Clustering SPIRES with EqRank

TL;DR

This paper introduces the EqRank clustering algorithm for directed graphs and applies it to the SPIRES citation database, resulting in a hierarchical clustering that supports a web service for scientific paper organization.

Contribution

The paper presents a novel clustering algorithm, EqRank, tailored for directed graphs and demonstrates its application to large-scale scientific citation data.

Findings

Hierarchical clustering of SPIRES using EqRank successfully organized papers.

The clustering results are integrated into a web service for accessing related scientific papers.

EqRank effectively captures the structure of citation networks.

Abstract

SPIRES is the largest database of scientific papers in the subject field of high energy and nuclear physics. It contains information on the citation graph of more than half a million of papers (vertexes of the citation graph). We outline the EqRank algorithm designed to cluster vertexes of directed graphs, and present the results of EqRank application to the SPIRES citation graph. The hierarchical clustering of SPIRES yielded by EqRank is used to set up a web service, which is also outlined.