Automatic and Interactive Browsing Hierarchy Construction for Scientific Publication Collections

Pre-constructed browsing hierarchies are often incapable of supplying the right set of terms to describe a new scientific publication collection. Even if a browsing hierarchy contains descriptive terms, they may not be organized in the same way as they are presented in the collection. Browsing hierarchies derived directly from the collection can be far more effective than pre-constructed ones. In this paper, we present a novel automatic browsing hierarchy construction algorithm which can derive browsing hierarchies that match the content of a collection of scientific publications. It also allows librarians or others who construct browsing hierarchies to interactively modify the hierarchies and, to some extent, teaches the algorithm to predict further human modifications. A user study and experimental results show that our algorithm is effective in creating hierarchies to support browsing activities for arbitrary collections.