CN108133390A - For predicting the method and apparatus of user behavior and computing device - Google Patents
For predicting the method and apparatus of user behavior and computing device Download PDFInfo
- Publication number
- CN108133390A CN108133390A CN201711400130.4A CN201711400130A CN108133390A CN 108133390 A CN108133390 A CN 108133390A CN 201711400130 A CN201711400130 A CN 201711400130A CN 108133390 A CN108133390 A CN 108133390A
- Authority
- CN
- China
- Prior art keywords
- user
- behavior
- data
- businessman
- behavioural characteristic
- 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
Links
- 238000000034 method Methods 0.000 title claims abstract description 45
- 230000003542 behavioural effect Effects 0.000 claims abstract description 51
- 238000006243 chemical reaction Methods 0.000 claims abstract description 32
- 230000009471 action Effects 0.000 claims abstract description 15
- 238000000605 extraction Methods 0.000 claims abstract description 12
- 230000006399 behavior Effects 0.000 claims description 131
- 238000012545 processing Methods 0.000 claims description 8
- 238000003066 decision tree Methods 0.000 claims description 5
- 238000004590 computer program Methods 0.000 claims description 4
- 238000007781 pre-processing Methods 0.000 claims description 4
- 238000010586 diagram Methods 0.000 description 12
- 238000012549 training Methods 0.000 description 9
- 230000007306 turnover Effects 0.000 description 7
- 230000007704 transition Effects 0.000 description 5
- 238000004891 communication Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000007796 conventional method Methods 0.000 description 2
- 230000005611 electricity Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000007634 remodeling Methods 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
Classifications
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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)
- Business, Economics & Management (AREA)
- Accounting & Taxation (AREA)
- Data Mining & Analysis (AREA)
- Development Economics (AREA)
- Finance (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Entrepreneurship & Innovation (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Economics (AREA)
- Game Theory and Decision Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The disclosure is directed to a kind of method for predicting user behavior, including:It handles current User action log and obtains the first user data of no goal behavior;Corresponding behavioural characteristic is extracted from first user data;And predict that the corresponding user behavior of first user data is converted to the probability of the goal behavior according to the corresponding behavioural characteristic of extraction and historical data.In accordance with an embodiment of the present disclosure, the goal behavior of user is predicted based on historical data, the precision of prediction user behavior conversion can be improved.
Description
Technical field
This disclosure relates to Internet technical field more particularly to it is a kind of for predict user behavior method and apparatus and
Computing device.
Background technology
With the development of Internet technology, O2O (Online to Offline, under line on line) has been risen as one
The business model that kind is received by consumers in general.The basic business logic of O2O is that the online upper mounting plate browsing of user is even advance
Payment, then to consumption experience under line.For the businessman in this business model, user browses under line from line and consumes
Conversion be a very important data, be collected into complete user's conversion data and its marketing effectiveness tracked for O2O businessmans
Work in every is waited to suffer from greatly helping.However, due to deposited in O2O business models on line under line between information not closed loop
The problem of, cause line upper mounting plate that can not get its all conversion data brought to businessman on line.
The collection of user's conversion data mainly passes through a kind of following method based on experience at present.User is in O2O platform (examples
Such as website or application) access a shops after, if consumer behavior on wired within follow-up a period of time, be counted as once having
It is converted on the line of effect;If without consumer behavior on line in user's follow-up a period of time, but there are other behaviors (for example to click
Comment is made a phone call to businessman), then the coefficient set based on experience can be used, these non-consumption behaviors being collected into platform
Calculate rough turn over number.For example, after user has browsed certain shops, some time on the day of has beaten an electricity to businessman
Words, based on experience setting learn that the probability for having about 10% after phone goes to shops to consume, then the behavior be counted as 0.1 time conversion.Most
Afterwards, the counting in the case of both the above is integrated, obtains final user's conversion data.
However, the above-mentioned user behavior based on experience setting coefficient prediction and practical consumption conversion behavior there may be compared with
Big difference, and the consumption transition probability under varying environment is not accounted for using fixed coefficient yet, therefore often lead
Consuming for family of applying is converted by the estimation of mistake.
Invention content
The purpose of the disclosure is to provide a kind of method and apparatus and computing device for being used to predict user behavior, Jin Erzhi
It is few to overcome the problems, such as caused by the limitation of the relevant technologies and defect one or more to a certain extent.
Other characteristics and advantages of the disclosure will be by the following detailed description apparent from or partially by the disclosure
Practice and acquistion.
According to the embodiment of the present disclosure in a first aspect, provide a kind of method for predicting user behavior, including:Processing is worked as
Preceding User action log obtains the first user data of no goal behavior;Corresponding behavior is extracted from first user data
Feature;And the corresponding user of first user data is predicted according to the corresponding behavioural characteristic of extraction and historical data
Behavior is converted to the probability of the goal behavior.
According to the second aspect of the embodiment of the present disclosure, a kind of device for being used to predict user behavior is provided, including:Pretreatment
Module, is set as handling current User action log and obtains the first user data of no goal behavior;Characteristic extracting module, if
It is set to and corresponding behavioural characteristic is extracted from first user data;And behavior prediction module, it is set as the institute according to extraction
Corresponding behavioural characteristic and historical data are stated to predict that the corresponding user behavior of first user data is converted to the target line
For probability.
According to the third aspect of the embodiment of the present disclosure, a kind of storage medium for being stored with computer program, the meter are provided
Calculation machine program makes the computing device perform the side as described in any of the above embodiment in the processor operation by computing device
Method.
According to the fourth aspect of the embodiment of the present disclosure, a kind of computing device is provided, including:Processor;Memory is stored with
The instruction that can be performed by the processor;Wherein described processor is configured as performing the side as described in any of the above embodiment
Method.
In accordance with an embodiment of the present disclosure, the target line of user is predicted by the nonlinear model trained based on historical data
For the precision that prediction user behavior converts can be improved.
It should be understood that above general description and following detailed description are only exemplary and explanatory, not
The disclosure can be limited.
Description of the drawings
Fig. 1 is the method flow diagram for being used to predict user behavior according to one embodiment of the disclosure.
Fig. 2 is the method flow diagram for being used to predict user behavior according to another embodiment of the disclosure.
Fig. 3 is the method flow diagram for being used to predict user behavior according to disclosure another embodiment.
Fig. 4 is the embodiment flow chart of model training in method shown in Fig. 3.
Fig. 5 is the embodiment flow chart that prediction is converted in method shown in Fig. 4.
Fig. 6 is the device schematic block diagram for being used to predict user behavior according to one embodiment of the disclosure.
Fig. 7 is the device schematic block diagram for being used to predict user behavior according to another embodiment of the disclosure.
Fig. 8 is the computing device schematic block diagram according to one embodiment of the disclosure.
Specific embodiment
The principle and spirit of the invention are described below with reference to several illustrative embodiments.It should be appreciated that provide this
A little embodiments are not with any just for the sake of better understood when those skilled in the art and then realize the present invention
Mode limits the scope of the invention.On the contrary, these embodiments are provided so that the disclosure is more thorough and complete, and energy
It is enough that the scope of the present disclosure is completely communicated to those skilled in the art.
One skilled in the art will appreciate that embodiments of the present invention can be implemented as a kind of system, device, equipment, method
Or computer program product.Therefore, the disclosure can be with specific implementation is as follows, i.e.,:Complete hardware, complete software
The form that (including firmware, resident software, microcode etc.) or hardware and software combine.
According to the embodiment of the present invention, it is proposed that it is a kind of for predict user behavior method and apparatus and medium and
Computing device.
Below with reference to several representative embodiments of the present invention, the principle and spirit of the invention are illustrated in detail.
Fig. 1 is the method flow diagram for being used to predict user behavior according to one embodiment of the disclosure, as shown in the figure, the present embodiment
Method include the following steps S101-S103.In one embodiment, the method for the present embodiment can by with user terminal communication
The computing devices such as server perform.
In step S101, handle current User action log and obtain the first user data of no goal behavior.
Current User action log is data record corresponding with pending all customer data, wherein both having included true
The fixed user data for having converted goal behavior also includes the user data of transition probability to be predicted.This step remembers these data
Record is handled, and therefrom distinguishes the first user data of no goal behavior (transition probability to be predicted) and has goal behavior (really
It is fixed to have converted) second user data, and carry out subsequent prediction processing using the first user data.
In step s 102, corresponding behavioural characteristic is extracted from the first user data.
This step is to extract corresponding behavioural characteristic for currently pending user data.For example, with O2O user
For the application scenarios for consuming conversion behavior prediction, it is assumed that behavioural characteristic includes behavioural information, user property, businessman's attribute on line
And the relationship information between user and businessman, then this step completion extract corresponding line uplink from the first user data
The features such as the relationship information between information, user property, businessman's attribute and user and businessman.
In step s 103, predict that the first user data is corresponding according to the correspondence behavioural characteristic of extraction and historical data
User behavior is converted to the probability of goal behavior.
In this step based on historical data to the correspondence behavioural characteristic that step S103 is extracted from the first user data into
Row processing, you can prediction obtains the probability that the corresponding user behavior of the first user data is converted to goal behavior.
In accordance with an embodiment of the present disclosure, the goal behavior of user is predicted based on historical data, prediction user can be improved
The precision of behavior conversion.
Fig. 2 is the method flow diagram for being used to predict user behavior according to another embodiment of the disclosure, as shown in the figure, this implementation
The method of example includes the following steps S201-S204.In one embodiment, the method for the present embodiment can by with user terminal communication
The computing devices such as server perform.
In step s 201, it is fitted to obtain behavioural characteristic and the non-linear relation model of goal behavior based on historical data.
It is different using fixed experience setting coefficient to carry out predictive behavior from conventional method, passes through historical data in the present embodiment
Training obtains the non-linear relation model between behavioural characteristic and goal behavior, is based further on non-linear relation model combination
The behavioural characteristic of subsequent extracted is predicted, so as to obtain the higher user's behavior prediction result of accuracy.
In one embodiment, goal behavior refers to the conversion results for the user behavior that needs are predicted.Energy in historical data
It enough determines to be converted into goal behavior known to which user behavior, once the behavioural characteristic for choosing user behavior, history can be based on
Data are fitted to obtain the non-linear relation model between these behavioural characteristics and goal behavior.
In step S202, handle current User action log and obtain the first user data of no goal behavior.
This step corresponds to the step S101 of Fig. 1 embodiments, and details are not described herein again.
In step S203, corresponding behavioural characteristic is extracted from the first user data.
Step S201 is the non-linear relation mould between the behavioural characteristic and goal behavior chosen based on historical data
Type, this step are to extract corresponding behavioural characteristic for currently pending user data.For example, with O2O customer consumptions
For the application scenarios of conversion behavior prediction, it is assumed that the behavioural characteristic chosen during step S201 training non-linear relation models includes
Relationship information on line between behavioural information, user property, businessman's attribute and user and businessman, then this step complete from the
The pass between behavioural information on corresponding line, user property, businessman's attribute and user and businessman is extracted in one user data
The features such as connection property information.
In step S204, according to the correspondence behavioural characteristic of extraction, the first user is predicted using non-linear relation model
The corresponding user behavior of data is converted to the probability of goal behavior.
Due to had been based in step S201 historical data train to obtain it is non-linear between behavioural characteristic and goal behavior
Relational model carries out the correspondence behavioural characteristic that step S203 is extracted from the first user data using the model in this step
Processing, you can prediction obtains the probability that the corresponding user behavior of the first user data is converted to goal behavior.
In accordance with an embodiment of the present disclosure, the target line of user is predicted by the nonlinear model trained based on historical data
For the precision that prediction user behavior converts can be improved.
The disclosure is described by taking the application scenarios of O2O customer consumptions conversion behavior prediction as an example below for predicting user behavior
Method another embodiment.
Fig. 3 is the method flow diagram for being used to predict user behavior according to another embodiment of the disclosure.As shown in the figure, this implementation
The method of example includes the following steps S301-S305.In one embodiment, the method for the present embodiment can be by the service of O2O platforms
The computing devices such as device perform.
In step S301, it is fitted to obtain behavioural characteristic and the non-linear relation model of consumer behavior based on historical data.
The present embodiment proposes a kind of scheme based on machine learning prediction customer consumption conversion.The scene consumed on user's line
With the scene consumed under user's line there is many intercommunities, this programme had from line and disappeared by the method for machine learning
Take in those users of behavior, learn to they in behavior, attribute and selection shops preference the characteristics of, so as to predict those
The user that browsed on line does not have consumer behavior on line after shops subsequently makes the possibility of consumer behavior, further reach it is desired by
The extensive purpose to consumption scene under user's line of scene is consumed on user's line.
In the definition of sample, this programme is by all relevant rows of single user shops browsed with it in one day
It is integrated for attribute as a sample.It did in this shops on the day of confirming user if based on historical data and was consumed on line, then
This sample is labeled as positive sample, this sample is just otherwise labeled as negative sample.It should be noted that browsing described here
Shops be not limited to shops's homepage, such as may also include the single group corresponding to shops's commodity.
For above-mentioned behavioural characteristic, the optional behavior for taking family when browsing some shops of one embodiment of this programme,
Four big category feature of relevance between user property, shops's attribute and user and shops, it is described in detail below.
(1) behavior on user's line, such as including user each mould of the page is browsed in residence time of shops's page, user
The number of block, user click number of each button of the page etc..
(2) user property, such as the age including user, gender, place city, level of consumption etc..
(3) shops's attribute, such as the level of aggregation including shops, taste star, evaluation number etc..
(4) relevance between user and shops, for example, including user browse shops when and shops distance, user preference
Commercial circle and the relevance of commercial circle where shops etc..Here relevance is for example including commercial circle where user preference commercial circle and shops
It is no consistent.
For above-mentioned non-linear relation model, GBDT (Gradient Boosting can be used in one embodiment of this programme
Decision Tree, gradient promote decision tree) it is converted to describe customer consumption.GBDT is a kind of point realized based on decision tree
Class regression algorithm, basic principle are through the new CART of training (Classification and Regression
Trees, Taxonomy and distribution) it is fitted the residual error of the conclusion sum of all trees obtained before the previous round iteration, allow sample
Loss as possible small.
In the scene of customer consumption conversion prediction, the present embodiment is mainly used for solving the problems, such as following two using GBDT.
First, in the behavioural characteristic of above-mentioned selection, being associated between user and shops is that (value shows as 0 to discrete feature mostly
Or 1), but user is then continuous in shops's this behavioural characteristic of page stay time, GBDT can handle simultaneously continuous type and
The feature of discrete type, without carrying out further artificial treatment to feature.Secondly, GBDT is a nonlinear mould
Type can be fitted the non-linear relation that above-mentioned behavioural characteristic consumes conversion with actual user very well;For example, user clicks the page very
Less but its possibility very little consumed is not indicated that, user is very long in shops's page residence time, but does not represent the possibility of its consumption
Property is very big;Linear model can not describe this relationship, and GBDT can be very good to solve the problems, such as this.
In addition, the businessman that enters of O2O platforms generally includes consumption and two kinds of support simultaneously under consumption, Support Line under Support Line
Three classifications of consumption pattern.In one embodiment, in order to make training sample purer, this programme is chosen as far as possible in sampling
On those Support Lines consumption or line on consumption proportion bigger shops.
Further, since positive sample ratio is much smaller than negative sample in real data, for improvement, in one embodiment
In, this programme can carry out repeated sampling to positive sample, so as to which positive and negative sample proportion be allowed to tend to be balanced, during ensureing model learning not
Negative sample can be excessively biased to.
In one embodiment, step S301 can be based on shown in Fig. 4 step S401-S406 is implemented.
In step S401, existing user property and shops's attribute are obtained from external data source;
In step S402, the relevance of user property and shops's attribute is handled, obtains Relating Characteristic;
In step S403, user behaviors log of the website with obtaining user in client log from line;
In step s 404, User action log is handled, calculates the behavioural characteristics such as stay time, while determines user at this
Whether shops has consumption to convert;
In step S405, merge regular all behavioural characteristics, distinguishing has consumption to convert and without the user's row for consuming conversion
For to generate data sample;
In step S406, GBDT models are obtained by repetitive exercise.
By above step, the non-linear relation mould between the behavioural characteristic of selection and consumption conversion behavior is finally obtained
Type.
In step s 302, handle current User action log obtain without consumption conversion behavior the first user data and
There are the second user data of goal behavior.
If only needing the consumption transition probability of prediction user, similar step S202 only need to be from current user behavior day
The first user data without consumption conversion behavior is distinguished in will.Disappear in this example, it is assumed that needing to count final user
Take turn over number, then perform step S302 and the first number of users without consumption conversion behavior is distinguished from current User action log
According to the second user data that have goal behavior, wherein the first user data is for obtaining the turn over number of subsequent prediction part, the
Two user data then can be used for directly obtaining the turn over number of determining part.
In step S303, corresponding behavioural characteristic is extracted from the first user data.
In step s 304, according to the correspondence behavioural characteristic of extraction, the first number of users is predicted using non-linear relation model
The probability of consumer behavior is converted to according to corresponding user behavior.
Step S303-S304 corresponds respectively to the step S203-S204 of embodiment illustrated in fig. 2.In one embodiment, it walks
The conversion prediction that rapid S303-S304 is carried out can be implemented based on step S501-S504 shown in fig. 5.
In step S501, user behaviors log of the website with obtaining user in client log from line;
In step S502, User action log is handled, the user shops pair for distinguishing wired upper conversion above converts with wireless
User shops pair;
In step S503, for the wirelessly upper user shops pair converted, obtain user behavior, user property and shops and belong to
Property, generate behavioural characteristic;
In step S504, the non-linear relation model of training handles above-mentioned behavioural characteristic before use, wireless to predict
The weak conversion results of the user shops pair of lower conversion.
By above step, the user behavior finally obtained corresponding to the first user data without consumption conversion behavior is final
The probability of consumer behavior may be converted to.
In step S305, the probability for counting second user data and above-mentioned prediction obtains all user behaviors to the mesh
The turn over number of mark behavior.
For example, for some shops, it is assumed that obtain having determined consumption conversion row based on the second user data of step S302
For turn over number be 16, obtaining the consumption transition probability of three users by step S303-S304 based on the first user data is
0.1st, 0.6,0.3, then it is 16+0.1 that step S305, which can finally show that current O2O platforms contribute number for the consumption conversion of the shops,
+ 0.6+0.3=17.It should be noted that being only the simple examples of Data Integration statistics herein, the disclosure is not limited to that.
It should be noted that although describing each step of method in the disclosure with particular order in the accompanying drawings,
This, which does not require that or implies, to perform these steps according to the particular order or have to carry out step shown in whole
It could realize desired result.Additional or alternative, it is convenient to omit multiple steps are merged into a step and held by certain steps
It goes and/or a step is decomposed into execution of multiple steps etc..In addition, being also easy to understand, these steps can be
Such as it is either synchronously or asynchronously performed in multiple module/process/threads.
A kind of device for being used to predict user behavior is further provided in this example embodiment.
Fig. 6 is the device schematic block diagram for being used to predict user behavior according to one embodiment of the disclosure.As shown in the figure, this implementation
The device of example includes preprocessing module 61, characteristic extracting module 62 and behavior prediction module 63.
In one embodiment, preprocessing module 61, which is set as handling current User action log, obtains no goal behavior
The first user data;Characteristic extracting module 62 is set as extracting corresponding behavioural characteristic from first user data;Behavior
Prediction module 63 is set as predicting first user data pair according to the corresponding behavioural characteristic and historical data of extraction
The user behavior answered is converted to the probability of the goal behavior.
In one embodiment, the goal behavior includes consumer consumption behavior, and the behavioural characteristic includes behavior on line
Relationship information between information, user property, businessman's attribute and user and businessman.
In one embodiment, behavior information includes following at least one on the line:User is in businessman's corresponding page
Residence time, user browse the number that the number of each module and user in the page click each button in the page.
In one embodiment, the relationship information between the user and businessman includes following at least one:User exists
Distance when browsing businessman's corresponding page with the businessman, the relevance of user preference commercial circle and commercial circle where the businessman.
In accordance with an embodiment of the present disclosure, the goal behavior of user is predicted based on historical data, prediction user can be improved
The precision of behavior conversion.
Fig. 7 is the device schematic block diagram for being used to predict user behavior according to another embodiment of the disclosure.As shown in the figure, scheming
On the basis of 6 illustrated embodiments, the device of the present embodiment further includes model training module 64 and conversion statistical module 65.
In one embodiment, model training module 64 is set as being fitted to obtain behavioural characteristic and target based on historical data
The non-linear relation model of behavior.Correspondingly, in the present embodiment, behavior prediction module 63 is set as the correspondence according to extraction
Behavioural characteristic, the non-linear relation model obtained using model training module 64 predict the corresponding use of first user data
Family behavior is converted to the probability of the goal behavior.
In another embodiment, preprocessing module 61, which is also configured to handle current User action log, has obtained target
The second user data of behavior.Correspondingly, conversion statistical module 65 is set as counting above-mentioned second user data and behavior prediction
The probability that module 63 is predicted obtains all user behaviors to the turn over number of goal behavior.
In one embodiment, the goal behavior includes consumer consumption behavior, and the behavioural characteristic includes behavior on line
Relationship information between information, user property, businessman's attribute and user and businessman.
In one embodiment, behavior information includes following at least one on the line:User is in businessman's corresponding page
Residence time, user browse the number that the number of each module and user in the page click each button in the page.
In one embodiment, the relationship information between the user and businessman includes following at least one:User exists
Distance when browsing businessman's corresponding page with the businessman, the relevance of user preference commercial circle and commercial circle where the businessman.
In one embodiment, the non-linear relation model promotes decision tree GBDT models for gradient.
In accordance with an embodiment of the present disclosure, the target line of user is predicted by the nonlinear model trained based on historical data
For the precision that prediction user behavior converts can be improved.
About the device in above-described embodiment, wherein modules perform the concrete mode of operation in related this method
Embodiment in be described in detail, explanation will be not set forth in detail herein.
It should be noted that although several modules or list for acting the equipment performed are referred in above-detailed
Member, but this division is not enforceable.In fact, according to embodiment of the present disclosure, it is above-described two or more
The feature and function of module either unit can embody in a module or unit.A conversely, above-described mould
Either the feature and function of unit can be further divided into being embodied by multiple modules or unit block.As module or list
The component of member display may or may not be physical unit, you can be located at a place or can also be distributed to
In multiple network element.Some or all of module therein can be selected according to the actual needs to realize disclosure scheme
Purpose.Those of ordinary skill in the art are without creative efforts, you can to understand and implement.
By the description of embodiment of above, those skilled in the art is it can be readily appreciated that example described above embodiment party
Formula can be realized by software, can also be realized in a manner that software is with reference to necessary hardware.
For example, in an example embodiment, a kind of computer readable storage medium is also provided, is stored thereon with calculating
Machine program, the step of method described in any one above-mentioned embodiment can be realized when which is executed by processor.The side
The specific steps of method can refer to the detailed description in previous embodiment, and details are not described herein again.The computer readable storage medium
Can be ROM, random access memory (RAM), CD-ROM, tape, floppy disk and optical data storage devices etc..
In another example embodiment, a kind of computing device is also provided, which can be mobile phone, tablet electricity
The terminal devices such as the mobile terminals such as brain or desktop computer, server do not limit this in this example embodiment
System.Fig. 8 is shown according to the schematic diagram of computing device 80 a kind of in disclosure example embodiment.For example, computing device 80 can be with
It is provided as the server with user terminal communication.With reference to Fig. 8, equipment 80 includes processing component 81, further comprises one
Or multiple processors and as the memory resource representated by memory 82, it can be by the execution of processing component 81 for storing
Instruction, such as application program.The application program stored in memory 82 can include each one or more correspondence
In the module of one group of instruction.It is above-mentioned for predicting user behavior to perform in addition, processing component 81 is configured as execute instruction
Method.The step of this method, can refer to the detailed description in preceding method embodiment, and details are not described herein again.
Computing device 80 can also include a power supply module 83 be configured as perform computing device 80 power management, one
A wired or wireless network interface 84 is configured as computing device 80 being connected to network and input and output (I/O) interface
85.Computing device 80 can be operated based on the operating system for being stored in memory 82.
Those skilled in the art will readily occur to the disclosure its after considering specification and putting into practice invention disclosed herein
Its embodiment.This application is intended to cover any variations, uses, or adaptations of the disclosure, these modifications, purposes or
Person's adaptive change follows the general principle of the disclosure and including the undocumented common knowledge in the art of the disclosure
Or conventional techniques.Description and embodiments are considered only as illustratively, and the true scope and spirit of the disclosure are by appended
Claim is pointed out.
Although exemplary embodiment describing the disclosure with reference to several, it is to be understood that, term used is explanation and shows
Example property, term and not restrictive.The spirit or reality that can be embodied in a variety of forms without departing from application due to the disclosure
Matter, it should therefore be appreciated that above-described embodiment is not limited to any aforementioned details, and should be in the spirit that appended claims are limited
With widely explained in range, therefore the whole variations fallen into claim or its equivalent scope and remodeling all should be accompanying power
Profit requirement is covered.
Claims (10)
1. a kind of method for predicting user behavior, including:
It handles current User action log and obtains the first user data of no goal behavior;
Corresponding behavioural characteristic is extracted from first user data;And
The corresponding user behavior of first user data is predicted according to the corresponding behavioural characteristic of extraction and historical data
It is converted to the probability of the goal behavior.
2. the method as described in claim 1 further includes:
It is fitted to obtain behavioural characteristic and the non-linear relation model of goal behavior based on historical data;
It is described that the corresponding user's row of first user data is predicted according to the corresponding behavioural characteristic and historical data of extraction
Probability to be converted to the goal behavior includes:
According to the corresponding behavioural characteristic of extraction, predict that first user data corresponds to using the non-linear relation model
User behavior be converted to the probability of the goal behavior.
3. the method as described in claim 1, wherein the current User action log of the processing further includes to have obtained target line
For second user data, the method further includes:
The probability for counting the second user data and the prediction obtains conversion of all user behaviors to the goal behavior
Number.
4. method as described in any one of claims 1 to 3, wherein the goal behavior includes consumer consumption behavior, the row
It is characterized including the relationship information on line between behavioural information, user property, businessman's attribute and user and businessman.
5. method as claimed in claim 4, wherein behavior information includes following at least one on the line:User is in businessman couple
The residence time of the page, user is answered to browse time that the number of each module and user in the page click each button in the page
Number.
6. method as claimed in claim 4, wherein the relationship information between the user and businessman includes following at least one
Kind:User is associated with when browsing businessman's corresponding page with the distance of the businessman, user preference commercial circle with commercial circle where the businessman
Property.
7. method as claimed in claim 2, wherein the non-linear relation model promotes decision tree GBDT models for gradient.
8. it is a kind of for predicting the device of user behavior, including:
Preprocessing module, is set as handling current User action log and obtains the first user data of no goal behavior;
Characteristic extracting module is set as extracting corresponding behavioural characteristic from first user data;And
Behavior prediction module is set as predicting first user according to the corresponding behavioural characteristic and historical data of extraction
The corresponding user behavior of data is converted to the probability of the goal behavior.
9. a kind of storage medium for being stored with computer program, the computer program is in the processor operation by computing device
When, the computing device is made to perform the method as described in any one of claim 1-7.
10. a kind of computing device, including:
Processor;
Memory is stored with the instruction that can be performed by the processor;
Wherein described processor is configured as performing the method as described in any one of claim 1-7.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711400130.4A CN108133390A (en) | 2017-12-22 | 2017-12-22 | For predicting the method and apparatus of user behavior and computing device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711400130.4A CN108133390A (en) | 2017-12-22 | 2017-12-22 | For predicting the method and apparatus of user behavior and computing device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108133390A true CN108133390A (en) | 2018-06-08 |
Family
ID=62391786
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711400130.4A Pending CN108133390A (en) | 2017-12-22 | 2017-12-22 | For predicting the method and apparatus of user behavior and computing device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108133390A (en) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109284864A (en) * | 2018-09-04 | 2019-01-29 | 广州视源电子科技股份有限公司 | Behavior sequence acquisition methods and device, user's conversion ratio prediction technique and device |
CN109919649A (en) * | 2019-01-11 | 2019-06-21 | 北京腾云天下科技有限公司 | A kind of target user determines method, apparatus, calculates equipment and medium |
CN110046932A (en) * | 2019-03-29 | 2019-07-23 | 阿里巴巴集团控股有限公司 | A kind of data processing method, device and equipment |
CN111523032A (en) * | 2020-04-22 | 2020-08-11 | 贝壳技术有限公司 | Method, device, medium and electronic equipment for determining user preference |
CN111724184A (en) * | 2019-03-20 | 2020-09-29 | 北京嘀嘀无限科技发展有限公司 | Transformation probability prediction method and device |
CN111767196A (en) * | 2019-04-01 | 2020-10-13 | 阿里巴巴集团控股有限公司 | Data processing method and device, electronic equipment and computing equipment |
CN113129080A (en) * | 2021-05-13 | 2021-07-16 | 北京大米科技有限公司 | Data processing method and device |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105868847A (en) * | 2016-03-24 | 2016-08-17 | 车智互联(北京)科技有限公司 | Shopping behavior prediction method and device |
CN106250403A (en) * | 2016-07-19 | 2016-12-21 | 北京奇艺世纪科技有限公司 | Customer loss Forecasting Methodology and device |
CN107369052A (en) * | 2017-08-29 | 2017-11-21 | 北京小度信息科技有限公司 | User's registration behavior prediction method, apparatus and electronic equipment |
-
2017
- 2017-12-22 CN CN201711400130.4A patent/CN108133390A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105868847A (en) * | 2016-03-24 | 2016-08-17 | 车智互联(北京)科技有限公司 | Shopping behavior prediction method and device |
CN106250403A (en) * | 2016-07-19 | 2016-12-21 | 北京奇艺世纪科技有限公司 | Customer loss Forecasting Methodology and device |
CN107369052A (en) * | 2017-08-29 | 2017-11-21 | 北京小度信息科技有限公司 | User's registration behavior prediction method, apparatus and electronic equipment |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109284864A (en) * | 2018-09-04 | 2019-01-29 | 广州视源电子科技股份有限公司 | Behavior sequence acquisition methods and device, user's conversion ratio prediction technique and device |
CN109284864B (en) * | 2018-09-04 | 2021-08-24 | 广州视源电子科技股份有限公司 | Behavior sequence obtaining method and device and user conversion rate prediction method and device |
CN109919649A (en) * | 2019-01-11 | 2019-06-21 | 北京腾云天下科技有限公司 | A kind of target user determines method, apparatus, calculates equipment and medium |
CN111724184A (en) * | 2019-03-20 | 2020-09-29 | 北京嘀嘀无限科技发展有限公司 | Transformation probability prediction method and device |
CN110046932A (en) * | 2019-03-29 | 2019-07-23 | 阿里巴巴集团控股有限公司 | A kind of data processing method, device and equipment |
CN111767196A (en) * | 2019-04-01 | 2020-10-13 | 阿里巴巴集团控股有限公司 | Data processing method and device, electronic equipment and computing equipment |
CN111523032A (en) * | 2020-04-22 | 2020-08-11 | 贝壳技术有限公司 | Method, device, medium and electronic equipment for determining user preference |
CN113129080A (en) * | 2021-05-13 | 2021-07-16 | 北京大米科技有限公司 | Data processing method and device |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108133390A (en) | For predicting the method and apparatus of user behavior and computing device | |
CN109902708A (en) | A kind of recommended models training method and relevant apparatus | |
US10565525B2 (en) | Collaborative filtering method, apparatus, server and storage medium in combination with time factor | |
Cao et al. | Customer demand prediction of service-oriented manufacturing incorporating customer satisfaction | |
CN108052979A (en) | The method, apparatus and equipment merged to model predication value | |
EP4242955A1 (en) | User profile-based object recommendation method and device | |
CN104933622A (en) | Microblog popularity degree prediction method based on user and microblog theme and microblog popularity degree prediction system based on user and microblog theme | |
CN110310163A (en) | A kind of accurate method, equipment and readable medium for formulating marketing strategy | |
CN110910201B (en) | Information recommendation control method and device, computer equipment and storage medium | |
CN111371767A (en) | Malicious account identification method, malicious account identification device, medium and electronic device | |
CN114663198A (en) | Product recommendation method, device and equipment based on user portrait and storage medium | |
CN114997263B (en) | Method, device, equipment and storage medium for analyzing training rate based on machine learning | |
CN113592593A (en) | Training and application method, device, equipment and storage medium of sequence recommendation model | |
CN110866625A (en) | Promotion index information generation method and device | |
CN107644042B (en) | Software program click rate pre-estimation sorting method and server | |
CN109934631A (en) | Question and answer information processing method, device and computer equipment | |
CN111652282B (en) | Big data-based user preference analysis method and device and electronic equipment | |
CN113850669A (en) | User grouping method and device, computer equipment and computer readable storage medium | |
CN117236624A (en) | Issue repairer recommendation method and apparatus based on dynamic graph | |
US20230351153A1 (en) | Knowledge graph reasoning model, system, and reasoning method based on bayesian few-shot learning | |
CN112989174A (en) | Information recommendation method and device, medium and equipment | |
CN113726963B (en) | Intelligent outbound harassment prevention method, device, equipment and medium | |
CN111768218A (en) | Method and device for processing user interaction information | |
CN112200602B (en) | Neural network model training method and device for advertisement recommendation | |
CN115185606A (en) | Method, device, equipment and storage medium for obtaining service configuration parameters |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180608 |