Community-Detection-New-Findings.md

New Findings About Social Network Analysis (From Statistical Perspective)

1. Static Network Random Generation Mechanism

• Community Detection
  Hot topic : stochastic block model,
  Recommended Paper :
  • Spectral clustering
    • Rohe K, Chatterjee S, Yu B. Spectral clustering and the high-dimensional stochastic blockmodel[J]. The Annals of Statistics, 2011: 1878-1915. (The most Influential spectral clustering paper)
    • Lei J, Rinaldo A. Consistency of spectral clustering in stochastic block models[J]. The Annals of Statistics, 2014, 43(1): 215-237.
    • Qin T, Rohe K. Regularized spectral clustering under the degree-corrected stochastic blockmodel[C]//Advances in Neural Information Processing Systems. 2013: 3120-3128.
  • SDP
    • Amini A A, Levina E. On semidefinite relaxations for the block model[J]. arXiv preprint arXiv:1406.5647, 2014. (Influential One)
    • Cai T T, Li X. Robust and computationally feasible community detection in the presence of arbitrary outlier nodes[J]. The Annals of Statistics, 2015, 43(3): 1027-1059.
  • Others
    • Jin J. Fast community detection by SCORE[J]. The Annals of Statistics, 2015, 43(1): 57-89.
    • Amini A A, Chen A, Bickel P J, et al. Pseudo-likelihood methods for community detection in large sparse networks[J]. The Annals of Statistics, 2013, 41(4): 2097-2122. (Good Paper)
    • Gao C, Ma Z, Zhang A Y, et al. Achieving Optimal Misclassification Proportion in Stochastic Block Model[J]. arXiv preprint arXiv:1505.03772, 2015. (spectral initialization, and then majority vote)
• Link Prediction
  Hot topic : edge probability matrix,
  Application : drug traffic analysis,

2. Relationship Between Static Network and Node Covariates

• Example: follow relationship along with user profiles such as gender, hobby, etc(Covariates).
• Topic1 : use network to analyze node covariates
• Topic2 : combine node covariates clustering structure with network clustering structure, enhance accuracy in finite sample
  (Since in reality, network is small, which could not guarantee enough nodes for 100% accuracy.)
• Topic3 : explorative tools for analysis

3. Dynamic Network Analysis

• Example : tweet, weibo, left-message, share and response in webchat, edges have timestamps

• Topic : assume stability(problem with event-based property)

• References

  • Jack Diamond, Overview About Social Network Analysis, https://www.zhihu.com/question/28939731
  • Jack Diamond, Classical Methods for Community Detection, https://www.zhihu.com/question/29042018