Network Analysis Applied to a Bibliographic Network

In this paper we study the ILPnet2 co-authorship network. The ILPnet2 on-line library (www.cs.bris.ac.uk/ILPnet2/Tools/Reports/) is a repository of more than 1,000 ILP-related articles by well over 500 authors, published between 1970 and 2003. Co-authorship networks constitute a specific view on bibliographic data, in which scientific publications are modeled as vertices, and two vertices are connected by an undirected edge whenever the two corresponding papers share at least one author. We investigate the largest connected component in this network, which contains 816 papers by 526 different authors. Properties of interest include degree distribution and degree centrality, eigenvector centrality, and PageRank. We furthermore study the community structure in this network by applying the Newman-Girvan algorithm. Our main conclusion is that, even with restricted information based purely on co-authorship, bibliographic network analysis can reveal useful and interesting patterns.