the first and second substances,
the user information acquiring module is used for acquiring the user behavior characteristics of the terminal; inputting the user behavior characteristics and information characteristics of a plurality of pieces of information in an information pool corresponding to the user behavior characteristics into a trained multi-task network model; the multitask network model comprises a regression task model and a classification task model, and is trained by a plurality of push targets and a sample with the plurality of push targets;
the user behavior predicting module is used for predicting a plurality of behavior categories of one or more pieces of information corresponding to the operation by the user according to the output of the multitask network model;
and the push information acquisition module is used for screening target display information from the plurality of pieces of information according to the behavior category and pushing the target display information to the terminal.
9. An electronic device, comprising:
one or more processors;
a storage device for storing one or more programs,
when executed by the one or more processors, cause the one or more processors to implement the method of any one of claims 1-7.
10. A computer-readable medium, on which a computer program is stored, which, when being executed by a processor, carries out the method according to any one of claims 1-7.
CN202210235884.3A 2022-03-10 2022-03-10 Information pushing method and device Pending CN114610996A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210235884.3A CN114610996A (en) 2022-03-10 2022-03-10 Information pushing method and device

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210235884.3A CN114610996A (en) 2022-03-10 2022-03-10 Information pushing method and device

Publications (1)

Publication Number Publication Date
CN114610996A true CN114610996A (en) 2022-06-10

Family

ID=81862908

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210235884.3A Pending CN114610996A (en) 2022-03-10 2022-03-10 Information pushing method and device

Country Status (1)

Country Link
CN (1) CN114610996A (en)

Similar Documents

Publication Publication Date Title
CN108197652B (en) Method and apparatus for generating information
CN109492772B (en) Method and device for generating information
CN111125574A (en) Method and apparatus for generating information
CN112650841A (en) Information processing method and device and electronic equipment
CN112036954A (en) Item recommendation method and device, computer-readable storage medium and electronic device
CN113763019A (en) User information management method and device
CN116541610A (en) Training method and device for recommendation model
CN115271757A (en) Demand information generation method and device, electronic equipment and computer readable medium
CN114511152A (en) Training method and device of prediction model
WO2024051707A1 (en) Recommendation model training method and apparatus, and resource recommendation method and apparatus
CN111768218B (en) Method and device for processing user interaction information
CN113822734A (en) Method and apparatus for generating information
CN111125502B (en) Method and device for generating information
CN111026849A (en) Data processing method and device
CN116109374A (en) Resource bit display method, device, electronic equipment and computer readable medium
CN112860999B (en) Information recommendation method, device, equipment and storage medium
CN111784377B (en) Method and device for generating information
CN111949860B (en) Method and apparatus for generating a relevance determination model
CN114610996A (en) Information pushing method and device
CN113742593A (en) Method and device for pushing information
CN113920397A (en) Method and device for training image classification model and method and device for image classification
CN113609397A (en) Method and device for pushing information
CN113792952A (en) Method and apparatus for generating a model
CN112308942A (en) Method and apparatus for generating image
CN111339432A (en) Recommendation method and device of electronic object and electronic equipment

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination