Identifying the subject of small, sparsely linked collections from a web community

摘要

This work deals with the problem of identifying the subject of small, sparsely linked collections of web documents from a web community. In the course of attempts to find solutions for many problems concerning the web, we are often left with a handful of pages dealing with something in common, but with very few links within them. This paper presents algorithms which work on such collections and output a set of words ordered in the decreasing order of relevance. The most relevant words provide a close approximation of the topic that the collection deals with.