Clustering research group website homepages

The majority of early exploratory webometrics studies have typically used simple network methods or multi-dimensional scaling to identify hyperlink or text-based relationships between collections of related academic websites. This paper uses unsupervised machine learning techniques to identify groups of computer science departments with similar interests through co-word occurrences in the homepages of the departmental research groups. The clustering results reflect inter-department research similarity reasonably well, at least as reflected online. This clustering approach may be useful for policy makers in identifying future collaborators with similar research interests or for monitoring research fields.