CN113128597B - Method and device for extracting user behavior characteristics and classifying and predicting user behavior characteristics - Google Patents

Method and device for extracting user behavior characteristics and classifying and predicting user behavior characteristics Download PDF

Info

Publication number
CN113128597B
CN113128597B CN202110432496.XA CN202110432496A CN113128597B CN 113128597 B CN113128597 B CN 113128597B CN 202110432496 A CN202110432496 A CN 202110432496A CN 113128597 B CN113128597 B CN 113128597B
Authority
CN
China
Prior art keywords
behavior
vector
target
target user
correction
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.)
Active
Application number
CN202110432496.XA
Other languages
Chinese (zh)
Other versions
CN113128597A (en
Inventor
朱海洋
周俊
陈为
严凡
钱中昊
夏祯锋
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.)
Zhongda Group Co ltd
Zhejiang University ZJU
Original Assignee
Zhongda Group Co ltd
Zhejiang University ZJU
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 Zhongda Group Co ltd, Zhejiang University ZJU filed Critical Zhongda Group Co ltd
Priority to CN202110432496.XA priority Critical patent/CN113128597B/en
Publication of CN113128597A publication Critical patent/CN113128597A/en
Application granted granted Critical
Publication of CN113128597B publication Critical patent/CN113128597B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • 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
    • G06F16/903—Querying
    • G06F16/9035—Filtering based on additional data, e.g. user or group profiles
    • 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
    • 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
    • G06N3/044—Recurrent networks, e.g. Hopfield networks
    • 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
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • 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
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0201—Market modelling; Market analysis; Collecting market data
    • G06Q30/0202—Market predictions or forecasting for commercial activities

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Strategic Management (AREA)
  • Development Economics (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Economics (AREA)
  • Finance (AREA)
  • Computational Linguistics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Accounting & Taxation (AREA)
  • Biomedical Technology (AREA)
  • Game Theory and Decision Science (AREA)
  • Databases & Information Systems (AREA)
  • Human Resources & Organizations (AREA)
  • General Business, Economics & Management (AREA)
  • General Health & Medical Sciences (AREA)
  • Software Systems (AREA)
  • Mathematical Physics (AREA)
  • Biophysics (AREA)
  • Computing Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Marketing (AREA)
  • Molecular Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The present specification provides a method and a device for extracting user behavior characteristics and predicting classification, wherein behavior information of a target user is obtained according to the method for extracting the user behavior characteristics; the behavior information comprises behavior categories and respective occurrence moments of each behavior in a plurality of behaviors of the target user in a preset time period; determining each embedded vector corresponding to each behavior based on the behavior information to obtain an initial vector sequence formed by each embedded vector; processing the initial vector sequence by using a target model to obtain a target vector sequence representing the behavior characteristics of the target user; the target vector sequence is used for classified predictive traffic involving the target user. Therefore, the finally obtained target vector sequence representing the behavior characteristics of the target user can express richer characteristic information, and the prediction capability of the classification prediction service for the target user is improved.

Description

Method and device for extracting user behavior characteristics and classifying and predicting user behavior characteristics
Technical Field
One or more embodiments of the present disclosure relate to the field of machine learning technologies, and in particular, to a method and an apparatus for extracting and classifying and predicting user behavior features.
Background
With the continuous development of internet technology, artificial intelligence technology is more and more applied to the work and life of people, and a great deal of convenience is provided for the work and life of people. Since the classification prediction service using the artificial intelligence technology largely depends on the extraction of features, it is very important how to extract features containing rich information.
Disclosure of Invention
In order to solve one of the above technical problems, one or more embodiments of the present specification provide a method and an apparatus for extracting and classifying and predicting user behavior characteristics.
According to a first aspect, there is provided a method for extracting user behavior features, including:
acquiring behavior information of a target user; the behavior information comprises behavior categories and respective occurrence moments of each behavior in a plurality of behaviors of the target user in a preset time period;
determining each embedded vector corresponding to each behavior based on the behavior information to obtain an initial vector sequence formed by each embedded vector;
processing the initial vector sequence by using a target model to obtain a target vector sequence representing the behavior characteristics of the target user; the target vector sequence is used for classified predictive traffic involving the target user.
Optionally, an embedding vector corresponding to any one of the behaviors is determined based on the behavior information in the following manner:
acquiring a category vector preset for a behavior category to which the behavior belongs;
determining a correction coefficient corresponding to the behavior based on the behavior type to which the behavior belongs and the occurrence time;
and correcting the category vector by using the correction coefficient to obtain an embedded vector corresponding to the behavior.
Optionally, the determining, based on the behavior category to which the behavior belongs and the occurrence time, a correction coefficient corresponding to the behavior includes:
acquiring a first correction vector and a second correction vector which are set according to the behavior category to which the behavior belongs;
determining a time factor corresponding to the behavior; the time factor is positively correlated with a target time difference; the target time difference is the time difference between the moment when the action occurs and the starting moment of the preset time period;
and determining a correction coefficient corresponding to the action based on the first correction vector, the second correction vector and the time factor.
Optionally, the determining a correction coefficient corresponding to the behavior based on the first correction vector, the second correction vector, and the time factor includes:
calculating a product of the first correction vector and the time factor;
inputting the sum of the product and the second correction vector as an argument to an activation function;
and determining the output result of the activation function as a correction coefficient corresponding to the behavior.
Optionally, the classification and prediction service related to the target user includes any one or more of the following:
a service for pre-estimating preset information related to the target user;
recommending information service to the target user;
providing planned services for the target user;
the service for classifying the target user;
and distributing service of service resources for the target user.
Optionally, the target model includes any one of: a recurrent neural network RNN, a long-short term memory neural network LSTM, and a gated recurrent unit GRU.
According to a second aspect, there is provided a method of classification prediction, comprising:
acquiring a target vector sequence representing the behavior characteristics of a target user; the target vector sequence is determined by the method of any one of the first aspect;
performing a classified predictive service involving the target user based on the target vector sequence.
Optionally, the classification and prediction service related to the target user includes any one or more of the following:
a service for pre-estimating preset information related to the target user;
recommending information service to the target user;
providing planned services for the target user;
the service for classifying the target user;
and distributing service of service resources for the target user.
According to a third aspect, there is provided an apparatus for extracting user behavior characteristics, comprising:
the acquisition module is used for acquiring the behavior information of the target user; the behavior information comprises behavior categories and respective occurrence moments of each behavior in a plurality of behaviors of the target user in a preset time period;
a determining module, configured to determine, based on the behavior information, each embedded vector corresponding to each behavior, and obtain an initial vector sequence formed by each embedded vector;
the processing module is used for processing the initial vector sequence by using a target model to obtain a target vector sequence representing the behavior characteristics of the target user; the target vector sequence is used for classified predictive traffic involving the target user.
According to a fourth aspect, there is provided an apparatus for class prediction, comprising:
the acquisition module is used for acquiring a target vector sequence representing the behavior characteristics of a target user; the target vector sequence is determined by the apparatus of the third aspect;
and the service module is used for executing classification prediction service related to the target user based on the target vector sequence.
According to a fifth aspect, there is provided a computer readable storage medium, storing a computer program which, when executed by a processor, implements the method of any of the first or second aspects described above.
According to a sixth aspect, there is provided an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the method of any of the first or second aspects when executing the program.
The technical scheme provided by the embodiment of the specification can have the following beneficial effects:
according to the method and the device for extracting the user behavior characteristics, the behavior information of the target user is obtained, each embedded vector corresponding to each behavior is determined based on the behavior information, an initial vector sequence formed by each embedded vector is obtained, and the initial vector sequence is processed by using the target model, so that a target vector sequence representing the behavior characteristics of the target user is obtained. In the embodiment, the influence of the moment of the behavior of the target user on the behavior characteristics is considered when the behavior characteristics of the target user are extracted, so that the finally obtained target vector sequence representing the behavior characteristics of the target user can express richer characteristic information, and the prediction capability of the classification prediction service for the target user is improved.
The method and the device for classified prediction provided by the embodiments of the present specification perform classified prediction services related to target users by obtaining a target vector sequence characterizing behavior characteristics of the target users and executing the classified prediction services based on the target vector sequence. Therefore, the prediction capability of the classified prediction service for the target user is improved, and the classified prediction service for the target user is more accurate and more reasonable.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the application.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
FIG. 1 is a flow chart illustrating a method for extracting user behavior features according to an exemplary embodiment;
FIG. 2 is a flow diagram illustrating another method for extracting user behavior features according to an exemplary embodiment of the present description;
FIG. 3 is a flow diagram illustrating a method of class prediction according to an exemplary embodiment of the present description;
FIG. 4 is a block diagram of an apparatus for extracting user behavior characteristics, according to an example embodiment;
FIG. 5 is a block diagram illustrating an apparatus for classification prediction according to an exemplary embodiment of the present specification.
Detailed Description
Reference will now be made in detail to the exemplary embodiments, examples of which are illustrated in the accompanying drawings. When the following description refers to the accompanying drawings, like numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present specification. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the specification, as detailed in the appended claims.
The terminology used in the description herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. As used in this specification and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should also be understood that the term "and/or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.
It is to be understood that although the terms first, second, third, etc. may be used herein to describe various information, such information should not be limited to these terms. These terms are only used to distinguish one type of information from another. For example, first information may also be referred to as second information, and similarly, second information may also be referred to as first information, without departing from the scope of the present application. The word "if" as used herein may be interpreted as "at … …" or "when … …" or "in response to a determination", depending on the context.
Fig. 1 is a flow chart illustrating a method for extracting user behavior characteristics according to an exemplary embodiment, which may be applied to any device, platform, server or device cluster having computing and processing capabilities. The method comprises the following steps:
in step 101, behavior information of a target user is acquired.
In this embodiment, the behavior information of the target user may include behavior types to which each behavior of the plurality of behaviors of the target user belongs and respective occurrence times within a preset time period. Specifically, the behavior of the user may be classified in advance to obtain a plurality of behavior categories. For example, the behavior category may include, but is not limited to, clicking on an advertisement, browsing news information, commenting on a message, searching for merchandise, purchasing merchandise, receiving and paying, sending and receiving mail, calling a car for appointment, ordering for take-out, playing audio, playing video, and so forth. Then, information of various operation behaviors performed by the target user through the terminal device within a preset time period can be acquired, so that behavior information of the target user is obtained.
In this embodiment, the preset time period may be any preset time period before the current time, and the duration of the preset time period may be one month, one week, or one day.
In step 102, each of the embedding vectors corresponding to each of the behaviors is determined based on the behavior information, and an initial vector sequence including each of the embedding vectors is obtained.
In this embodiment, each of the embedding vectors corresponding to each of the behaviors may be determined based on the behavior information, and an initial vector sequence formed by each of the embedding vectors may be obtained. Specifically, in one implementation, the embedded vector corresponding to any one of the behaviors may be determined based on the behavior information as follows: first, a category vector set in advance for a behavior category to which the behavior belongs is acquired. Then, based on the behavior type to which the behavior belongs and the occurrence time, a correction coefficient corresponding to the behavior is determined. And finally, correcting the category vector by using the correction coefficient to obtain an embedded vector corresponding to the behavior.
In another implementation manner, a category vector set in advance for a behavior category to which the behavior belongs may be obtained, a time dimension is added to the category vector, and a value corresponding to the time dimension is determined based on a time when the behavior occurs, so that an embedded vector corresponding to the behavior is obtained.
It can be understood that each embedding vector corresponding to each behavior may also be determined based on the behavior information in any other reasonable manner, and the embodiment is not limited to the specific manner of determining each embedding vector.
In this embodiment, after obtaining each embedded vector corresponding to each behavior, the embedded vectors corresponding to each behavior may be sorted according to the sequence of the time corresponding to each behavior, so as to obtain an initial vector sequence, where the initial vector sequence is a time sequence.
In step 103, the initial vector sequence is processed by using the target model to obtain a target vector sequence representing the behavior characteristics of the target user.
In this embodiment, the initial vector sequence may be input to the target model for processing, so as to obtain a target vector sequence representing behavior characteristics of the target user. Wherein the target model may comprise an encoder and a decoder, and the target model may be any one of the following types of models: RNN (Recurrent Neural Network), LSTM (Long Short-Term Memory Neural Network), GRU (Gated Recurrent Unit).
In this embodiment, the target vector sequence may be used for a classified prediction service related to the target user, and the classified prediction service related to the target user may include any one or more of the following: a service for pre-estimating preset information related to a target user; recommending information service to a target user; providing planned services for the target user; classifying the target user; and distributing service of service resources for the target user.
For example, the services for predicting the preset information related to the target user may include, but are not limited to, click through rate prediction, consumption prediction, and the like. The service of recommending information to the target user may include, but is not limited to, advertisement push, friend recommendation, coupon push, and the like. Services that provide planning for a target user may include, but are not limited to, trip planning, entertainment play planning, and the like. The business of classifying the target user may include, but is not limited to, circle people, match with passengers, and the like. The traffic to allocate service resources to the target users may include, but is not limited to, allocating drivers, allocating dispatchers, and the like.
It can be understood that the classification predicted service related to the target user may not be any other reasonable classification predicted service, and the embodiment does not limit the specific form of the classification predicted service.
It should be noted that the above steps may be performed in the application stage, for example, after the training of the target model is completed, the above steps 101 to 103 may be performed, so as to obtain the target vector sequence of the target user. The resulting target vector sequence may be stored or sent to other devices for classified predictive traffic involving the target user based on the target vector sequence.
The above steps may also be performed in a training phase of the target model, for example, when the target model is trained, the above steps 101 to 103 may be performed, and after the target vector sequence is obtained, the target vector sequence is input into the softmax classifier, so that a loss is obtained. Then, the parameters of the target model are adjusted based on the loss, and the above steps 101 to 103 are re-executed until the condition for stopping training is satisfied. Therefore, the present embodiment does not limit the stage of performing the above steps.
For the embodiment, a specific application scenario may be that, in the intelligent supply chain integrated service system, behavior information of the target user may be collected through the intelligent supply chain management platform, for example, information of the target user clicking an advertisement, information of browsing news information, information of commenting a message, information of searching for a commodity, and the like is collected. Then, based on the behavior information of the target user, the method for extracting the user behavior features provided by the embodiment is applied to obtain a target vector sequence representing the behavior features of the target user. And the target vector sequence is applied to carry out classification prediction on the target user, and according to the result of the classification prediction, the service which is more in line with the user requirement is provided for the target user.
The present embodiment is not limited to the above application scenario, and may be applied to other scenarios. In the method for extracting user behavior features provided in the above embodiments of the present specification, by obtaining behavior information of a target user, determining each embedded vector corresponding to each behavior based on the behavior information, obtaining an initial vector sequence formed by each embedded vector, and processing the initial vector sequence by using a target model, obtaining a target vector sequence representing behavior features of the target user. In the embodiment, the influence of the moment of the behavior of the target user on the behavior characteristics is considered when the behavior characteristics of the target user are extracted, so that the finally obtained target vector sequence representing the behavior characteristics of the target user can express richer characteristic information, and the prediction capability of the classification prediction service for the target user is improved.
Fig. 2 is a flow chart illustrating another method for extracting user behavior characteristics according to an exemplary embodiment, which describes a process of determining an embedded vector corresponding to any behavior, and is applicable to any device, platform, server or device cluster with computing and processing capabilities. The method comprises the following steps:
in step 201, a category vector set in advance for a behavior category to which the behavior belongs is acquired.
In this embodiment, a corresponding category vector is set in advance for each divided behavior category to distinguish different behavior categories. For any behavior, a category vector set in advance for the behavior category to which the behavior belongs may be acquired. For example, the behavior categories may include click, browse, buy, and taxi. Then a category vector (1, 0, 0, 0) may be set for clicks, a category vector (0, 1, 0, 0) for browsing, a category vector (0, 0, 1, 0) for purchases, and a category vector (0, 0, 0, 1) for taxi-taking. If the behavior category to which one behavior belongs is purchase, a category vector (0, 0, 1, 0) may be acquired for purchase setting.
In step 202, a correction coefficient corresponding to the behavior is determined based on the behavior type to which the behavior belongs and the occurrence time.
In this embodiment, the correction coefficient corresponding to the behavior may be determined based on the behavior type to which the behavior belongs and the occurrence time. Specifically, first, a first correction vector and a second correction vector set for a behavior class to which the behavior belongs may be acquired. In this case, the first correction vector and the second correction vector may be set in advance for each of the divided behavior categories. The first correction vector and the second correction vector are both randomly set vectors, and the values of the first correction vector and the second correction vector on all dimensions meet normal distribution. And the first correction vector and the second correction vector have the same dimension.
Then, a time factor corresponding to the behavior is determined, the time factor is positively correlated with a target time difference, and the target time difference is the time difference between the moment when the behavior occurs and the starting moment of the preset time period. For example, in one implementation, let t be the time at which the action occurs1The starting time of the preset time interval is t0If the target time difference is Δ t ═ t1-t0. Assuming that the duration of the preset time period is T and the time factor corresponding to the behavior is τ, τ may be Δ T/T.
Finally, a correction factor corresponding to the behavior may be determined based on the first correction vector, the second correction vector, and a time factor. Specifically, the product of the first correction vector and the time factor may be calculated, and the sum of the product and the second correction vector may be input to the Sigmoid activation function as an argument. And determining the output result of the Sigmoid activation function as a correction coefficient corresponding to the behavior.
For example, in one implementation, the time factor corresponding to the behavior is τ, and the first correction vector and the second correction vector that are set in advance for the behavior class to which the behavior belongs are respectively τ
Figure BDA0003031918270000101
And
Figure BDA0003031918270000102
if the Sigmoid activation function is represented by y ═ Sigmoid (x) (x is an argument of the Sigmoid activation function, and y is a variable of the Sigmoid activation function), the correction coefficient δ corresponding to the behavior can be represented as:
Figure BDA0003031918270000103
in step 203, the class vector is modified by using the modification coefficient to obtain an embedded vector corresponding to the behavior.
In this embodiment, the class vector may be modified by using the modification coefficient to obtain an embedded vector corresponding to the behavior. For example, the modification coefficient is multiplied by the category vector, so as to modify the category vector, and obtain the embedded vector corresponding to the behavior.
For example, in one implementation, the category vector corresponding to the behavior category to which the behavior belongs is set as
Figure BDA0003031918270000104
Correction coefficient corresponding to the behavior
Figure BDA0003031918270000105
Then the corresponding embedded vector for that behavior
Figure BDA0003031918270000106
In the method for extracting user behavior features provided in the above embodiments of the present specification, a category vector set in advance for a behavior category to which a behavior belongs is obtained, a correction coefficient corresponding to the behavior is determined based on the behavior category to which the behavior belongs and a time of occurrence, and the category vector is corrected by using the correction coefficient, so that an embedded vector corresponding to the behavior is obtained. Therefore, the finally obtained target vector sequence representing the behavior characteristics of the target user can express richer characteristic information, and the prediction capability of the classification prediction service for the target user is improved.
FIG. 3 is a flow diagram illustrating a method of class prediction, as shown in FIG. 3, which may be applied to any computing, processing capable device, platform, server, or cluster of devices, according to an example embodiment. The method comprises the following steps:
in step 301, a target vector sequence characterizing the behavior of a target user is obtained.
In this embodiment, a target vector sequence representing behavior characteristics of a target user obtained by the method in the embodiment of fig. 1 or fig. 2 may be obtained. For example, the target vector sequence transmitted by another device may be received, or the target vector sequence may be read from a storage device. It can be understood that the target vector sequence may be obtained in any reasonable manner, and the embodiment is not limited in the specific manner of obtaining the target vector sequence representing the behavior characteristics of the target user.
In step 302, a classified prediction service involving a target user is performed based on a sequence of target vectors.
In this embodiment, a classified prediction service involving a target user may be performed based on a target vector sequence. For example, a sequence of target vectors may be input into one or more models trained in advance to perform a class prediction business involving a target user.
In this embodiment, the target vector sequence may be used for a classified prediction service related to the target user, and the classified prediction service related to the target user may include any one or more of the following: a service for pre-estimating preset information related to a target user; recommending information service to a target user; providing planned services for the target user; classifying the target user; and distributing service of service resources for the target user.
For example, the services for predicting the preset information related to the target user may include, but are not limited to, click through rate prediction, consumption prediction, and the like. The service of recommending information to the target user may include, but is not limited to, advertisement push, friend recommendation, coupon push, and the like. Services that provide planning for a target user may include, but are not limited to, trip planning, entertainment play planning, and the like. The business of classifying the target user may include, but is not limited to, circle people, match with passengers, and the like. The traffic to allocate service resources to the target users may include, but is not limited to, allocating drivers, allocating dispatchers, and the like.
It can be understood that the classification predicted service related to the target user may not be any other reasonable classification predicted service, and the embodiment does not limit the specific form of the classification predicted service.
In the method for classified prediction provided by the foregoing embodiment of the present specification, a target vector sequence characterizing behavior characteristics of a target user is obtained, and a classified prediction service related to the target user is executed based on the target vector sequence. Therefore, the prediction capability of the classified prediction service for the target user is improved, and the classified prediction service for the target user is more accurate and more reasonable.
It should be noted that although in the above-described embodiments, the operations of the methods of the embodiments of the present specification are described in a particular order, this does not require or imply that these operations must be performed in this particular order, or that all of the illustrated operations must be performed, to achieve desirable results. Rather, the steps depicted in the flowcharts may change the order of execution.
Corresponding to the embodiments of the method for extracting the user behavior features and the method for classifying and predicting, the present specification also provides embodiments of a device for extracting the user behavior features and a device for classifying and predicting.
As shown in fig. 4, fig. 4 is a block diagram of an apparatus for extracting user behavior characteristics, according to an exemplary embodiment, the apparatus may include: an acquisition module 401, a determination module 402 and a processing module 403.
The obtaining module 401 is configured to obtain behavior information of a target user, where the behavior information includes a behavior category to which each behavior of multiple behaviors of the target user belongs and a time when each behavior occurs in a preset time period.
A determining module 402, configured to determine, based on the behavior information, each embedded vector corresponding to each behavior, to obtain an initial vector sequence formed by each embedded vector.
A processing module 403, configured to process the initial vector sequence by using a target model to obtain a target vector sequence representing behavior characteristics of a target user, where the target vector sequence is used for a classified prediction service related to the target user.
In some embodiments, the determining module 402 determines the embedded vector corresponding to any one of the behaviors based on the behavior information by: acquiring a category vector preset for the behavior category to which the behavior belongs, determining a correction coefficient corresponding to the behavior based on the behavior category to which the behavior belongs and the occurrence time, and correcting the category vector by using the correction coefficient to obtain an embedded vector corresponding to the behavior.
In other embodiments, the determining module 402 determines the modification factor corresponding to the behavior based on the behavior category to which the behavior belongs and the occurrence time as follows: the method comprises the steps of obtaining a first correction vector and a second correction vector which are set according to a behavior category to which a behavior belongs, determining a time factor corresponding to the behavior, wherein the time factor is positively correlated with a target time difference, and the target time difference is the time difference between the moment when the behavior occurs and the starting moment of a preset time period. And determining a correction coefficient corresponding to the action based on the first correction vector, the second correction vector and the time factor.
In other embodiments, the determining module 402 determines the correction factor corresponding to the behavior based on the first correction vector, the second correction vector, and the time factor by: and calculating a product of the first correction vector and the time factor, taking the sum of the product and the second correction vector as an argument, inputting the argument into an activation function, and determining the output result of the activation function as a correction coefficient corresponding to the behavior.
In other embodiments, the classification forecast traffic relating to the target user may include any one or more of: a service for pre-estimating preset information related to a target user; recommending information service to a target user; providing planned services for the target user; classifying the target user; and distributing service of service resources for the target user.
In other embodiments, the target model may include any one of: a recurrent neural network RNN, a long-short term memory neural network LSTM, and a gated recurrent unit GRU.
It should be understood that the above-mentioned apparatus may be preset in the computing device, and may also be loaded into the computing device by downloading or the like. Corresponding modules in the device can be matched with modules in the computing equipment to realize the extraction scheme of the user behavior characteristics.
As shown in fig. 5, fig. 5 is a block diagram of an apparatus for class prediction according to an exemplary embodiment, and the apparatus may include: an acquisition module 501 and a service module 502.
The obtaining module 501 is configured to obtain a target vector sequence representing behavior characteristics of a target user, where the target vector sequence is determined by the apparatus in fig. 4.
A traffic module 502 for performing a classified prediction traffic involving the target user based on the target vector sequence.
In some embodiments, the classification prediction traffic relating to the target user may include any one or more of: a service for pre-estimating preset information related to a target user; recommending information service to a target user; providing planned services for the target user; classifying the target user; and distributing service of service resources for the target user.
It should be understood that the above-mentioned apparatus may be preset in the computing device, and may also be loaded into the computing device by downloading or the like. Corresponding modules in the above-described apparatus may cooperate with modules in the computing device to implement a classification prediction scheme.
For the device embodiments, since they substantially correspond to the method embodiments, reference may be made to the partial description of the method embodiments for relevant points. The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of one or more embodiments of the present specification. One of ordinary skill in the art can understand and implement it without inventive effort.
The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for the system embodiment, since it is substantially similar to the method embodiment, the description is simple, and for the relevant points, reference may be made to the partial description of the method embodiment.
The foregoing description has been directed to specific embodiments of this disclosure. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims may be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing may also be possible or may be advantageous.
It will be further appreciated by those of ordinary skill in the art that the elements and algorithm steps of the examples described in connection with the embodiments disclosed herein may be embodied in electronic hardware, computer software, or combinations of both, and that the components and steps of the examples have been described in a functional general in the foregoing description for the purpose of illustrating clearly the interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application. The software modules may reside in Random Access Memory (RAM), memory, Read Only Memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.
The above-mentioned embodiments, objects, technical solutions and advantages of the present application are described in further detail, it should be understood that the above-mentioned embodiments are merely exemplary embodiments of the present application, and are not intended to limit the scope of the present application, and any modifications, equivalent substitutions, improvements and the like made within the spirit and principle of the present application should be included in the scope of the present application.

Claims (7)

1. A method for extracting user behavior features, the method comprising:
acquiring behavior information of a target user; the behavior information comprises behavior categories and respective occurrence moments of each behavior in a plurality of behaviors of the target user in a preset time period;
determining each embedded vector corresponding to each behavior based on the behavior information to obtain an initial vector sequence formed by each embedded vector;
processing the initial vector sequence by using a target model to obtain a target vector sequence representing the behavior characteristics of the target user; the target vector sequence is used for classified prediction service related to the target user;
wherein an embedded vector corresponding to any one of the plurality of behaviors is determined based on the behavior information by:
acquiring a category vector preset for a behavior category to which the behavior belongs;
determining a correction coefficient corresponding to the behavior based on the behavior type to which the behavior belongs and the occurrence time;
correcting the category vector by using the correction coefficient to obtain an embedded vector corresponding to the behavior;
wherein, the determining the correction coefficient corresponding to the behavior based on the behavior category to which the behavior belongs and the occurrence time includes:
acquiring a first correction vector and a second correction vector which are set according to the behavior category to which the behavior belongs;
determining a time factor corresponding to the behavior; the time factor is positively correlated with a target time difference; the target time difference is the time difference between the moment when the action occurs and the starting moment of the preset time period;
determining a correction coefficient corresponding to the behavior based on the first correction vector, the second correction vector and the time factor;
wherein the determining a correction coefficient corresponding to the behavior based on the first correction vector, the second correction vector, and the time factor comprises:
calculating a product of the first correction vector and the time factor;
inputting the sum of the product and the second correction vector as an argument to an activation function;
and determining the output result of the activation function as a correction coefficient corresponding to the behavior.
2. The method of claim 1, wherein the classified predicted traffic related to the target user comprises any one or more of:
a service for pre-estimating preset information related to the target user;
recommending information service to the target user;
providing planned services for the target user;
the service for classifying the target user;
and distributing service of service resources for the target user.
3. The method of claim 1, wherein the object model comprises any one of: a recurrent neural network RNN, a long-short term memory neural network LSTM, and a gated recurrent unit GRU.
4. A method of classification prediction, the method comprising:
acquiring a target vector sequence representing the behavior characteristics of a target user; the target vector sequence is determined by the method of claim 1;
performing a classified predictive service involving the target user based on the target vector sequence.
5. The method of claim 4, wherein the classified predicted traffic related to the target user comprises any one or more of:
a service for pre-estimating preset information related to the target user;
recommending information service to the target user;
providing planned services for the target user;
the service for classifying the target user;
and distributing service of service resources for the target user.
6. An apparatus for extracting user behavior features, the apparatus comprising:
the acquisition module is used for acquiring the behavior information of the target user; the behavior information comprises behavior categories and respective occurrence moments of each behavior in a plurality of behaviors of the target user in a preset time period;
a determining module, configured to determine, based on the behavior information, each embedded vector corresponding to each behavior, and obtain an initial vector sequence formed by each embedded vector;
the processing module is used for processing the initial vector sequence by using a target model to obtain a target vector sequence representing the behavior characteristics of the target user; the target vector sequence is used for classified prediction service related to the target user;
wherein the determining module determines the embedded vector corresponding to any one of the plurality of behaviors based on the behavior information by:
acquiring a category vector preset for a behavior category to which the behavior belongs;
determining a correction coefficient corresponding to the behavior based on the behavior type to which the behavior belongs and the occurrence time;
correcting the category vector by using the correction coefficient to obtain an embedded vector corresponding to the behavior;
the determining module determines a correction coefficient corresponding to the behavior based on the behavior category to which the behavior belongs and the occurrence time in the following manner:
acquiring a first correction vector and a second correction vector which are set according to the behavior category to which the behavior belongs;
determining a time factor corresponding to the behavior; the time factor is positively correlated with a target time difference; the target time difference is the time difference between the moment when the action occurs and the starting moment of the preset time period;
determining a correction coefficient corresponding to the behavior based on the first correction vector, the second correction vector and the time factor;
wherein the determining module determines a correction coefficient corresponding to the behavior based on the first correction vector, the second correction vector, and the time factor by:
calculating a product of the first correction vector and the time factor;
inputting the sum of the product and the second correction vector as an argument to an activation function;
and determining the output result of the activation function as a correction coefficient corresponding to the behavior.
7. An apparatus for classification prediction, the apparatus comprising:
the acquisition module is used for acquiring a target vector sequence representing the behavior characteristics of a target user; the target vector sequence is determined by the apparatus of claim 6;
and the service module is used for executing classification prediction service related to the target user based on the target vector sequence.
CN202110432496.XA 2021-04-21 2021-04-21 Method and device for extracting user behavior characteristics and classifying and predicting user behavior characteristics Active CN113128597B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110432496.XA CN113128597B (en) 2021-04-21 2021-04-21 Method and device for extracting user behavior characteristics and classifying and predicting user behavior characteristics

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110432496.XA CN113128597B (en) 2021-04-21 2021-04-21 Method and device for extracting user behavior characteristics and classifying and predicting user behavior characteristics

Publications (2)

Publication Number Publication Date
CN113128597A CN113128597A (en) 2021-07-16
CN113128597B true CN113128597B (en) 2021-10-08

Family

ID=76778787

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110432496.XA Active CN113128597B (en) 2021-04-21 2021-04-21 Method and device for extracting user behavior characteristics and classifying and predicting user behavior characteristics

Country Status (1)

Country Link
CN (1) CN113128597B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116342193A (en) * 2023-03-29 2023-06-27 上海数禾信息科技有限公司 User classification method, device, computer equipment and medium based on user journey

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106971194A (en) * 2017-02-16 2017-07-21 江苏大学 A kind of driving intention recognition methods based on the double-deck algorithms of improvement HMM and SVM
CN109359137A (en) * 2018-10-16 2019-02-19 大连理工大学 Based on user's growth of Feature Selection and semi-supervised learning portrait construction method
CN111310676A (en) * 2020-02-21 2020-06-19 重庆邮电大学 Video motion recognition method based on CNN-LSTM and attention

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP3627399B1 (en) * 2018-09-19 2024-08-14 Tata Consultancy Services Limited Systems and methods for real time configurable recommendation using user data
CN109684543B (en) * 2018-12-14 2021-04-02 北京百度网讯科技有限公司 User behavior prediction and information delivery method, device, server and storage medium
CN109815980A (en) * 2018-12-18 2019-05-28 北京三快在线科技有限公司 Prediction technique, device, electronic equipment and the readable storage medium storing program for executing of user type
CN111274501B (en) * 2020-02-25 2023-04-18 支付宝(杭州)信息技术有限公司 Method, system and non-transitory storage medium for pushing information
CN112488155A (en) * 2020-11-09 2021-03-12 北京三快在线科技有限公司 User information prediction method, device, equipment and medium

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106971194A (en) * 2017-02-16 2017-07-21 江苏大学 A kind of driving intention recognition methods based on the double-deck algorithms of improvement HMM and SVM
CN109359137A (en) * 2018-10-16 2019-02-19 大连理工大学 Based on user's growth of Feature Selection and semi-supervised learning portrait construction method
CN111310676A (en) * 2020-02-21 2020-06-19 重庆邮电大学 Video motion recognition method based on CNN-LSTM and attention

Also Published As

Publication number Publication date
CN113128597A (en) 2021-07-16

Similar Documents

Publication Publication Date Title
CN109903103B (en) Method and device for recommending articles
CN110880127B (en) Consumption level prediction method and device, electronic equipment and storage medium
CN112528110A (en) Method and device for determining entity service attribute
CN111008335B (en) Information processing method, device, equipment and storage medium
CN110598120A (en) Behavior data based financing recommendation method, device and equipment
CN112232546A (en) Recommendation probability estimation method and device, electronic equipment and storage medium
CN110827138A (en) Push information determining method and device
CN112068962B (en) Cloud rendering resource exchange method based on deep learning
CN111161104A (en) Generation method and device of community user portrait
US20230316106A1 (en) Method and apparatus for training content recommendation model, device, and storage medium
CN112288554A (en) Commodity recommendation method and device, storage medium and electronic device
Montazery et al. Learning user preferences in matching for ridesharing
CN113034168B (en) Content item delivery method, device, computer equipment and storage medium
CN113592593A (en) Training and application method, device, equipment and storage medium of sequence recommendation model
CN113032676A (en) Recommendation method and system based on micro-feedback
CN114581114A (en) Popularization resource allocation method, system, device and storage medium
CN113128597B (en) Method and device for extracting user behavior characteristics and classifying and predicting user behavior characteristics
CN111444930B (en) Method and device for determining prediction effect of two-classification model
CN112269943B (en) Information recommendation system and method
CN112785390B (en) Recommendation processing method, device, terminal equipment and storage medium
CN111831892A (en) Information recommendation method, information recommendation device, server and storage medium
CN111325614B (en) Recommendation method and device of electronic object and electronic equipment
CN117422553A (en) Transaction processing method, device, equipment, medium and product of blockchain network
CN115983902A (en) Information pushing method and system based on user real-time event
CN114463590A (en) Information processing method, apparatus, device, storage medium, and program product

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
GR01 Patent grant
GR01 Patent grant