#### DMCA

## Detecting communities and their evolutions in dynamic social networks -- a Bayesian approach (2010)

Venue: | MACH LEARN |

Citations: | 45 - 2 self |

### Citations

5032 |
Stochastic relaxation, Gibbs distribution and the Bayesian restoration of images
- Geman, Geman
- 1984
(Show Context)
Citation Context ...ntractable problem to optimize the posterior probabilities in the offline and online inference algorithms introduced in the previous section. As a consequence, we appeal to the Gibbs sampling method (=-=Geman and Geman 1984-=-; Griffiths and Steyvers 2004) for the solutions. In Gibbs sampling, we need to compute the conditional probability of the community assignment of each node conditioned on the community assignments of... |

3722 | Normalized cuts and image segmentation
- Shi, Malik
- 1997
(Show Context)
Citation Context ...lgorithms—static stochastic block models (SSBM, Holland and Leinhardt 1976) for DSBM, static graphfactorization clustering (SGFC, Yu et al. 2005) for FacetNet, and static spectral clustering (SSpect, =-=Shi and Malik 2000-=-) for EvolSpect. Figure 5 presents the performance, in terms of the normalized mutual information with respect to the ground truth over the 10 time steps, of all the algorithms for the four datasets w... |

1055 |
Finding scientific topics
- Griffiths, Steyvers
- 2004
(Show Context)
Citation Context ...optimize the posterior probabilities in the offline and online inference algorithms introduced in the previous section. As a consequence, we appeal to the Gibbs sampling method (Geman and Geman 1984; =-=Griffiths and Steyvers 2004-=-) for the solutions. In Gibbs sampling, we need to compute the conditional probability of the community assignment of each node conditioned on the community assignments of other nodes. We will first d... |

947 | Modularity and community structure in networks - Newman |

531 | Graphs over time: Densification laws, shrinking diameters and possible explanations - LESKOVEC, KLEINBERG, et al. - 2005 |

399 |
Pattern Recognition and Machine Learning (Information Science and Statistics
- Bishop
- 2006
(Show Context)
Citation Context ...imated parameters. This is usually called point estimation in statistics, and is notorious for its instability when data is noisy. We address the limitation of point estimation by Bayesian inference (=-=Bishop 2006-=-). Instead of using the most likely values for the model parameters, we utilize the distribution of model parameters when computing the prediction. 5.1 Conjugate prior for Bayesian inference We first ... |

364 | Mixed membership stochastic blockmodels
- AIROLDI, BLEI, et al.
- 2008
(Show Context)
Citation Context ...g et al. 1985; Hoetal.2002; Shortreed et al. 2006; Snijders 2002; Wasserman and Pattison 1996). The SBM model has been successfully applied in various areas such as bioinformatics and social science (=-=Airoldi et al. 2006-=-; Fienberg et al. 1985;Hoetal.2002). Researchers have extended the stochastic block model in different directions. For example, Airoldi et al. (2006) proposed a mixed-membership stochastic block model... |

356 | On the bursty evolution of blogspace - Kumar, Novak, et al. - 2003 |

310 | Logit models and logistic regressions for social networks: I. An introduction to Markov graphs and p
- Wasserman, Pattison
- 1996
(Show Context)
Citation Context ...lock model (SBM). This model was originally proposed by Holland and Leinhardt (1976) and was further extended by others, e.g. (Fienberg et al. 1985; Hoetal.2002; Shortreed et al. 2006; Snijders 2002; =-=Wasserman and Pattison 1996-=-). The SBM model has been successfully applied in various areas such as bioinformatics and social science (Airoldi et al. 2006; Fienberg et al. 1985;Hoetal.2002). Researchers have extended the stochas... |

287 | Efficient identification of web communities
- Flake, Lawrence, et al.
- 2000
(Show Context)
Citation Context ...As another example, communities also play an important role in Web analysis, where a Web community is defined as “a set of sites that have more links to members of the community than to non-members” (=-=Flake et al. 2000-=-). Social networks are usually represented by graphs where nodes represent individuals and edges represent relationships and interactions among individuals. Based on this graph representation, there e... |

179 | Markov Chain Monte Carlo estimation of exponential random graph models
- Snijders
- 2002
(Show Context)
Citation Context ...he stochastic block model (SBM). This model was originally proposed by Holland and Leinhardt (1976) and was further extended by others, e.g. (Fienberg et al. 1985; Hoetal.2002; Shortreed et al. 2006; =-=Snijders 2002-=-; Wasserman and Pattison 1996). The SBM model has been successfully applied in various areas such as bioinformatics and social science (Airoldi et al. 2006; Fienberg et al. 1985;Hoetal.2002). Research... |

163 | A spectral clustering approach to finding communities in graphs
- White, Smyth
- 2005
(Show Context)
Citation Context ...efore, a higher modularity value indicates a community structure that better explains the observed social network. Many existing studies, such as (Brandes et al. 2008; Chenetal.2009; Lin et al. 2008; =-=White and Smyth 2005-=-), have used this metric for performance analysis. 8.2 Experiments on synthetic datasets 8.2.1 Data generator We generate synthetic data by following a procedure suggested by Newman and Girvan (2004).... |

158 | Discovering evolutionary theme patterns from text: an exploration of temporal text mining - Mei, Zhai - 2005 |

151 | P.S.: Graphscope: parameter-free mining of large time-evolving graphs - Sun, Faloutsos, et al. |

140 | On modularity clustering
- Brandes, Delling, et al.
- 2008
(Show Context)
Citation Context ... structure versus generated by a random process. Therefore, a higher modularity value indicates a community structure that better explains the observed social network. Many existing studies, such as (=-=Brandes et al. 2008-=-; Chenetal.2009; Lin et al. 2008; White and Smyth 2005), have used this metric for performance analysis. 8.2 Experiments on synthetic datasets 8.2.1 Data generator We generate synthetic data by follow... |

125 | Evolutionary clustering - Chakrabarti, Kumar, et al. - 2006 |

113 | Dynamic social network analysis using latent space models - Sarkar, Moore - 2005 |

112 | D.: A framework for community identification in dynamic social networks - Tantipathananandh, Berger-Wolf, et al. |

92 | An event-based framework for characterizing the evolutionary behavior of interaction graphs
- Asur, Parthasarathy, et al.
(Show Context)
Citation Context ... see that such a naive approach by itself does not give performances comparable to our algorithms. 8.3.3 The paper co-authorship (DBLP) data This data was extracted from DBLP and has been studied in (=-=Asur et al. 2007-=-; Lin et al. 2008). It contains the co-authorship information among the papers in 28 conferences over 10 years (1997–2006). The 28 conferences span three main areas—data mining (DM), database (DB), an... |

89 | Evolutionary spectral clustering by incorporating temporal smoothness, KDD
- Chi, Song, et al.
- 2007
(Show Context)
Citation Context ...des In dynamic social networks, at a given time, new individuals may join in the network and old ones may leave. To handle insertion of new nodes and deletion of old ones, existing algorithm such as (=-=Chi et al. 2007-=-) and (Lin et al. 2008) use some heuristics, e.g., by assuming that all the nodes are in the network all the time but in some time steps certain nodes have no incident links. In comparison, in both th... |

80 | Statistical analysis of multiple sociometric relations
- Fienberg, Meyer, et al.
- 1985
(Show Context)
Citation Context ...eld of statistics, a well-studied probabilistic model is the stochastic block model (SBM). This model was originally proposed by Holland and Leinhardt (1976) and was further extended by others, e.g. (=-=Fienberg et al. 1985-=-; Hoetal.2002; Shortreed et al. 2006; Snijders 2002; Wasserman and Pattison 1996). The SBM model has been successfully applied in various areas such as bioinformatics and social science (Airoldi et al... |

74 | Quantifying social group evolution’, Nature 446 - Palla, Barabási, et al. - 2006 |

68 |
Bayesian approach to network modularity
- Hofman, Wiggins
- 2008
(Show Context)
Citation Context ...e steps. 3.2 Stochastic block model (SBM) We first briefly review the Stochastic Block Model (SBM). SBM is a well studied statistical model that has been successfully used in social network analysis (=-=Hofman and Wiggins 2008-=-; Holland and Leinhardt 1976). In the SBM model, a network is generated in the following way. First, each node is assigned to a community following a probability π ={π1,...,πK} where πk is the probabi... |

64 | Monic: modeling and monitoring cluster transitions - Spiliopoulou, Ntoutsi, et al. - 2006 |

63 | Community evolution in dynamic multimode networks - Tang, Liu, et al. - 2008 |

59 | Facetnet: a framework for analyzing communities and their evolutions in dynamic networks
- Lin, Chi, et al.
(Show Context)
Citation Context ...lem for documents. In their model, the probabilities transiting between clusters are considered independent, while we consider the transition follows certain distribution. Lin et al. (2009b) extends (=-=Lin et al. 2008-=-) by modeling of content of documents. Tang et al. (2008) used joint matrix factorization method to discover the community evolution. Kim and Han (2009) proposed the particle-and-density based method ... |

50 | Dynamic non-parametric mixture models and the recurrent chinese restaurant process with applications to evolutionary clustering - Ahmed, Xing - 2008 |

44 | Finding social groups: A meta-analysis of the southern women data
- Freeman
- 2003
(Show Context)
Citation Context ... to the aggregated data and we set the number of communities to be 2, the number used in most previous studies. Not surprisingly, we obtain the same result as most social science methods reported in (=-=Freeman 2003-=-), that is, women 1–9 belong to one community and women 10–18 belong to the other community. Next, based on the number of events that occurred, we partition the time period into 3 time steps: (1) Febr... |

37 |
Document clustering by concept factorization
- Xu, Gong
- 2004
(Show Context)
Citation Context ...and a higher MI value indicates that the result given by the algorithm C ′ is closer to the ground truth C. This metric MI has been commonly used in the information retrieval field (Gong and Xu 2007; =-=Xu and Gong 2004-=-). Where there is no ground truth available in the dataset, we measure the performance by using the metric of modularity which is proposed by Newman and Girvan (2004) for measuring community partition... |

35 | A particle-and-density based evolutionary clustering method for dynamic networks - Kim, Han - 2009 |

34 | Metafac: community discovery via relational hypergraph factorization - Lin, Sun, et al. - 2009 |

32 | M.: Extracting evolution of web communities from a series of web archives - Toyoda, Kitsuregawa - 2003 |

30 | Detecting communities in social networks using max-min modularity - Chen, Zaı̈ane, et al. - 2009 |

27 | Soft clustering on graphs
- Yu, Yu, et al.
- 2005
(Show Context)
Citation Context ..., we also provide the performance of the static versions for all the algorithms—static stochastic block models (SSBM, Holland and Leinhardt 1976) for DSBM, static graphfactorization clustering (SGFC, =-=Yu et al. 2005-=-) for FacetNet, and static spectral clustering (SSpect, Shi and Malik 2000) for EvolSpect. Figure 5 presents the performance, in terms of the normalized mutual information with respect to the ground t... |

15 |
Machine learning for multimedia content analysis
- Gong, Xu
- 2007
(Show Context)
Citation Context ...s between 0 and 1 and a higher MI value indicates that the result given by the algorithm C ′ is closer to the ground truth C. This metric MI has been commonly used in the information retrieval field (=-=Gong and Xu 2007-=-; Xu and Gong 2004). Where there is no ground truth available in the dataset, we measure the performance by using the metric of modularity which is proposed by Newman and Girvan (2004) for measuring c... |

15 | A bayesian approach toward finding communities and their evolutions in dynamic social networks - Yang, Chi, et al. - 2009 |

8 |
Local structure in social networks. Sociological Methodology
- Holland, Leinhardt
- 1976
(Show Context)
Citation Context ...lock model (SBM) We first briefly review the Stochastic Block Model (SBM). SBM is a well studied statistical model that has been successfully used in social network analysis (Hofman and Wiggins 2008; =-=Holland and Leinhardt 1976-=-). In the SBM model, a network is generated in the following way. First, each node is assigned to a community following a probability π ={π1,...,πK} where πk is the probability for a node to be assign... |

2 | Statistical analysis of multiple sociometric relations. Latent Space Approaches to Social Network Analysis - Ho, Raftery, et al. - 2002 |

2 | Memo 2004-019 - AI |

2 |
A particle-and-density based evolutionary clustering method for dynamic networks
- Shortreed, Handcock, et al.
- 2006
(Show Context)
Citation Context ...robabilistic model is the stochastic block model (SBM). This model was originally proposed by Holland and Leinhardt (1976) and was further extended by others, e.g. (Fienberg et al. 1985; Hoetal.2002; =-=Shortreed et al. 2006-=-; Snijders 2002; Wasserman and Pattison 1996). The SBM model has been successfully applied in various areas such as bioinformatics and social science (Airoldi et al. 2006; Fienberg et al. 1985;Hoetal.... |