CN104239327B - A kind of mobile Internet user behavior analysis method and device based on positional information - Google Patents
A kind of mobile Internet user behavior analysis method and device based on positional information Download PDFInfo
- Publication number
- CN104239327B CN104239327B CN201310239737.4A CN201310239737A CN104239327B CN 104239327 B CN104239327 B CN 104239327B CN 201310239737 A CN201310239737 A CN 201310239737A CN 104239327 B CN104239327 B CN 104239327B
- Authority
- CN
- China
- Prior art keywords
- theme
- place
- application service
- information
- parameter matrix
- 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
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9537—Spatial or temporal dependent retrieval, e.g. spatiotemporal queries
Abstract
The present invention relates to a kind of mobile Internet user behavior analysis method and device based on positional information, including step：S1, obtain and store the IMEI code of mobile interchange network users, IMSI codes；S2, the historical position information and historical usage information on services for obtaining and storing each identity mobile interchange network users；S3, according to the historical position information initialize place theme, according to the historical usage information on services initialize application service theme；S4, get parms matrix Φ and parameter matrix Β；S5, according to parameter acquiring parameter matrix Θ；S6, undated parameter matrix Φ, Β, Θ；S7, judge whether Φ, Β, Θ restrain, undated parameter matrix Φ, Β, Θ value is until convergence；S8, according to convergency value Βf、Θf、ΦfThe apply service element maximum with the current location information degree of association is obtained, and is pushed to the user.Application service is pushed present invention utilizes the geographical location information of mobile interchange network users, the accuracy of marketing is improved.
Description
【Technical field】
The present invention relates to Data Mining, more particularly to a kind of mobile Internet user behavior point based on positional information
Analyse method and device.
【Background technology】
Nowadays the content type in network service is increasingly enriched covers all trades and professions there is provided the webpage of various information, can
Various application services to operate in mobile-terminal platform are powerful and species is various, and user's choosing is enriched in these application services
While selecting, while also bringing the challenge that information content explodes to user.To enable users to from the application service of numerous and complicated
Selection meets the service of self-demand, and operator wishes to provide personalized service for different users, to realize accurately
Marketing strategy, so that maximum revenue.Therefore need accurately to analyze the behavior pattern of user, know different user group
The behavioral characteristic and its use habit of body.
The method to Internet user's behavioural analysis mainly has behavioural analysis and base based on conventional internet user at present
In the major class of behavioural analysis two of wireless interconnected network users.
Behavioural analysis presence based on conventional internet user can not effectively identify sole user and obtain its personal attribute
And behavioral data, can only to single behavior in itself carry out analyze prediction shortcoming.In addition, some use such as neutral net at present
The user behavior analysis scheme of tagsort algorithm needs to predefine the class of service that user uses, and setting network user makes
Corresponding relation between class of business and networks congestion control classification, this way needs in advance to classify to business,
The automatic identification of class of business can not be accomplished, cost of labor is higher, and increase with user and class of business is constantly soaring,
The scalability of the program is not strong.
Behavioural analysis based on wireless interconnected network users also has many in terms of subscriber segmentation, user characteristics extraction can be with
The space of lifting, is also merely resting on the stage that point heap is carried out according to the statistical indicator of user to the subdivision of client at present, to
The excavation that family pent-up demand is carried out is not careful enough and gos deep into, and seldom combines and uses in particularly current user behavior analysis scheme
The geographical location information at family.Because the behavior of Internet user with the geographical position where the user has great correlation, because
The geographical location information of this user has great significance for user behavior analysis.
【The content of the invention】
Present invention seek to address that above-mentioned problems of the prior art, propose a kind of mobile interchange based on positional information
Network users behavior analysis method and device.
One aspect of the present invention proposes a kind of mobile Internet user behavior analysis method based on positional information, including step
Suddenly：S1, obtain and store the IMEI code of mobile interchange network users, IMSI codes, mobile Internet user identity is identified；
S2, the historical position information and historical usage information on services for obtaining and storing each identity mobile interchange network users, it is described to go through
History positional information includes some place elements and its frequency, the place that the place element representation mobile interchange network users pass through,
The historical usage information on services includes some apply service elements and its frequency, and the apply service element represents mobile interchange
The used application service of network users；S3, according to the historical position information initialize place theme Li(i=1,2 ...), according to
The historical usage information on services initialization application service theme Aj(j=1,2 ...), wherein each place theme LiRepresent
The set of place element described in identical type, each application service theme AjRepresent apply service element described in identical type
Set；S4, the historical usage information on services is sampled, connected applications service theme AjGet parms matrix Φ, and
The historical position information is sampled, with reference to place theme LiGet parms matrix Β, and wherein Φ represents to apply at each
The probability of each apply service element is produced under service theme, Β represents to produce each place element under each place theme
Probability；S5, using Gibbs samplings, according to application service theme Aj, place theme Li, parameter matrix Φ, Β get parms square
Battle array Θ, Θ represents to produce the probability of each place theme under each application service theme；S6, using Gibbs samplings, and base
In Maximum-likelihood estimation criterion, undated parameter matrix Φ, Β, Θ；S7, judge whether parameter matrix Φ, Β, Θ value restrain, if
Otherwise repeat step S4 to S6, undated parameter matrix Φ, Β, Θ value are until convergence；S8, the IMEI for obtaining mobile interchange network users
Code, IMSI codes and its current location information, according to the parameter matrix Β for belonging to the userf、Θf、ΦfObtain and current location
The maximum apply service element of information relevance, and pushed to the user, wherein, Φf、Βf、ΘfRespectively Φ, Β, Θ are received
Optimal value after holding back.
Another aspect of the present invention proposes a kind of mobile Internet user behavior analysis device based on positional information, including deposits
Module, customer attribute information acquisition module, historical information acquisition module, data processing module, application service pushing module are stored up, its
In, the customer attribute information acquisition module obtains the IMEI code of mobile interchange network users, IMSI codes, the memory module storage
The IMEI code, IMSI codes；The historical information acquisition module obtains the history bit confidence of each identity mobile interchange network users
Breath and historical usage information on services, the historical position information include some place elements and its frequency, the place element
The place that mobile interchange network users pass through is represented, the historical usage information on services includes some apply service elements and its frequency
Secondary, the apply service element represents the used application service of mobile interchange network users, is gone through described in the memory module storage
History positional information and historical usage information on services；The data processing module initializes place according to the historical position information
Theme Li(i=1,2 ...), and application service theme A is initialized according to the historical usage information on servicesj(j=1,2 ...), wherein
Each place theme LiThe set of place element described in identical type is represented, each application service theme AjRepresent phase
With the set of apply service element described in species；The data processing module is sampled to the historical usage information on services,
Connected applications service theme AjGot parms matrix Φ, and the historical position information is sampled, with reference to place theme LiObtain
Parameter matrix Β, wherein Φ is taken to represent to produce the probability of each apply service element under each application service theme, Β is represented
The probability of each place element is produced under each place theme；According to application service theme Aj, place theme Li, parameter matrix
Φ, Β, based on Gibbs samplings, the data processing module is got parms matrix Θ, and Θ is represented in each application service theme
The lower probability for producing each place theme；Based on Gibbs samplings and Maximum-likelihood estimation criterion, the data processing module is more
New parameter matrix Φ, Β, Θ value is until convergence, obtains Φ optimal value Φf, Β optimal value ΒfAnd Θ optimal value Θf；
The customer attribute information acquisition module obtains IMEI code, IMSI codes and its current location information of mobile interchange network users, institute
Data processing module is stated according to the parameter matrix Β for belonging to the userf、Θf、ΦfObtain with the current location information degree of association most
Big application service；The application service pushing module pushes the application maximum with the current location information degree of association to the user
Service.
Mobile Internet user behavior analysis method and device proposed by the present invention based on positional information make use of movement
The geographical location information of Internet user, and based on mobile interchange network users geographical location information and its in the geographical position institute
Its application service paid high attention to is pushed to user using the high correlation of application service species, the accuracy of marketing is improved；
The present invention program extracts theme using improved probability topic model is automatic from the ground point set of user and behavior set simultaneously,
And the class of business of user need not be classified in advance, the cost of labor that manual classification is caused is reduced, business is realized
The automatic identification of species, with stronger scalability.
【Brief description of the drawings】
Fig. 1 is the mobile Internet user behavior analysis method flow diagram based on positional information of one embodiment of the invention；
Fig. 2 is the acquisition of one embodiment of the invention application service maximum with the mobile Internet current location information degree of association
The method flow diagram of element；
Fig. 3 is the mobile Internet user behavior analysis structure drawing of device based on positional information of one embodiment of the invention.
【Embodiment】
In order that the objects, technical solutions and advantages of the present invention become apparent from, it is right below in conjunction with specific embodiment and accompanying drawing
The present invention is described in further detail.It should be appreciated that the specific embodiment described in text is only to explain skill of the invention
Art scheme, and it is not construed as limitation of the present invention.
One aspect of the present invention provides a kind of mobile Internet user behavior analysis method based on positional information, such as Fig. 1 institutes
Show, this method comprises the following steps：S1, obtain and store the IMEI code of mobile interchange network users, IMSI codes, to mobile Internet
User identity is identified；S2, the historical position information and history for obtaining and storing each identity mobile interchange network users should
Use information on services；S3, according to the historical position information initialize place theme Li(i=1,2 ...), according to the historical usage
Information on services initialization application service theme Aj(j=1,2,…)；S4, the historical usage information on services is sampled, with reference to
Application service theme AjGot parms matrix Φ, and the historical position information is sampled, and combines place theme LiObtain
Parameter matrix Β；S5, using Gibbs samplings, according to application service theme Aj, place theme Li, parameter matrix Φ, Β obtain
Parameter matrix Θ；S6, using Gibbs samplings, and based on Maximum-likelihood estimation criterion, undated parameter matrix Φ, Β, Θ；S7、
Judge whether parameter matrix Φ, Β, Θ value restrain, if otherwise repeat step S4 to S6, undated parameter matrix Φ, Β, Θ value are straight
To convergence；S8, IMEI code, IMSI codes and its current location information for obtaining mobile interchange network users, according to belonging to the user
Parameter matrix Βf、Θf、ΦfThe apply service element maximum with the current location information degree of association is obtained, and is pushed away to the user
Send, wherein, Φf、Βf、ΘfOptimal value respectively after Φ, Β, Θ convergence.
The technical scheme to the above-mentioned mobile Internet user behavior analysis method based on positional information is made into one below
Step is discussed in detail.
In step sl, the IMEI code of mobile interchange network users is obtained and stored by mobile operator data servers
（International Mobile Equipment Identity, International Mobile Equipment Identity code）, IMSI codes
（International Mobile Subscriber Identity, international mobile subscriber identity）, due to being likely to occur one
The situation of card or a card multimachine more than machine, in the present embodiment, by IMEI code and IMSI codes collectively as mobile interchange network users
Identity.
In step s 2, obtained by mobile operator data servers and store each identity mobile interchange network users
Historical position information and historical usage information on services, the historical position information include some place elements and its frequency, institute
Place or place that place element representation mobile interchange network users pass through in the past period T are stated, for example " first people cure
Institute ", " Suning market ", " Xinhua Bookstore " etc., the frequency herein be the Internet user in the past period T by upper
State the number of times of each place element；The historical usage information on services includes some apply service elements and its frequency, described to answer
The used application service in the past period T of mobile interchange network users, such as " QQ ", " Google are represented with service element
Map ", " masses' comment " etc., the frequency herein is that the Internet user is answered in the past period T using above-mentioned each
With the number of times of service.Preferably, the data sample included for the guarantee historical position information and historical usage information on services
It is sufficiently large, to the value of the time T is with some moons or longer is advisable.
In step s3, place theme L is initialized according to the historical position information obtained in step s 2i(i=1,
2 ...), wherein each place theme LiThe set of place element described in identical type is represented, mobile Internet is used as described
The places such as " First People's Hospital ", " the second the People's Hospital ", " healthcare hospital for women ＆ children " were once gone at family in a period of time T, then place
Some above-mentioned place elements such as theme " hospital " representative " First People's Hospital ", " the second the People's Hospital ", " healthcare hospital for women ＆ children "
Set；Meanwhile, in step s3, application service theme A is initialized according to the historical usage information on servicesj(j=1,2 ...),
Wherein each application service theme AjThe set of apply service element described in identical type is represented, as described mobile Internet
User once used the application services such as " QQ ", " wechat ", " Skype " in a period of time T, then application service theme " chat class "
Represent the set of some above-mentioned apply service elements such as " QQ ", " wechat ", " Skype ".
Preferably, the application service theme Aj(j=1,2 ...) obeys Multinomial (π) distributions, and π is obeyed
Dirichlet (α) is distributed, and α is Dirichlet distributed constants.
In step s 4, the historical usage information on services is sampled, with reference to the application service theme AjObtain
Parameter matrix Φ；The historical position information is sampled simultaneously, with reference to place theme LiGet parms matrix Β, wherein joining
Matrix number Φ represented to produce the probability of each apply service element under each application service theme, and parameter matrix Β is represented each
The probability of each place element is produced under individual place theme.
In step s 5, using Gibbs samplings, according to the application service theme Aj(j=1,2 ...), place theme Li
(i=1,2 ...), parameter matrix Φ, Β get parms matrix Θ, and Θ represents to produce each place under each application service theme
The probability of theme.
In step s 6, using Gibbs samplings, and based on Maximum-likelihood estimation criterion, undated parameter matrix Φ, Β,
Θ；In the step s 7, judge whether parameter matrix Φ, Β, Θ value restrain, if otherwise repeat step S4 to S6, undated parameter square
Battle array Φ, Β, Θ value is until convergence.Gibbs samplings are to calculate LDA（Latent Dirichlet Allocation）Theme mould
A kind of mathematics implementation method of shape parameter, this method passes through DSMC（Monte Carlo method）, accumulated using class
Divide method, by substantial amounts of circulation random sampling, by the use of last result of calculation as prior probability, calculate posteriority again afterwards general
Rate, according to Bayes and statistics correlation theory, when cycle-index is enough by result of calculation approaching to reality value.Gibbs takes out
Sample method is cyclically updated number of times K and can preset, and K values are bigger, parameter matrix optimal value Φf、Βf、ΘfIt is more accurate, in this implementation
In example, defining K value can be set with specific reference to the operational capability of data processor.
In step s 8, IMEI code, IMSI codes and its current location information of mobile interchange network users are obtained, according to belonging to
The parameter matrix Β of the userf、Θf、ΦfObtain the apply service element maximum with the current location information degree of association, and to institute
User's push is stated, wherein, Φf、Βf、ΘfOptimal value respectively after Φ, Β, Θ convergence.Obtain mobile interchange network users
IMEI code, IMSI codes are used for the identity for determining the mobile interchange network users, mutual due to having stored the movement in step sl
The IMEI code of on-line customer, IMSI codes are identified to its identity, therefore in step s 8, according to belonging to the mobile Internet
The parameter matrix Β of userf、Θf、ΦfObtain the application maximum with the mobile Internet user current location information relevance
Service element, and push the application service to the mobile interchange network users.
Preferably, as shown in Fig. 2 step S8 specifically includes following steps：S81, according to the current location information and
Parameter matrix ΒfObtain the place theme L maximum with the current location information degree of associationc；S82, according to parameter matrix ΘfObtain
Take and LcThe maximum application service theme A of the degree of associationc；S83, according to parameter matrix ΦfObtain and AcThe maximum application clothes of the degree of association
Be engaged in element Ec, the apply service element EcIt is i.e. maximum with the mobile Internet user current location information relevance.Specifically
Ground, in step S81, place element and parameter matrix Β in the current location informationfObtain and the place
The maximum place theme L of elements correlation degreec, wherein parameter matrix ΒfRow represent place element, row represent place theme, according to
The place element, from parameter matrix ΒfIt is middle to extract the corresponding column vector of place element, the wherein element representation in column vector
The place element belongs to the probability distribution situation of different location theme, and column vector is sorted, and obtains general belonging to the place element
The maximum place theme L of ratec；In step S82, according to parameter matrix ΘfObtain and LcThe maximum application service theme of the degree of association
Ac, wherein parameter matrix ΘfRow represent application service theme, row represent place theme, according to the place obtained in step S81
Theme Lc, from parameter matrix ΘfMiddle extraction place theme LcElement representation in corresponding row vector, wherein row vector is represented
Place theme LcThe probability distribution situation of correspondence different application service theme, row vector is sorted, place theme L is obtainedcIt is corresponding
The application service theme A of maximum probabilityc；In step S83, according to parameter matrix ΦfObtain and AcThe maximum application clothes of the degree of association
Be engaged in element Ec, the apply service element EcMaximum, the wherein parameter matrix Φ with the user current location information relevancef's
Row represents apply service element, and row represent application service theme, according to the application service theme obtained in step S82, from parameter
Matrix ΦfMiddle extraction application service theme AcElement representation application service theme A in corresponding row vector, wherein row vectorc
Comprising different application service element probability distribution situation, row vector is sorted, the bigger apply service element table of probability
Show in application service theme AcLower user is higher using the possibility of the apply service element, chooses the application service of maximum probability
Element EcAs the application service maximum with the mobile Internet user current location information relevance, and it is mutual to the movement
On-line customer pushes.
Another aspect of the present invention provides a kind of mobile Internet user behavior analysis device based on positional information, such as Fig. 3
Shown, described device includes：Memory module 100, customer attribute information acquisition module 200, historical information acquisition module 300, number
According to processing module 400, application service pushing module 500.
Below by the company between the mobile Internet user behavior analysis device modules based on positional information
Connect relation and operation principle is described in further detail.
The customer attribute information acquisition module 200 is obtained by mobile operator data servers and stores mobile interchange
The IMEI code of network users（International Mobile Equipment Identity, International Mobile Equipment Identity code）、
IMSI codes（International Mobile Subscriber Identity, international mobile subscriber identity）, due to possible
There is the situation of one-telephone multi-card or a card multimachine, in the present embodiment, by IMEI code and IMSI codes collectively as mobile Internet
The identity of user.The memory module 100 stores the IMEI code, IMSI codes, as to the mobile interchange network users
Identity is identified.
The historical information acquisition module 300 is obtained by mobile operator data servers and stores the movement of each identity
The historical position information and historical usage information on services of Internet user, the historical position information includes some place elements
And its frequency, place or place that the place element representation mobile interchange network users pass through in the past period T, for example
" First People's Hospital ", " Suning market ", " Xinhua Bookstore " etc., the frequency herein were the Internet user at one section of past
Between in T by each above-mentioned place element number of times；The historical usage information on services include some apply service elements and its
The frequency, the apply service element represents the used application service in the past period T of mobile interchange network users, for example
" QQ ", " Google Maps ", " masses' comment " etc., the frequency herein is that the Internet user uses in the past period T
The number of times of each above-mentioned application service.Preferably, it is that the guarantee historical position information and historical usage information on services are included
Data sample it is sufficiently large, to the value of the time T is with some moons or longer is advisable.The memory module 100 stores described
Historical position information and historical usage information on services.
The data processing module 400 initializes place theme L according to the historical position informationi(i=1,2 ...), its
In each place theme LiThe set of place element described in identical type is represented, mobile interchange network users are at one section as described
Once the places such as " First People's Hospital ", " the second the People's Hospital ", " healthcare hospital for women ＆ children " were gone in time T, then place theme " doctor
Institute " represents the set of some above-mentioned place elements such as " First People's Hospital ", " the second the People's Hospital ", " healthcare hospital for women ＆ children "；Together
When, the data processing module 400 initializes application service theme A according to the historical usage information on servicesj(j=1,2 ...),
Wherein each application service theme AjThe set of apply service element described in identical type is represented, as described mobile Internet
User once used the application services such as " QQ ", " wechat ", " Skype " in a period of time T, then application service theme " chat class "
Represent the set of some above-mentioned apply service elements such as " QQ ", " wechat ", " Skype ".
Preferably, the application service theme Aj(j=1,2 ...) obeys Multinomial (π) distributions, and π is obeyed
Dirichlet (α) is distributed, and α is Dirichlet distributed constants.
400 pairs of the data processing module historical usage information on services is sampled, with reference to the application service master
Inscribe AjGet parms matrix Φ；The historical position information is sampled simultaneously, with reference to place theme LiGet parms matrix
Β, wherein parameter matrix Φ represent to produce the probability of each apply service element, parameter matrix under each application service theme
Β represents to produce the probability of each place element under each place theme.
According to the application service theme Aj(j=1,2 ...), place theme Li(i=1,2 ...), parameter matrix Φ, Β, and
Based on Gibbs samplings, the data processing module 400 is got parms matrix Θ, and Θ is represented under each application service theme
Produce the probability of each place theme.
Based on Gibbs samplings and Maximum-likelihood estimation criterion, the data processing module 400 updates the parameter matrix
Φ, Β, Θ value are until convergence, obtains Φ optimal value Φf, Β optimal value ΒfAnd Θ optimal value Θf.Gibbs samples
Method is to calculate LDA（Latent Dirichlet Allocation）A kind of mathematics implementation method of topic model parameter, this method
Pass through DSMC（Monte Carlo method）, using class integration method, pass through substantial amounts of circulation random sampling, profit
With last result of calculation as prior probability, posterior probability is calculated again afterwards, according to Bayes and statistics correlation theory,
By result of calculation approaching to reality value when cycle-index is enough.Gibbs samplings are cyclically updated number of times K and can preset, K values
It is bigger, parameter matrix optimal value Φf、Βf、ΘfIt is more accurate, in the present embodiment, can be with specific reference to the data processing module
400 operational capability sets defining K value.
The customer attribute information acquisition module 200 obtains the IMEI codes of mobile interchange network users, IMSI codes and its currently
Positional information, the data processing module 400 is according to the parameter matrix Β for belonging to the userf、Θf、ΦfObtain and present bit
The maximum apply service element of information relevance is put, the application service pushing module 500 pushes the application to the user and taken
Business.The IMEI code of mobile interchange network users, IMSI codes are used for the identity for determining the mobile interchange network users, due to the storage
Module 100, which has stored the IMEI code of the mobile interchange network users, IMSI codes, to be used to be identified user identity, therefore basis
Belong to the parameter matrix Β of the mobile interchange network usersf、Θf、ΦfObtain and believe with the mobile Internet user current location
The maximum apply service element of the degree of association is ceased, the application service pushing module 500 is pushed to the mobile interchange network users should
Application service.
Preferably, the data processing module 400 is according to the current location information and parameter matrix ΒfObtain and institute
State the maximum place theme L of the current location information degree of associationc, according to parameter matrix ΘfObtain and LcThe maximum application clothes of the degree of association
Be engaged in theme Ac, according to parameter matrix ΦfObtain and AcThe maximum apply service element E of the degree of associationc, the apply service element EcI.e.
For the application service maximum with the current location information degree of association.
Specifically, place element and parameter square of the data processing module 400 in the current location information
Battle array ΒfObtain the place theme L maximum with the place elements correlation degreec, wherein parameter matrix ΒfRow represent place element,
Row represent place theme, according to the place element, from parameter matrix ΒfMiddle extraction corresponding column vector of place element, its
The element representation place element in middle column vector belongs to the probability distribution situation of different location theme, and column vector is sorted, obtained
To the place theme L of the maximum probability belonging to the place elementc；The data processing module 400 is according to parameter matrix ΘfObtain
With LcThe maximum application service theme A of the degree of associationc, wherein parameter matrix ΘfRow represent application service theme, row represent place
Theme, according to the place theme L obtained in step S71c, from parameter matrix ΘfMiddle extraction place theme LcCorresponding row to
Amount, wherein the element representation in row vector represents place theme LcThe probability distribution situation of correspondence different application service theme, will
Row vector sorts, and obtains place theme LcThe application service theme A of corresponding maximum probabilityc；The data processing module 400
According to parameter matrix ΦfObtain and AcThe maximum apply service element E of the degree of associationc, the apply service element EcWork as with the user
Front position information relevance is maximum, wherein parameter matrix ΦfRow represent apply service element, row represent application service theme,
According to application service theme Ac, from parameter matrix ΦfMiddle extraction application service theme AcCorresponding row vector, wherein row vector
In element representation application service theme AcComprising different application service element probability distribution situation, row vector is arranged
Sequence, the bigger apply service element of probability is represented in application service theme AcLower user uses the possibility of the apply service element
It is higher, choose the apply service element E of maximum probabilitycAs with the mobile Internet user current location information relevance most
Big application service, the application service pushing module 500 pushes the application service to the mobile interchange network users.
Mobile Internet user behavior analysis method and device proposed by the present invention based on positional information make use of movement
The geographical location information of Internet user, and based on mobile interchange network users geographical location information and its in the geographical position institute
Its application service paid high attention to is pushed to user using the high correlation of application service species, the accuracy of marketing is improved；
The present invention program extracts theme using improved probability topic model is automatic from the ground point set of user and behavior set simultaneously,
And the class of business of user need not be classified in advance, the cost of labor that manual classification is caused is reduced, business is realized
The automatic identification of species, with stronger scalability.
Although the present invention is described with reference to current better embodiment, those skilled in the art should be able to manage
Solution, above-mentioned better embodiment is only used for explanation and illustration technical scheme, and is not used for limiting the guarantor of the present invention
Shield scope, any within the scope of the spirit and principles in the present invention, any modification, equivalence replacement, deformation, improvement for being done etc.,
Within the claims that should be included in the present invention.
Claims (6)
1. a kind of mobile Internet user behavior analysis method based on positional information, comprises the following steps：
S1, obtain and store the IMEI code of mobile interchange network users, IMSI codes, mobile Internet user identity is identified；
S2, the historical position information and historical usage information on services for obtaining and storing each identity mobile interchange network users, institute
Stating historical position information includes some place elements and its frequency, the ground that the place element representation mobile interchange network users pass through
Point, the historical usage information on services includes some apply service elements and its frequency, and the apply service element represents mobile
The used application service of Internet user；
S3, according to the historical position information initialize place theme Li(i=1,2 ...), according to the historical usage information on services
Initialize application service theme Aj(j=1,2 ...), wherein each place theme LiRepresent place element described in identical type
Set, each application service theme AjRepresent the set of apply service element described in identical type；
S4, the historical usage information on services is sampled, connected applications service theme AjGet parms matrix Φ, and to institute
State historical position information to be sampled, with reference to place theme LiGot parms matrix Β, and wherein Φ is represented in each application service
The probability of each apply service element is produced under theme, Β represents to produce the general of each place element under each place theme
Rate；
S5, using Gibbs samplings, according to application service theme Aj, place theme Li, parameter matrix Φ, Β get parms matrix
Θ, Θ represent to produce the probability of each place theme under each application service theme；
S6, using Gibbs samplings, and based on Maximum-likelihood estimation criterion, undated parameter matrix Φ, Β, Θ；
S7, judge whether parameter matrix Φ, Β, Θ value restrain, if otherwise repeat step S4 to S6, undated parameter matrix Φ, Β,
Θ values are until convergence；
S8, IMEI code, IMSI codes and its current location information for obtaining mobile interchange network users, according to the ginseng for belonging to the user
Matrix number Βf、Θf、ΦfThe apply service element maximum with the current location information degree of association is obtained, and is pushed to the user,
Wherein, Φf、Βf、ΘfOptimal value respectively after Φ, Β, Θ convergence.
2. the mobile Internet user behavior analysis method according to claim 1 based on positional information, it is characterised in that
Application service theme AjMultinomial (π) distributions are obeyed, π obeys Dirichlet (α) distributions, and α is that Dirichlet is distributed ginseng
Number.
3. the mobile Internet user behavior analysis method according to claim 1 based on positional information, step S8 includes
Following steps：
S81, according to the current location information and parameter matrix ΒfObtain and current location information degree of association maximum
Place theme Lc；
S82, according to parameter matrix ΘfObtain and LcThe maximum application service theme A of the degree of associationc；
S83, according to parameter matrix ΦfObtain and AcThe maximum apply service element E of the degree of associationc, the apply service element EcWith
The user current location information relevance is maximum.
4. a kind of mobile Internet user behavior analysis device based on positional information, including：Memory module, customer attribute information
Acquisition module, historical information acquisition module, data processing module, application service pushing module, wherein,
The customer attribute information acquisition module obtains the IMEI code of mobile interchange network users, IMSI codes, and the memory module is deposited
Store up the IMEI code, IMSI codes；
The historical information acquisition module obtains the historical position information and historical usage of each identity mobile interchange network users
Information on services, the historical position information includes some place elements and its frequency, the place element representation mobile Internet
The place that user passes through, the historical usage information on services includes some apply service elements and its frequency, the application service
The used application service of element representation mobile interchange network users, the memory module stores the historical position information and gone through
History application service information；
The data processing module initializes place theme L according to the historical position informationi(i=1,2 ...), and according to described
Historical usage information on services initialization application service theme Aj(j=1,2 ...), wherein each place theme LiRepresent identical
The set of place element described in species, each application service theme AjRepresent the collection of apply service element described in identical type
Close；
The data processing module is sampled to the historical usage information on services, connected applications service theme AjGet parms
Matrix Φ, and the historical position information is sampled, with reference to place theme LiGot parms matrix Β, and wherein Φ is represented
The probability of each apply service element is produced under each application service theme, Β represents to produce each ground under each place theme
The probability of point element；
According to application service theme Aj, place theme Li, parameter matrix Φ, Β, based on Gibbs samplings, the data processing mould
Block gets parms matrix Θ, and Θ represents to produce the probability of each place theme under each application service theme；
Based on Gibbs samplings and Maximum-likelihood estimation criterion, data processing module undated parameter matrix Φ, Β, Θ value
Until convergence, obtains Φ optimal value Φf, Β optimal value ΒfAnd Θ optimal value Θf；
The customer attribute information acquisition module obtains IMEI code, IMSI codes and its present bit confidence of mobile interchange network users
Breath, the data processing module is according to the parameter matrix Β for belonging to the userf、Θf、ΦfAcquisition is associated with current location information
Spend maximum application service；
Application service pushing module pushes the application service maximum with the current location information degree of association to the user.
5. the mobile Internet user behavior analysis device according to claim 4 based on positional information, it is characterised in that
Application service theme AjMultinomial (π) distributions are obeyed, π obeys Dirichlet (α) distributions, and α is that Dirichlet is distributed ginseng
Number.
6. the mobile Internet user behavior analysis device according to claim 4 based on positional information, it is characterised in that
The data processing module is according to the current location information and parameter matrix ΒfAcquisition is associated with the current location information
Spend maximum place theme Lc, according to parameter matrix ΘfObtain and LcThe maximum application service theme A of the degree of associationc, according to parameter
Matrix ΦfObtain and AcThe maximum apply service element E of the degree of associationc, the apply service element EcAs and current location information
The maximum application service of the degree of association.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310239737.4A CN104239327B (en) | 2013-06-17 | 2013-06-17 | A kind of mobile Internet user behavior analysis method and device based on positional information |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310239737.4A CN104239327B (en) | 2013-06-17 | 2013-06-17 | A kind of mobile Internet user behavior analysis method and device based on positional information |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104239327A CN104239327A (en) | 2014-12-24 |
| CN104239327B true CN104239327B (en) | 2017-11-07 |
Family
ID=52227416
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310239737.4A Active CN104239327B (en) | 2013-06-17 | 2013-06-17 | A kind of mobile Internet user behavior analysis method and device based on positional information |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104239327B (en) |
Families Citing this family (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106488493B (en) * | 2015-08-24 | 2020-06-02 | 阿里巴巴集团控股有限公司 | Method and device for identifying network hotspot type of user and electronic equipment |
| CN107071181B (en) * | 2017-04-27 | 2020-01-10 | 长沙军鸽软件有限公司 | Method for automatically matching communication contact persons |
| CN107704586B (en) * | 2017-10-09 | 2020-02-07 | 陈包容 | User portrait method, device and system based on user activity address |
| CN107613084B (en) * | 2017-10-09 | 2020-02-07 | 陈包容 | Method, device and system for automatically grouping contacts in address list |
| CN110070371B (en) * | 2017-11-20 | 2022-11-18 | 腾讯科技（深圳）有限公司 | Data prediction model establishing method and equipment, storage medium and server thereof |
| CN111209531B (en) * | 2018-11-21 | 2023-08-08 | 百度在线网络技术（北京）有限公司 | Correlation degree processing method, device and storage medium |
| CN110377521B (en) * | 2019-07-23 | 2022-11-18 | 郑州阿帕斯科技有限公司 | Target object verification method and device |
| CN110852556A (en) * | 2019-09-20 | 2020-02-28 | 国网浙江省电力有限公司 | 95598 automatic order dispatching method for first-aid repair work order |
| CN113946753B (en) * | 2021-10-29 | 2024-04-12 | 平安科技（深圳）有限公司 | Service recommendation method, device, equipment and storage medium based on location fence |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101986298A (en) * | 2010-10-28 | 2011-03-16 | 浙江大学 | Information real-time recommendation method for online forum |
| CN102355623A (en) * | 2011-06-29 | 2012-02-15 | 青岛百灵信息科技有限公司 | System and method for changing applicable desktop theme of mobile terminal according to position of mobile terminal |
| CN102591917A (en) * | 2011-12-16 | 2012-07-18 | 华为技术有限公司 | Data processing method and system and related device |
| CN102902700A (en) * | 2012-04-05 | 2013-01-30 | 中国人民解放军国防科学技术大学 | Online-increment evolution topic model based automatic software classifying method |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2009059480A1 (en) * | 2007-11-08 | 2009-05-14 | Shanghai Hewlett-Packard Co., Ltd | Url and anchor text analysis for focused crawling |
-
2013
- 2013-06-17 CN CN201310239737.4A patent/CN104239327B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101986298A (en) * | 2010-10-28 | 2011-03-16 | 浙江大学 | Information real-time recommendation method for online forum |
| CN102355623A (en) * | 2011-06-29 | 2012-02-15 | 青岛百灵信息科技有限公司 | System and method for changing applicable desktop theme of mobile terminal according to position of mobile terminal |
| CN102591917A (en) * | 2011-12-16 | 2012-07-18 | 华为技术有限公司 | Data processing method and system and related device |
| CN102902700A (en) * | 2012-04-05 | 2013-01-30 | 中国人民解放军国防科学技术大学 | Online-increment evolution topic model based automatic software classifying method |
Non-Patent Citations (2)
| Title |
|---|
| Categorizing software applications for maintenance;MeMillan,C etc;《International Conference on Software Maintenance》;20110930;343-352 * |
| 基于地理位置的社交网络潜在用户和位置推荐模型研究;朱荣鑫;《中国优秀硕士学位论文全文数据库 信息科技辑》;20130615;I139-251 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104239327A (en) | 2014-12-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104239327B (en) | A kind of mobile Internet user behavior analysis method and device based on positional information | |
| Yin et al. | Joint modeling of user check-in behaviors for real-time point-of-interest recommendation | |
| Sobolevsky et al. | General optimization technique for high-quality community detection in complex networks | |
| CN106506705A (en) | Listener clustering method and device based on location-based service | |
| CN104102719B (en) | The method for pushing and device of a kind of trace information | |
| CN104731917B (en) | A kind of recommendation method and device | |
| CN104394118B (en) | A kind of method for identifying ID and system | |
| Chen et al. | Effective and efficient user account linkage across location based social networks | |
| Ahmed et al. | Nested chinese restaurant franchise process: Applications to user tracking and document modeling | |
| CN105045916A (en) | Mobile game recommendation system and recommendation method thereof | |
| CN110191416A (en) | For analyzing the devices, systems, and methods of the movement of target entity | |
| KR102654884B1 (en) | Method for providing interactive information service and apparatus therefor | |
| CN106101222A (en) | The method for pushing of information and device | |
| CN113412608B (en) | Content pushing method and device, server and storage medium | |
| CN104143005A (en) | Related searching system and method | |
| CN107562947A (en) | A kind of Mobile Space-time perceives the lower dynamic method for establishing model of recommendation service immediately | |
| CN105512156B (en) | Click model generation method and device | |
| CN113343127B (en) | Tourism route recommendation method, system, server and storage medium | |
| CN105468598A (en) | Friend recommendation method and device | |
| CN104915354A (en) | Multimedia file pushing method and device | |
| CN106777946B (en) | Personalized health service recommendation method based on hierarchal model | |
| EP3358474A1 (en) | Route search method, device and apparatus, and non-volatile computer storage medium | |
| CN110263840B (en) | Line analysis method, device, program product and storage medium | |
| CN105069077A (en) | Search method and device | |
| CN113011471A (en) | Social group dividing method, social group dividing system and related devices |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |