CN102929942A - Social network overlapping community finding method based on ensemble learning - Google Patents
Social network overlapping community finding method based on ensemble learning Download PDFInfo
- Publication number
- CN102929942A CN102929942A CN2012103718768A CN201210371876A CN102929942A CN 102929942 A CN102929942 A CN 102929942A CN 2012103718768 A CN2012103718768 A CN 2012103718768A CN 201210371876 A CN201210371876 A CN 201210371876A CN 102929942 A CN102929942 A CN 102929942A
- Authority
- CN
- China
- Prior art keywords
- community
- network
- center
- scheme
- overlapping
- 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.)
- Granted
Links
Images
Abstract
The invention provides a social network overlapping community finding method based on ensemble learning and belongs to the technical field of social network. The social network overlapping community finding method comprises the following steps of: firstly, for a sosical network dataset, carrying out community division on a network by utilizing a KASP method to obtain a plurality of different lambda community division candidate schemes; then utilizing a CCChooser choosing method to choose lambda community division candidate schemes to be polymerized from the lambda community division candidate schemes; and finally, carrying out layer soft clustering on the lambda community division candidate schemes to be polymerized and outputting a generation cluster corresponding to the optimal cutting-off point as a final network overlapping community structure. Compared with the network overlapping community finding method based on an individual clustering device, the method disclosed by the invention can find the more effective network overlapping community structure. The social network overlapping community finding method is applied to various social platforms including micro-blog networks, mail networks, BBS (Bulletin Board System) forum networks and the like, and can be used for optimizing an information network structure, improving the information initiative service quality, enhancing the network culture safety, etc.
Description
Technical field
The present invention relates to the overlapping community mining in the information exchange platform under a kind of Web2.0, particularly a kind of overlapping community discovery method that uses the integrated study theory belongs to the community network technical field.
Background technology
Community network (Social Network is called for short SN) is a kind of relational network that is used for representing Social Individual member interactive relationship, extensively is present in human society, and the form of expression of different tissues structures such as family, club, residential quarter, city is arranged.The Web technology of fast development has greatly been enriched the form of expression of community network, and the various social networks such as microblogging network, mail network, BBS forum network emerge in an endless stream.How from these numerous and complicated mixed and disorderly community networks, to find hiding potential valuable community structure pattern, become a popular research direction that attracts vertical many scholars to participate in.
A distinguishing feature of community structure pattern is exactly that the community internal node connects and closely connects loosely between community, and these characteristics induce a large amount of community network community discovery methods.In general, overlappingly community discovery method can be divided into two classes according to whether allowing between the community, first kind method supposition community network individuality only belongs to certain community, utilizes various hard clustering algorithms the community network individuality to be divided into the community of non-overlapping copies.For example, based on the K-Means method of dividing cluster, based on the GN dividing method of limit convergence factor, based on the SM Spectral Clustering of algebraic graph theory, etc.These class methods have been ignored owing to the community network individuality can be under the jurisdiction of community's plyability that a plurality of different communities cause simultaneously, thereby can't find Fiel's plot structure hiding in the community network.For example, the microblogging person can be divided into little group of different communities according to the theme of microblogging microblogging that the person sends out, exist identical microblogging person between different little group of communities.The Equations of The Second Kind method has been eliminated the hypothesis in the first kind method, can find the overlapping community structure of community network.For example, the people such as Palla at first propose to find by rolling K complete graph the CPM method of overlapping community, the people such as Shen Huawei propose to utilize hierarchical clustering thought to realize the EAGLE algorithm of overlapping community discovery, and the people such as Magdon-Ismail propose the overlapping community discovery algorithm SSDE based on spectral clustering thought.Existing algorithm in these class methods has the different defectives such as computation complexity is high, community's quality is on the low side as a result, can not be advantageously applied to the overlapping community mining of actual community network.
In a word, although exist the correlation technique of from community network, finding community in the prior art, but these methods are not the overlapping attributes that can't react community, have exactly the number of drawbacks that affects its practical application, thereby are not suitable for finding from community network overlapping community.
Summary of the invention
The objective of the invention is fast and effeciently to find in order to overcome community discovery method of the prior art the defective of the overlapping community structure of community network, provide a kind of community network based on integrated study overlapping community discovery method.
To achieve these goals, the invention provides the overlapping community discovery method of a kind of community network based on integrated study, be applied to the social networks under the Web2.0, it is characterized in that, described method synthesis integrated study strategy and Spectral Clustering are realized the overlapping community discovery of community network, may further comprise the steps:
A. use quick Spectral Clustering KASP to calculate the Λ kind community splitting scheme that obtains community network;
B. use the CCChooser system of selection to divide to select the candidate scheme from various communities and treat polymerization
Plant community's splitting scheme, wherein
C. the community that treats in the polymerization community splitting scheme carries out the soft cluster of level, exports generation corresponding to optimum truncation points bunch as network overlapped community structure.
The invention has the beneficial effects as follows: compared with traditional community discovery method, the Rational Composition that the method that the present invention proposes can have various communities splitting scheme takes full advantage of and effective integration, can find the community structure that more is consistent with the Web Community real structure.The present invention is applied to various social platforms such as microblogging network, mail network, BBS forum networks, and initiatively service quality, the enhancing Internet culture wait safely can to optimize information network structure, lifting information.
Description of drawings
Fig. 1 is the general flow chart of the overlapping community discovery method of community network based on integrated study of the present invention;
Fig. 2 is the overlapping community structure of the community network karate of the inventive method discovery;
Fig. 3 is the overlapping community structure of the community network dolphins of the inventive method discovery;
Fig. 4 is the overlapping community structure of the community network HLM of the inventive method discovery;
Fig. 5 in size be on 5000 the first kind network scale parameter on the impact of the inventive method validity;
Fig. 6 in size be on 5000 the Equations of The Second Kind network scale parameter on the impact of the inventive method validity;
Fig. 7 in size be on 10000 the first kind network ratio of compression on the impact of the inventive method validity;
Fig. 8 in size be on 10000 the Equations of The Second Kind network ratio of compression on the impact of the inventive method validity.
Embodiment:
Below in conjunction with the drawings and specific embodiments the present invention is explained.
In order conveniently to elaborate the present invention, at first unified explanation relating basic concepts.
Figure: the community network among the present invention represents with graph data structure, and its form is
, V is the node set that consists of network, namely
, | V| represents the number of node, and E is the set of the limit e between the node, namely
,
The measure function of limit e, in order to two node u and the distance of v, the i.e. tightness degree of the relationship of the two of estimating that limit e connects.Figure among the present invention represents with incidence matrix.
Have following step based on the overlapping community discovery method one of the community network of integrated study:
Step 1: call Λ quick spectral clustering KASP same community network is carried out the k division, obtain a splitting scheme that comprises k non-intersect community at every turn.The quick spectral clustering flow process of KASP is as follows:
1) node set of community network is carried out the K-Means cluster, obtain k community center's set
, to all nodes
Set up as follows node to the mapping table of community center: node with
Corresponding community center is
, wherein
, the expression node
With community center
Between Euclidean distance;
2) community center is gathered
Carry out the SM cluster and obtain the cluster of community center;
3) the community center's cluster that obtains to mapping table and the SM clustering method of community center according to node is returned the community network node bunch, forms a splitting scheme that comprises k non-intersect community.
K-Means clustering method flow process is as follows:
1) random initializtion k
Dimensional vector is as community center
2) to each node among the community network figure
, calculate successively it and all community centers
Distance
3) each node is joined community with its nearest community center representative, thereby obtain community's splitting scheme of community network
4) upgrade each community center
5) repeat 2), 3) with 4) until the node in each community no longer change.
SM clustering method flow process is as follows:
1) sets up the similarity matrix S of community center and diagonal matrix D thereof, wherein
Expression community center
Similarity,
,
2) the variant Laplacian matrix of the structure similarity matrix S of community center
, and compute matrix
Proper vector;
3) select maximum k eigenwert characteristic of correspondence vector to construct lower dimensional space as column vector
4) i the element of community center being concentrated
The i that corresponds among the U is capable
(i=1 ..., | V|);
5) will with the K-Means clustering method
(i=1 ..., | V|) poly-for k bunch.
Step 2: use the CCChooser system of selection to divide to select the candidate scheme from various communities and treat polymerization
Plant community's splitting scheme, wherein
CCChooser system of selection flow process is as follows:
1) calculates each community and divide candidate scheme
Representativeness
, select to have community's splitting scheme of maximum ANMI as the current splitting scheme C* of community, wherein
Expression community divides candidate scheme
Y iWith
Y jThe standardization mutual information;
2) community organization that all communities is divided in the candidate scheme becomes community's set and gives the wherein random initial score V of community;
3) the following process of iteration, until the scoring convergence: the i of community in community's set is carried out threshold value is
Random chance select, if do not choose, then from community's set, choose the community with maximum scores V, and with this community current community splitting scheme voted, form new community's splitting scheme
, and recomputate the ANMI of this scheme, and be designated as ANMI_new, calculate the repayment R=ANMI_new-ANMI that chooses current community, upgrade the ANMI=ANMI_new of this scheme, upgrade the scoring of this community
4) calculate all communities divide candidate schemes with
NMI, and divide candidate scheme according to NMI descending sort community, before therefrom selecting
Individual forecast scheme configuration is treated polymerization
Plant community's splitting scheme
Step 3: treat all communities that community's splitting scheme of polymerization comprises and carry out the soft cluster of level, produce the tree construction of clustering cluster, its calculation process is:
1) calculate generation and treat that polymerization community splitting scheme comprises intercommunal similarity matrix L,
, wherein
Expression belongs to community
And belong to
Nodes,
Expression neither belongs to community
Do not belong to community yet
Nodes,
Expression belongs to community
But community not
Nodes,
Represent not community
But belong to community
Nodes;
2) each community is initialized as one bunch, the following operation of iteration is until all communities all are integrated into one bunch: find two similarity maximums bunch and with the two be merged into a high level bunch, any bunch
With bunch
Between calculating formula of similarity be
, wherein
3) the corresponding network overlapped community structure PT of the T layer of cluster result genealogical tree calculated it and block fitness
, choose optimum truncation points
Network overlapped community structure corresponding to (LEVEL is the height of tree of cluster result genealogical tree) exported as net result.
Performance evaluating:
Experiment of the present invention is with 3 live network (Karate networks, Dolphins network and HLM network) and 3 artificial network (SynNet_1, SynNet_2 and SynNet_3) be data set, from Cluster Validity and two aspects of robustness algorithm is estimated.
Clustering Validity Analysis
Cluster Validity for experimental evaluation algorithm SCEA, we introduce NMI (Normalized Mutual Information) evaluation criterion, and the structural similarity of the community structure of finding by comparing cell Fiel plot structure and SCEA algorithm is come the checking of implementation algorithm validity.
We utilize the SCEA algorithm successively network Karate, Dolphins, HLM, SynNet_1, SynNet_2 and SynNet_3 to be carried out community's plyability analysis, and the overlapping community structure of its result as shown in Figure 1.For the Karate network, only having node 3 in its overlapping community structure (Fig. 2) is the node of two communities, and other nodes all belong in its true community exactly, the NMI of this community structure=0.903; For the Dolphins network, only having node 8 and 40 in its overlapping community structure (Fig. 3) is that two communities share, and other nodes all belong in its true community exactly, at this moment the NMI of community structure=0.824; Can find from the final community (Fig. 4) of HLM network: the shared node " history marquis " of the Shi Fu of community and Rong Guofu, there are shared node " You Erjie " in Ningguo mansion community and flourish state mansion community, " Jia Yuan " and " merchant drills ", node " grandmother Liu " is shared by mansion of a prince community and flourish state mansion community, " Wang Xifeng " and " Wangfu people ", node " Xing Xiuyan " and " Xue Baochai " are shared by Xue mansion community and flourish state mansion community, node " aunt Xue " is shared by Xue mansion community and mansion of a prince community. can find out from these shared nodes, the Four Great families in the A Dream of Red Mansions mainly are deep-rooted with the marriage connection through one's female relatives and a group that is difficult to cut apart that form: " You Erjie " marries " Jia Lian ", " Wang Xifeng " marries " Jia Lian ", " Wangfu people " marries " Jia Zheng ", " aunt Xue " marries " princes and dukes' sons ", " Xing Xiuyan " marries " Xue Ke ", also have history marquis's daughter " merchant is female " to marry " Jia Daishan ", certainly also have brotherhood " merchant drills " and " Jia Yuan ", what is interesting is that " grandmother Liu " becomes a shared node. this moment community structure NMI=0.861; For artificial network SynNet_1, SynNet_2 and SynNet_3, the shared node number that its corresponding overlapping community structure contains is respectively 28,25,22, and its corresponding NMI is respectively 0.863,0.884,0.892.
For the validity of evaluation algorithms better, we compare the SCEA algorithm on 6 data sets with algorithm CPM, Link, COPRA, SSDE, as can be seen from Table 1, no matter be live network or artificial network, the community structure degree consistent with true community that the SCEA algorithm calculates gained is higher than other algorithms far away.
Table 1. algorithm CPM, Link, COPRA, the Cluster Validity of SSDE and SCEA are relatively
The algorithm robust analysis
Because scale parameter plays very important effect in traditional spectral clustering, a little different scale parameter value can cause cluster result far from each other, owing to from real data, being difficult to obtain the priori that relevant scale parameter is chosen, this has greatly limited the practical application of spectral clustering, and whether the SCEA as the spectral clustering Integrated Algorithm is faced with same problem so.For probing into this problem, we produce the artificial network that two classes have the different topology feature by the Adoption Network Data Generator, the first kind is the network that mixing constant (the total limit of limit number/network number between community) progressively increases progressively change, and Equations of The Second Kind is the network that overlapping nodes ratio (overlapping nodes number/network node sum) progressively increases progressively change.From Fig. 5 and Fig. 6 as can be known, in the different network of each mixing constant (overlapping nodes ratio) value, the NMI of the as a result community value of SCEA does not progressively increase progressively along with scale parameter and certain linear or nonlinear change occurs, but constant value, although mixing constant (overlapping nodes ratio) has such impact to SCEA validity: less mixing constant (overlapping nodes ratio) can cause larger NMI value, between the corresponding NMI of each mixing constant (overlapping nodes ratio) is in 0.65 to 0.85 in so better interval.This shows, with regard to scale parameter, SCEA has very strong robustness in the network overlapped community of excavation.To it is worthy of note in addition, mixing constant and overlapping nodes ratio on the impact of SCEA can be according to it the two definition make such explanation: mixing constant is larger, namely the limit number is larger between community, this means that mixed-media network modules mixed-media is lower, and the overlapping nodes ratio is larger, overlapping degree is higher between community, boundary is fuzzyyer between this meaning Web Community, no matter be low modularity or high ambiguity all can increase the difficulty of community mining problem, thereby can reduce the as a result validity of community of mining algorithm.
The time efficiency of algorithm SCEA and ratio of compression parameter have closely and contact, what kind of relation object is the validity of ratio of compression and SCEA have again like the analysis of scale parameter so, we utilize the network maker to generate 5 first kind networks and 5 Equations of The Second Kind networks, number of network node all is 10000, can find out from experimental result Fig. 7 and Fig. 8, in the different network of each mixing constant (overlapping nodes ratio) value, the NMI of the as a result community value of SCEA does not significantly reduce along with the increase of ratio of compression, but the fluctuation of small amplitude appears, is in 0.7 to 0.88 such interval.This shows, with regard to ratio of compression, SCEA has very strong robustness in the network overlapped community of excavation.What is interesting is especially, the variation of ratio of compression can be eliminated the impact of mixing constant (overlapping nodes ratio) within the specific limits, for example, in Fig. 7, when ratio of compression is 4, mixing constant 0.4 is the same with 0.5 corresponding NMI, and when ratio of compression was 8, the NMI of mixing constant 0.2 correspondence became 5 kinds of best case in the value condition; Similar phenomenon also is present among Fig. 8, and when ratio of compression was 4, the NMI of overlapping nodes ratio 0.1 correspondence became 5 kinds of best case in the value condition, and when ratio of compression was 8, the NMI value that overlapping nodes ratio 0.2 is corresponding with 0.25 equated.The above only is preferred embodiment of the present invention, and all equalizations of doing according to the present patent application claim change and modify, and all should belong to covering scope of the present invention.
Claims (6)
1. the overlapping community discovery method of the community network based on integrated study is applied to the social networks under the Web2.0, it is characterized in that, described method synthesis integrated study strategy and Spectral Clustering are realized the overlapping community discovery of community network, may further comprise the steps:
A. use quick Spectral Clustering KASP to calculate the Λ kind community splitting scheme that obtains community network;
B. use the CCChooser system of selection to divide to select the candidate scheme from various communities and treat polymerization
Plant community's splitting scheme, wherein
C. the community that treats in the polymerization community splitting scheme carries out the soft cluster of level, exports generation corresponding to optimum truncation points bunch as network overlapped community structure.
2. the overlapping community discovery method of the community network based on integrated study as claimed in claim 1 is characterized in that, the KASP clustering method flow process in the described steps A is as follows:
Step 21: the node set to community network is carried out the K-Means cluster, obtains k community center's set
, to all nodes
Set up as follows node to the mapping table of community center: node with
Corresponding community center is
, wherein
, the expression node
With community center
Between Euclidean distance;
Step 22: community center is gathered
Carry out the SM cluster and obtain the cluster of community center;
Step 23: the community center's cluster that obtains to mapping table and the SM clustering method of community center according to node is returned the community network node bunch, forms a splitting scheme that comprises k non-intersect community.
3. the overlapping community discovery method of the community network based on integrated study as claimed in claim 2 is characterized in that, described SM clustering method flow process is as follows:
Step 31: set up the similarity matrix S of community center and diagonal matrix D thereof, wherein
Expression community center
Similarity,
,
Step 32: the variant Laplacian matrix that makes up the similarity matrix S of community center
, and compute matrix
Proper vector;
Step 33: select maximum k eigenwert characteristic of correspondence vector to construct lower dimensional space as column vector
Step 34: i the element that community center is concentrated
The i that corresponds among the U is capable
(i=1 ..., | V|);
Step 35: will with the K-Means clustering method
(i=1 ..., | V|) poly-for k bunch.
4. the overlapping community discovery method of the community network based on integrated study as claimed in claim 3 is characterized in that, described K-Means clustering method flow process is as follows:
Step 41: random initializtion k
Dimensional vector is as community center
Step 42: to each node among the community network figure
, calculate successively it and all community centers
Distance
Step 43: each node is joined community with its nearest community center representative, thereby obtain community's splitting scheme of community network
Step 44: upgrade each community center
Step 45: repeat 42,43 and 44 until the node in each community no longer change.
5. the overlapping community discovery method of the community network based on integrated study as claimed in claim 1 is characterized in that, the CCChooser system of selection flow process among the described step B is as follows:
Step 51: calculate each community and divide candidate scheme
Representativeness
, select to have community's splitting scheme of maximum ANMI as the current splitting scheme C* of community, wherein
Expression community divides candidate scheme
Y iWith
Y jThe standardization mutual information;
Step 52: the community organization that all communities are divided in the candidate scheme becomes community's set and gives the wherein random initial score V of community;
Step 53: the following process of iteration, until the scoring convergence: the i of community in community's set is carried out threshold value is
Random chance select, if do not choose, then from community's set, choose the community with maximum scores V, and with this community current community splitting scheme voted, form new community's splitting scheme
, and recomputate the ANMI of this scheme, and be designated as ANMI_new, calculate the repayment R=ANMI_new-ANMI that chooses current community, upgrade the ANMI=ANMI_new of this scheme, upgrade the scoring of this community
Step 54: calculate all communities divide candidate schemes with
NMI, and divide candidate scheme according to NMI descending sort community, before therefrom selecting
Individual forecast scheme configuration is treated polymerization
Plant community's splitting scheme.
6. the overlapping community discovery method of the community network based on integrated study as claimed in claim 1 is characterized in that, the implementation method among the described step C is:
Step 61: calculate to generate and treat that polymerization community splitting scheme comprises intercommunal similarity matrix L,
, wherein
Expression belongs to community
And belong to
Nodes,
Expression neither belongs to community
Do not belong to community yet
Nodes,
Expression belongs to community
But community not
Nodes,
Represent not community
But belong to community
Nodes;
Step 62: each community is initialized as one bunch, and the following operation of iteration is until all communities all are integrated into one bunch: find two similarity maximums bunch and with the two be merged into a high level bunch, any bunch
With bunch
Between calculating formula of similarity be
, wherein
Step 63: the corresponding network overlapped community structure PT of the T layer of cluster result genealogical tree calculated it block fitness
, choose optimum truncation points
Corresponding network overlapped community structure is exported as net result, and wherein, LEVEL is the height of tree of cluster result genealogical tree.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210371876.8A CN102929942B (en) | 2012-09-27 | 2012-09-27 | The overlapping community discovery method of a kind of community network based on integrated study |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210371876.8A CN102929942B (en) | 2012-09-27 | 2012-09-27 | The overlapping community discovery method of a kind of community network based on integrated study |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102929942A true CN102929942A (en) | 2013-02-13 |
| CN102929942B CN102929942B (en) | 2015-08-12 |
Family
ID=47644740
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210371876.8A Expired - Fee Related CN102929942B (en) | 2012-09-27 | 2012-09-27 | The overlapping community discovery method of a kind of community network based on integrated study |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102929942B (en) |
Cited By (28)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103425737A (en) * | 2013-07-03 | 2013-12-04 | 西安理工大学 | Overlapping community discovery method for network |
| CN103455612A (en) * | 2013-09-07 | 2013-12-18 | 西安电子科技大学 | Method for detecting non-overlapping network communities and overlapping network communities based on two-stage strategy |
| CN103838804A (en) * | 2013-05-09 | 2014-06-04 | 电子科技大学 | Social network user interest association rule mining method based on community division |
| CN103838803A (en) * | 2013-04-28 | 2014-06-04 | 电子科技大学 | Social network community discovery method based on node Jaccard similarity |
| CN103888541A (en) * | 2014-04-01 | 2014-06-25 | 中国矿业大学 | Method and system for discovering cells fused with topology potential and spectral clustering |
| CN104850728A (en) * | 2015-02-10 | 2015-08-19 | 湖南大学 | Social network degree analysis method based on mean field theory and spreading coefficient |
| CN105282011A (en) * | 2015-09-30 | 2016-01-27 | 广东工业大学 | Social group finding method based on cluster fusion algorithm |
| CN105678626A (en) * | 2015-12-30 | 2016-06-15 | 南京理工大学 | Overlapped community excavation method and apparatus |
| CN103793489B (en) * | 2014-01-16 | 2017-01-18 | 西北工业大学 | Method for discovering topics of communities in on-line social network |
| CN103678669B (en) * | 2013-12-25 | 2017-02-08 | 福州大学 | Evaluating system and method for community influence in social network |
| CN106780053A (en) * | 2015-11-20 | 2017-05-31 | 香港中文大学深圳研究院 | A kind of overlap community discovery method and system based on node connection preference |
| CN106909619A (en) * | 2017-01-16 | 2017-06-30 | 中国科学院声学研究所 | It is a kind of based on offset adjusted and the mixing social networks clustering method bidded and system |
| CN107153713A (en) * | 2017-05-27 | 2017-09-12 | 合肥工业大学 | Overlapping community detection method and system based on similitude between node in social networks |
| CN107577727A (en) * | 2017-08-22 | 2018-01-12 | 武汉大学 | A kind of One-male unit behavioral trait analysis method |
| CN107705213A (en) * | 2017-07-17 | 2018-02-16 | 西安电子科技大学 | A kind of overlapping Combo discovering method of static social networks |
| CN107844939A (en) * | 2017-11-28 | 2018-03-27 | 广州真知码信息科技有限公司 | Sampling estimation cargo numbering method |
| CN108600013A (en) * | 2018-04-26 | 2018-09-28 | 北京邮电大学 | The overlapping community discovery method and device of dynamic network |
| CN108833158A (en) * | 2018-06-08 | 2018-11-16 | 成都理工大学 | A kind of similitude community discovery method based on k-means |
| CN108846543A (en) * | 2018-04-26 | 2018-11-20 | 深圳大学 | A kind of calculation method and device of non-overlap community set quality Measure Indexes |
| CN108898264A (en) * | 2018-04-26 | 2018-11-27 | 深圳大学 | A kind of calculation method and device being overlapped community's set quality Measure Indexes |
| CN109325814A (en) * | 2017-07-31 | 2019-02-12 | 上海诺悦智能科技有限公司 | A method of for finding suspicious trade network |
| CN110442800A (en) * | 2019-07-22 | 2019-11-12 | 哈尔滨工程大学 | A kind of semi-supervised community discovery method of aggregators attribute and graph structure |
| CN110535681A (en) * | 2019-07-18 | 2019-12-03 | 西安电子科技大学 | Based on the parallel overlapping community detection method of open operation language OpenCL |
| CN110910262A (en) * | 2019-10-31 | 2020-03-24 | 北京达佳互联信息技术有限公司 | Community grouping method and device, content isolation method and device and server |
| CN111008338A (en) * | 2019-11-11 | 2020-04-14 | 重庆邮电大学 | Social network multi-scale structure mining method and medium based on hierarchical clustering |
| CN111427999A (en) * | 2020-03-19 | 2020-07-17 | 中国科学院自动化研究所 | Theme real-time influence evaluation method and system for comprehensive integration discussion environment |
| CN112269922A (en) * | 2020-10-14 | 2021-01-26 | 西华大学 | Community public opinion key character discovery method based on network representation learning |
| CN112384920A (en) * | 2018-07-11 | 2021-02-19 | 维萨国际服务协会 | Privacy preserving graph compression with automatic fuzzy variable detection |
-
2012
- 2012-09-27 CN CN201210371876.8A patent/CN102929942B/en not_active Expired - Fee Related
Non-Patent Citations (4)
| Title |
|---|
| LU ZHUBING等: "An Overview on Overlapping Community Detection", 《THE 7TH INTERNATIONAL CONFERENCE ON COMPUTER SCIENCE & EDUCATION (ICCSE 2012)》 * |
| MALIK MAGDON-ISMAIL等: "SSDE-Cluster: Fast Overlapping Clustering of Networks Using Sampled Spectral Distance Embedding and GMMs", 《2011 IEEE INTERNATIONAL CONFERENCE ON PRIVACY, SECURITY, RISK, AND TRUST, AND IEEE INTERNATIONAL CONFERENCE ON SOCIAL COMPUTING》 * |
| 黄发良等: "Web社区谱聚类的比较研究", 《小型微型计算机系统》 * |
| 黄发良等: "基于线图与PSO的网络重叠社区发现", 《自动化学报》 * |
Cited By (40)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103838803A (en) * | 2013-04-28 | 2014-06-04 | 电子科技大学 | Social network community discovery method based on node Jaccard similarity |
| CN103838804A (en) * | 2013-05-09 | 2014-06-04 | 电子科技大学 | Social network user interest association rule mining method based on community division |
| CN103425737A (en) * | 2013-07-03 | 2013-12-04 | 西安理工大学 | Overlapping community discovery method for network |
| CN103455612B (en) * | 2013-09-07 | 2016-08-17 | 西安电子科技大学 | Based on two-stage policy non-overlapped with overlapping network community detection method |
| CN103455612A (en) * | 2013-09-07 | 2013-12-18 | 西安电子科技大学 | Method for detecting non-overlapping network communities and overlapping network communities based on two-stage strategy |
| CN103678669B (en) * | 2013-12-25 | 2017-02-08 | 福州大学 | Evaluating system and method for community influence in social network |
| CN103793489B (en) * | 2014-01-16 | 2017-01-18 | 西北工业大学 | Method for discovering topics of communities in on-line social network |
| CN103888541A (en) * | 2014-04-01 | 2014-06-25 | 中国矿业大学 | Method and system for discovering cells fused with topology potential and spectral clustering |
| CN103888541B (en) * | 2014-04-01 | 2017-02-08 | 中国矿业大学 | Method and system for discovering cells fused with topology potential and spectral clustering |
| CN104850728A (en) * | 2015-02-10 | 2015-08-19 | 湖南大学 | Social network degree analysis method based on mean field theory and spreading coefficient |
| CN105282011A (en) * | 2015-09-30 | 2016-01-27 | 广东工业大学 | Social group finding method based on cluster fusion algorithm |
| CN106780053A (en) * | 2015-11-20 | 2017-05-31 | 香港中文大学深圳研究院 | A kind of overlap community discovery method and system based on node connection preference |
| CN105678626A (en) * | 2015-12-30 | 2016-06-15 | 南京理工大学 | Overlapped community excavation method and apparatus |
| CN105678626B (en) * | 2015-12-30 | 2020-02-21 | 南京理工大学 | Method and device for mining overlapped communities |
| CN106909619A (en) * | 2017-01-16 | 2017-06-30 | 中国科学院声学研究所 | It is a kind of based on offset adjusted and the mixing social networks clustering method bidded and system |
| CN106909619B (en) * | 2017-01-16 | 2020-04-10 | 中国科学院声学研究所 | Hybrid social network clustering method and system based on offset adjustment and bidding |
| CN107153713A (en) * | 2017-05-27 | 2017-09-12 | 合肥工业大学 | Overlapping community detection method and system based on similitude between node in social networks |
| CN107153713B (en) * | 2017-05-27 | 2018-02-23 | 合肥工业大学 | Overlapping community detection method and system based on similitude between node in social networks |
| CN107705213A (en) * | 2017-07-17 | 2018-02-16 | 西安电子科技大学 | A kind of overlapping Combo discovering method of static social networks |
| CN107705213B (en) * | 2017-07-17 | 2022-01-28 | 西安电子科技大学 | Overlapped community discovery method of static social network |
| CN109325814A (en) * | 2017-07-31 | 2019-02-12 | 上海诺悦智能科技有限公司 | A method of for finding suspicious trade network |
| CN107577727A (en) * | 2017-08-22 | 2018-01-12 | 武汉大学 | A kind of One-male unit behavioral trait analysis method |
| CN107577727B (en) * | 2017-08-22 | 2020-10-30 | 武汉大学 | Group movement behavior characteristic analysis method |
| CN107844939A (en) * | 2017-11-28 | 2018-03-27 | 广州真知码信息科技有限公司 | Sampling estimation cargo numbering method |
| CN107844939B (en) * | 2017-11-28 | 2022-01-04 | 广州真知码信息科技有限公司 | Sampling estimation cargo numbering method |
| CN108846543A (en) * | 2018-04-26 | 2018-11-20 | 深圳大学 | A kind of calculation method and device of non-overlap community set quality Measure Indexes |
| CN108898264A (en) * | 2018-04-26 | 2018-11-27 | 深圳大学 | A kind of calculation method and device being overlapped community's set quality Measure Indexes |
| CN108600013A (en) * | 2018-04-26 | 2018-09-28 | 北京邮电大学 | The overlapping community discovery method and device of dynamic network |
| CN108898264B (en) * | 2018-04-26 | 2021-10-29 | 深圳大学 | Method and device for calculating quality metric index of overlapping community set |
| CN108846543B (en) * | 2018-04-26 | 2021-10-29 | 深圳大学 | Computing method and device for non-overlapping community set quality metric index |
| CN108833158A (en) * | 2018-06-08 | 2018-11-16 | 成都理工大学 | A kind of similitude community discovery method based on k-means |
| CN112384920A (en) * | 2018-07-11 | 2021-02-19 | 维萨国际服务协会 | Privacy preserving graph compression with automatic fuzzy variable detection |
| CN110535681B (en) * | 2019-07-18 | 2021-04-27 | 西安电子科技大学 | Open computing language OpenCL parallel-based overlapping community detection method |
| CN110535681A (en) * | 2019-07-18 | 2019-12-03 | 西安电子科技大学 | Based on the parallel overlapping community detection method of open operation language OpenCL |
| CN110442800A (en) * | 2019-07-22 | 2019-11-12 | 哈尔滨工程大学 | A kind of semi-supervised community discovery method of aggregators attribute and graph structure |
| CN110910262A (en) * | 2019-10-31 | 2020-03-24 | 北京达佳互联信息技术有限公司 | Community grouping method and device, content isolation method and device and server |
| CN111008338A (en) * | 2019-11-11 | 2020-04-14 | 重庆邮电大学 | Social network multi-scale structure mining method and medium based on hierarchical clustering |
| CN111427999A (en) * | 2020-03-19 | 2020-07-17 | 中国科学院自动化研究所 | Theme real-time influence evaluation method and system for comprehensive integration discussion environment |
| CN111427999B (en) * | 2020-03-19 | 2023-05-12 | 中国科学院自动化研究所 | Method and system for evaluating real-time influence of theme facing comprehensive integrated discussion environment |
| CN112269922A (en) * | 2020-10-14 | 2021-01-26 | 西华大学 | Community public opinion key character discovery method based on network representation learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102929942B (en) | 2015-08-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102929942B (en) | The overlapping community discovery method of a kind of community network based on integrated study | |
| CN101383748B (en) | Community division method in complex network | |
| CN105512289B (en) | Image search method based on deep learning and Hash | |
| WO2021134871A1 (en) | Forensics method for synthesized face image based on local binary pattern and deep learning | |
| CN107464210B (en) | Image style migration method based on generating type countermeasure network | |
| CN108510012A (en) | A kind of target rapid detection method based on Analysis On Multi-scale Features figure | |
| CN108446794A (en) | One kind being based on multiple convolutional neural networks combination framework deep learning prediction techniques | |
| CN108062560A (en) | A kind of power consumer feature recognition sorting technique based on random forest | |
| CN101739721B (en) | Time change and disordered multi-image-based four-dimensional modeling method | |
| CN104008165A (en) | Club detecting method based on network topology and node attribute | |
| CN110353675A (en) | The EEG signals emotion identification method and device generated based on picture | |
| CN105931046A (en) | Suspected transaction node set detection method and device | |
| CN104268629B (en) | Complex network community detecting method based on prior information and network inherent information | |
| CN111753207B (en) | Collaborative filtering method for neural map based on comments | |
| CN105388402A (en) | Cross-linked cable partial discharge mode recognition method based on parameter optimization SVM (Support Vector Machine) algorithm | |
| CN103413174A (en) | Short-term wind speed multi-step prediction method based on deep learning method | |
| CN107784327A (en) | A kind of personalized community discovery method based on GN | |
| CN107895202A (en) | Short-term wind speed forecasting method based on various visual angles wind speed mode excavation | |
| CN108627798A (en) | WLAN indoor positioning algorithms based on linear discriminant analysis and gradient boosted tree | |
| CN103955580B (en) | Parametric Yield of VLSI IC method of estimation based on reliability rule base reasoning | |
| CN106251230A (en) | A kind of community discovery method propagated based on election label | |
| CN110765582B (en) | Self-organization center K-means microgrid scene division method based on Markov chain | |
| CN106294418A (en) | Search method and searching system | |
| CN104715034A (en) | Weighed graph overlapping community discovery method based on central persons | |
| CN107391594A (en) | A kind of image search method based on the sequence of iteration vision |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20150812 Termination date: 20170927 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |