In this work, community structure extraction essentially resorts to its solution to graph partition problem. The authors explore two different approaches. The spectral approach is based on the minimization of balanced cut and its resulting solution comes from the spectral decomposition of the graph Laplacian. The modularity based approach is based on the maximization of modularity and implemented in a hierarchical fashion. In practice, the approach can extract useful information from the community structure, such as what is the most influential component in a given community. Being able to identify and group friends on social networks, the technique can provide a customized advertisement based on their interests. This can have a big return in terms of marketing efficiency. Community structure can also be used for network visualization and navigation. As a result, it can be seen which groups or which pages have more interaction, thus giving a clear image for navigation purposes.
|Title of host publication||Social Networking and Community Behavior Modeling|
|Subtitle of host publication||Qualitative and Quantitative Measures|
|Number of pages||17|
|State||Published - 1 Dec 2011|