CN103700018B - A kind of crowd division methods in mobile community network - Google Patents
A kind of crowd division methods in mobile community network Download PDFInfo
- Publication number
- CN103700018B CN103700018B CN201310686372.XA CN201310686372A CN103700018B CN 103700018 B CN103700018 B CN 103700018B CN 201310686372 A CN201310686372 A CN 201310686372A CN 103700018 B CN103700018 B CN 103700018B
- Authority
- CN
- China
- Prior art keywords
- user
- matrix
- mrow
- social networks
- similarity
- 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.)
- Expired - Fee Related
Links
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses crowd's division methods in a kind of mobile community network, comprise the following steps:A, user's geography trace logs file is imported, calculates the discrete geography track similarity s between any two useri,j;B, the discrete geographical track similarity based on any two users, establishes User User similarity matrixs G, Locate User matrix LsUWith User Locate matrix UsL;C, user social contact activity log is imported, social networks model is established according to the handset call historical record between user, calculates social networks adjacency matrix F;D, calculate matrix in block form and divide crowd.The present invention can be applied to individualized content commending system, individualized content pushed to targeted customer by precisely being divided to cell phone network or based on the user in the online social networks in position.
Description
Technical field
The present invention relates to data mining and user behavior analysis technology, and in particular to the crowd in a kind of mobile community network
Division methods.By precisely being divided to cell phone network or based on the user in the online social networks in position, personalization can be applied to
Content recommendation system.
Background technology
Corporations or cluster in accurate division crowd push away for optimizing the individualized contents such as location-based Internet service
The system of recommending is of great significance, and according to crowd's division result, can push individualized content to targeted customer exactly.
The method of existing division crowd can be generally divided into based on community network community detection method and based on data characteristics
Clustering method.
On the one hand, the method for the Detecting Community based on community network according to the tightness degree of the social networks between individual come
Traditional corporations are divided, the social pass inside and outside corporations can be intuitively reflected very much due to the corporations that this method obtains
System, therefore be easy to for analyzing and explaining some universal phenomena, it can very easily be applied to pure social networking service and recommend
Or in other the problem of being closely related with social networks researchs, but crowd might not only have social attribute, other belong to
Property such as geographical attribute, social property etc. equally influence the behavior of individual or even the division of corporations, and then influence its application surface.
On the other hand, many applications have been had in Data Mining based on the data clustering method of common trait.It is this
Method obtains the cluster of the common trait between individual based on the similitude of the attribute of Different Individual dimension, so as to divide difference
The crowd of type.This method is had been widely used in the personalized recommendation field such as e-commerce, content of multimedia push,
But this method specific aim is relatively strong, it is only capable of analyzing and detects behavior of the user in particular community dimension.
The content of the invention
In view of the deficiencies of the prior art, it is an object of the invention to propose the crowd division side in a kind of mobile community network
Method, this method is based on discrete geographical track similitude, more smart in combination with the geographical attribute and social attribute of individual, division
It is accurate.
To realize above goal of the invention, the present invention uses following technical scheme:
A kind of crowd's division methods in mobile community network, comprise the following steps:
A, user's geography trace logs file is imported, calculates the discrete geographical track similarity between any two user
si,j, wherein, give total number of users N, User Activity region is completely covered by M base station, user's geography trace logs by with
The base station sequence number list of family ID and User Activity forms;
B, the discrete geographical track similarity based on any two users, establishes User-User similarity matrixs G, Locate-
User matrix Ls and User-Locate matrix Us, wherein, N*M ties up the matrix element PV of User-Locate matrix UsijRepresent user i
The probability of occurrence of j in position;M*N ties up the matrix element LV of Locate-User matrix LsijRepresent that user j occurs general on the i of position
Rate;N*N ties up User-User similarity matrixs G:
Wherein, each matrix element corresponds to the geographical similarity of the user couple of correspondence position in user list;
C, user social contact activity log is imported, social networks is established according to the handset call historical record between user
Model, calculates social networks adjacency matrix F, wherein, social networks adjacency matrix F ties up matrix, matrix element V for N*NijRepresent to use
The weight of the social networks of family i, j;
D, calculate matrix in block form and divide crowd.
Compared with prior art, the present invention has following technique effect:Due to combine multidimensional data come cluster possess it is similar
The individual of behavior, division are more accurate.
Brief description of the drawings
With reference to following explanation, with reference to attached drawing, there can be optimal understanding to the present invention.In the accompanying drawings, identical part
It can be represented by identical label.
Fig. 1 is crowd's division methods general frame schematic diagram proposed by the present invention;
Fig. 2 is social networks adjacency matrix calculation flow chart;
Fig. 3 is the flow chart of crowd's partiting step.
Embodiment
In order to make the purpose , technical scheme and advantage of the present invention be clearer, below in conjunction with attached drawing and exemplary reality
Example is applied, the present invention will be described in further detail.It should be appreciated that exemplary embodiment described herein is only explaining this
Invention, the scope of application being not intended to limit the present invention.
As shown in Figure 1, crowd's division methods in the mobile community network of the embodiment of the present invention comprise the following steps:
A, user's geography trace logs file is imported, calculates the discrete geographical track similarity between any two user.
Wherein, user's geography trace logs are made of the base station sequence number list of User ID and User Activity.Give total user
Number N, User Activity region are completely covered by M base station, then any user i, and the discrete geographical track similarity between j is expressed as
The sum of cosine similarity index of both all common locations:
Wherein pi,lIt is the probability that user occurs in position l, | | pi,l| | it is pi,lNorm.In the present embodiment, which is
2 norms.
Step a is specially:It is first directed to the history base station sequence number list of any two user, both common base stations of contrast
Quantity and respective ratio is accounted for, calculate the probability of occurrence p of each all shared base station location l of leisurei,l, then calculate corresponding
Cosine similarity index, finally summation obtain the discrete geographical track similarity of any two users.
B, the discrete geographical track similarity based on any two users, establishes User-User similarity matrixs G, Locate-
User matrix LsUWith User-Locate matrix UsL。
Step b is specifically included:According to the discrete geography track similarity s between individuali,jEstablish User-User similitude squares
Battle array G, while establish User-Locate matrix UsLWith Locate-User matrix LsU, utilize symmetrical Algorithms of Non-Negative Matrix Factorization
(SNMF) matrix decomposition is carried out to User-User similarity matrixs G, obtain the Preliminary division of individual geographic track, i.e., it is affiliated each
The degree of adhering to separately of matrix in block form.
Wherein, N*M ties up User-Locate matrix UsLMatrix element PVijRepresent user i in position PVjProbability of occurrence;
M*N ties up Locate-User matrix LsUMatrix element LVijRepresent position LViThe probability that upper user j occurs;N*N ties up User-
User similarity matrixs G is:
Wherein, each matrix element corresponds to the geographical similarity of the user couple of correspondence position in user list.
C, user social contact activity log is imported, social networks is established according to the handset call historical record between user
Model, calculates social networks adjacency matrix F.
Wherein, social networks adjacency matrix F ties up matrix, matrix element V for N*NijRepresent the power of the social networks of user i, j
Weight.In the present embodiment, the weight can be the social interaction frequency of user i, j, i.e. number of communications between user i, j and
The duration of call.
Step c is specially:The doings daily record of N number of user is imported, initializes N*N social networks adjacency matrix F=0,
Corresponding targeted customer in the doings daily record of each user of circulation reading, according to the weight of social networks between individual, renewal
Corresponding matrix element in social networks adjacency matrix, so as to establish the social networks adjacency matrix of colony, as shown in Figure 2.
D, calculate matrix in block form and divide crowd.
Step d is specially:User-User similarity matrix G, User-Locate matrix U in steps for importing b,
Social networks adjacency matrix F in Locate-User matrix Ls and step c, the initial value X of sets target matrix in block form is step b
In to User-User similarity matrixs G carry out matrix decomposition obtained from degree of adhering to separately matrix, using joint Non-negative Matrix Factorization calculate
Method (JNMF) does matrix decomposition, iterates until target segment matrixThe mould of the difference of iteration is less than twice before and after meeting
Predetermined value, exports each matrix in block form and its degree of adhering to separately belonging to each node, that is, obtains the dividing condition of colony, as shown in Figure 3.
According to such scheme, input aforementioned four matrix G, U, L, F do joint Non-negative Matrix Factorization, its target segment matrix
The calculating of (N*K matrixes) is as follows:
In the present embodiment, preset value α=β=0..5, k=1,2 ..., K, K are default geographical corporations number, and scalar 1 is
The unit matrix (i.e. the diagonal element of 2 β L subtracts 1) of corresponding dimension.Required mesh is obtained after the final stabilization that iterates
Mark matrix in block formWherein the i-th row represents degree of membership of the user i for each geographical corporations.
After step d obtains crowd's division result, individualized content commending system can be used according to the crowd of division to target
Family pushes individualized content.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all essences in the present invention
All any modification, equivalent and improvement made within refreshing and principle etc., should all be included in the protection scope of the present invention.
Claims (1)
1. crowd's division methods in a kind of mobile community network, comprise the following steps:
A, user's geography trace logs file is imported, calculates the discrete geography track similarity s between any two useri,j, its
In, give total number of users N, User Activity region is completely covered by M base station, user's geography trace logs by User ID and
The base station sequence number list composition of User Activity;
Specially:Be first directed to the history base station sequence number list of any two user, both common base station numbers of contrast and
Respective ratio is accounted for, calculates the probability of occurrence p of each comfortable shared base station location li,l, then calculate corresponding cosine similarity and refer to
Mark, finally summation obtain the discrete geographical track similarity of any two users, wherein,
Discrete geographical track similarity between any user i, j is expressed as the cosine similarity index of both all common locations
The sum of:
<mrow>
<msub>
<mi>S</mi>
<mrow>
<mi>i</mi>
<mi>j</mi>
</mrow>
</msub>
<mo>=</mo>
<munder>
<mo>&Sigma;</mo>
<mrow>
<mi>l</mi>
<mo>&Element;</mo>
<mi>L</mi>
<mo>,</mo>
<mi>i</mi>
<mo>,</mo>
<mi>j</mi>
<mo>&Element;</mo>
<mi>N</mi>
</mrow>
</munder>
<mfrac>
<mrow>
<msub>
<mi>p</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>l</mi>
</mrow>
</msub>
<msub>
<mi>p</mi>
<mrow>
<mi>j</mi>
<mo>,</mo>
<mi>l</mi>
</mrow>
</msub>
</mrow>
<mrow>
<mo>|</mo>
<mo>|</mo>
<msub>
<mi>p</mi>
<mrow>
<mi>i</mi>
<mo>,</mo>
<mi>l</mi>
</mrow>
</msub>
<mo>|</mo>
<mo>|</mo>
<mo>&times;</mo>
<mo>|</mo>
<mo>|</mo>
<msub>
<mi>p</mi>
<mrow>
<mi>j</mi>
<mo>,</mo>
<mi>l</mi>
</mrow>
</msub>
<mo>|</mo>
<mo>|</mo>
</mrow>
</mfrac>
</mrow>
Wherein, pi,lIt is the probability that user occurs in shared base station location l, | | pi,l| | it is pi,lNorm;
B, the discrete geographical track similarity based on any two users, establishes User-User similarity matrixs G, Locate-User
Matrix LUWith User-Locate matrix UsL, wherein, N*M dimension User-Locate matrix UsLMatrix element PVijRepresent that user i exists
Position PVjProbability of occurrence;M*N ties up Locate-User matrix LsUMatrix element LVijRepresent position LViWhat upper user j occurred
Probability;N*N ties up User-User similarity matrixs G:
Wherein, each matrix element corresponds to the geographical track similarity of the user couple of correspondence position in user list,
Specifically, carrying out matrix decomposition to User-User similarity matrixs G using symmetrical Algorithms of Non-Negative Matrix Factorization, obtain a
The Preliminary division of body geography track, i.e., the degree of adhering to separately of affiliated each matrix in block form;
C, user social contact activity log is imported, social networks mould is established according to the handset call historical record between user
Type, calculates social networks adjacency matrix F, wherein, social networks adjacency matrix F ties up matrix, matrix element V for N*NijRepresent user
The weight of the social networks of i, j;
The user i, the weights of the social networks of j are specially user i, the social interaction frequency of j, i.e. user i, logical between j
Believe number and the duration of call;
Specially:The doings daily record of N number of user is imported, initializes N*N social networks adjacency matrix F=0, circulation is read every
Corresponding targeted customer in the doings daily record of a user, according to the weight of social networks between individual, renewal social networks is adjacent
Corresponding matrix element in matrix is connect, so as to establish the social networks adjacency matrix of colony;
D, calculate matrix in block form and divide crowd, it is specially:User-User similarity matrixs G, User- in steps for importing b
Locate matrix UsL, Locate-User matrix LsUAnd the social networks adjacency matrix F in step c, sets target matrix in block form
Initial value X be step b in User-User similarity matrixs G carry out matrix decomposition obtained from degree of adhering to separately matrix, using joint
Algorithms of Non-Negative Matrix Factorization does matrix decomposition, iterates until target segment matrixThe difference of iteration twice before and after meeting
Mould be less than predetermined value, export each matrix in block form and its degree of adhering to separately belonging to each node, that is, obtain the dividing condition of colony.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310686372.XA CN103700018B (en) | 2013-12-16 | 2013-12-16 | A kind of crowd division methods in mobile community network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310686372.XA CN103700018B (en) | 2013-12-16 | 2013-12-16 | A kind of crowd division methods in mobile community network |
Publications (2)
Publication Number | Publication Date |
---|---|
CN103700018A CN103700018A (en) | 2014-04-02 |
CN103700018B true CN103700018B (en) | 2018-04-24 |
Family
ID=50361537
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201310686372.XA Expired - Fee Related CN103700018B (en) | 2013-12-16 | 2013-12-16 | A kind of crowd division methods in mobile community network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103700018B (en) |
Families Citing this family (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104035978B (en) * | 2014-05-26 | 2017-06-30 | 南京泰锐斯通信科技有限公司 | Combo discovering method and system |
CN104518930B (en) * | 2015-01-09 | 2017-11-21 | 哈尔滨工程大学 | A kind of abnormal user and message Simultaneous Detection towards microblogging |
CN104765873B (en) * | 2015-04-24 | 2019-03-26 | 百度在线网络技术(北京)有限公司 | User's similarity determines method and apparatus |
CN104820905B (en) * | 2015-05-19 | 2018-11-20 | 威海北洋电气集团股份有限公司 | Personnel's management-control method and system based on space tracking big data analysis |
CN104881459A (en) * | 2015-05-22 | 2015-09-02 | 电子科技大学 | Friend recommendation method of mobile social network |
CN107316044A (en) * | 2016-04-27 | 2017-11-03 | 中国电信股份有限公司 | Similar users recognition methods and device |
CN106570764A (en) * | 2016-11-09 | 2017-04-19 | 广州杰赛科技股份有限公司 | User relationship predicting method and device |
CN106600053B (en) * | 2016-12-12 | 2020-04-10 | 西安交通大学 | User attribute prediction system based on space-time trajectory and social network |
CN106506705B (en) * | 2016-12-29 | 2020-07-28 | 平安科技(深圳)有限公司 | Crowd classification method and device based on location service |
CN106846228A (en) * | 2017-02-13 | 2017-06-13 | 浪潮通信信息系统有限公司 | A kind of public safety prewarning method and system based on positional information and social networks |
CN107613520B (en) * | 2017-08-29 | 2020-08-04 | 重庆邮电大学 | Telecommunication user similarity discovery method based on L DA topic model |
CN110750730B (en) * | 2019-09-10 | 2023-04-07 | 合肥工业大学 | Group detection method and system based on space-time constraint |
CN112417076B (en) * | 2020-11-24 | 2022-08-05 | 杭州东信北邮信息技术有限公司 | Building personnel affiliation identification method based on big data mining technology |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102880719A (en) * | 2012-10-16 | 2013-01-16 | 四川大学 | User trajectory similarity mining method for location-based social network |
CN103049496A (en) * | 2012-12-07 | 2013-04-17 | 北京百度网讯科技有限公司 | Method, apparatus and device for dividing multiple users into user groups |
KR101259604B1 (en) * | 2011-11-15 | 2013-05-14 | 경기대학교 산학협력단 | Contents recommendation system and method based on social network |
-
2013
- 2013-12-16 CN CN201310686372.XA patent/CN103700018B/en not_active Expired - Fee Related
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR101259604B1 (en) * | 2011-11-15 | 2013-05-14 | 경기대학교 산학협력단 | Contents recommendation system and method based on social network |
CN102880719A (en) * | 2012-10-16 | 2013-01-16 | 四川大学 | User trajectory similarity mining method for location-based social network |
CN103049496A (en) * | 2012-12-07 | 2013-04-17 | 北京百度网讯科技有限公司 | Method, apparatus and device for dividing multiple users into user groups |
Non-Patent Citations (1)
Title |
---|
Mining user similarity based on location history;Li Quannan .etc;《In Proceedings of the 16th ACM SIGSPATIAL international conference on Advances in Geographic Information Systems》;20081231;摘要,正文第3-7页 * |
Also Published As
Publication number | Publication date |
---|---|
CN103700018A (en) | 2014-04-02 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103700018B (en) | A kind of crowd division methods in mobile community network | |
CN103488678B (en) | Friend recommendation system based on user sign-in similarity | |
Pappalardo et al. | Using big data to study the link between human mobility and socio-economic development | |
CN103617279A (en) | Method for achieving microblog information spreading influence assessment model on basis of Pagerank method | |
Zhang et al. | User community discovery from multi-relational networks | |
CN103678669A (en) | Evaluating system and method for community influence in social network | |
Zhang et al. | Quality-aware user recruitment based on federated learning in mobile crowd sensing | |
CN107341261B (en) | Interest point recommendation method oriented to location social network | |
CN102695121A (en) | Method and system for pushing friend information for user in social network | |
CN104394118A (en) | User identity identification method and system | |
CN102571485A (en) | Method for identifying robot user on micro-blog platform | |
CN107391670A (en) | A kind of mixing recommendation method for merging collaborative filtering and user property filtering | |
CN107203901A (en) | The method and device of product information is pushed to user | |
CN102663047A (en) | Method and device for mining social relationship during mobile reading | |
CN105404943A (en) | User behavior prediction method facing mobile social environment | |
Zhu et al. | Learning geographical and mobility factors for mobile application recommendation | |
Ju et al. | Relationship strength estimation based on Wechat Friends Circle | |
Song et al. | Friendship influence on mobile behavior of location based social network users | |
CN103905507A (en) | Service information recommendation system and service information recommendation method | |
Park et al. | How use of location-based social network (LBSN) services contributes to accumulation of social capital | |
Li | Cost analysis and minimization of movement-based location management schemes in wireless communication networks: a renewal process approach | |
CN101925044A (en) | Method for judging wireless user attributes | |
Ma et al. | Social account linking via weighted bipartite graph matching | |
CN107181672A (en) | The friend recommendation method based on Annual distribution relative entropy in the social networks of position | |
CN111401478A (en) | Data abnormity identification method and device |
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 | ||
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20180424 Termination date: 20181216 |