Detecting Community Structure in Scientific Network - A Citation Network Analysis for Scientific

The study of the underlying community structure in citation networks affords plentiful aids for the progress of the research field, with building a basis for forthcoming research through the credit of previous research activities; identifying gaps in research for students and researchers; improving the incorporation between practice and theory, and so on. In recent years, a growing numeral of clustering algorithms for categorical data has been introduced to learn the community structure of the scientiuc networks. However, many were ineffective for large scale data, thereby making it harder to visualize the community, itself. Additionally, one of the major limitations of the citation network related works is that, none of them provide a visualization of the community structure, which makes it harder for the researchers to perceive the connections. This paper aims at investigating the community structure of various subdivision fields in the citation network of articles published on different scientific journals. Here, a top-down agglomerative clustering algorithm (Louvain) is applied to reveal major communities that correspond to obviously discernable subfields of Science. The measure of modularity is used to define the power of the community structure, thereby demonstrating that the algorithm used is highly efficient and effective at discovering community structure in the citation network data.