Community Mining in Co-authorship Network

Community is formed by individuals such that those within a group interact with each other more frequently than with those outside the group. Community mining or detection involves discovering groups in a network where an individual's group memberships are not explicitly given. Much of the research on detecting communities in co-authorship networks has been done using publicly available datasets. Discovering communities by using a publication list available from the researchers homepage is a challenging task. As the data available from homepages will be in an unstructured form, an interpreter is required to extract the structured co-authorship network to detect communities. The focus of this work is to detect communities in co-authorship network, for which the co-authorship network will be constructed using the publication list of researchers homepage. Network centric community detection algorithms: Multi-dimensional scaling, Block model, Spectral clustering and Modularity maximization was applied on coauthorship network to mine communities. The results of evaluation of these algorithms using the measures Internal density, Expansion and Conductance will be discussed in this work.