The field of social network analysis is in constant flux, with new issues emerging inspired by practice operating networks, or by theoretical ideas needing an application area. This multi-author work focuses on community detection or evolution (seven chapters), as well as link prediction, influence, and information propagation (four chapters). It comprises selected papers from the 2013 IEEE/ACM International Conference on Advances in Social Network Analysis.
Most chapters do not offer insight into the algorithms besides simple statistics, social network analysis, and keyword frequency analysis used in some case studies, but a few do. Some interesting chapters combine well-known graph theory-based social network analysis with natural language processing, for example, to identify dominant nodes in a network. Some others track node behavior over time via probability evolutions or semantic networks, making the split between passive and active actors. A theoretical chapter deals with similarity relationships in behavior over time via attribute interaction matrices and an entanglement index. A basic result is provided on consensual communities, shown not to exist in random graphs as one would expect; this depends, however, on the consensus metric used. A highly speculative chapter proposes an approach to link predictions in heterogeneous collaboration networks, involving a re-weighting of nodes based on features extracted from patterns of prominent interactions across the network; this relies on a diversity of unsupervised proximity metrics and diffusion assumptions. Another chapter on link prediction makes the assumption of microblog networks with narrower topic relevance. One chapter deals with influence in the narrow sense of hierarchically maximizing product adoption within a social network under a constrained advertising budget.
In conclusion, this volume presents interesting approaches in relation to communities interacting via social networks, but most chapters lack the validation results or a critical review of the assumptions made, which would help the reader narrow down his or her options. The volume may be of interest to specialists mining social network data and to some advanced students wishing to further their research. Each chapter has its own references, and the collective work has a common glossary and index.