EqRank: Theme Evolution in Citation Graphs

Time evolution of the classification scheme generated by the EqRank algorithm is studied with hep-th citation graph as an example. Intuitive expectations about evolution of an adequate classification scheme for a growing set of objects are formulated. Evolution compliant with these expectations is called natural. It is demonstrated that EqRank yields a naturally evolving classification scheme. We conclude that EqRank can be used as a means to detect new scientific themes, and to track their development.