An Automatic Method of Finding Topic Boundaries

Abstract

This article outlines a new method of locating discourse boundaries based on lexical cohesion and a graphical technique called dotplotting. The application of dotplotting to discourse segmentation can be performed either manually, by examining a graph, or automatically, using an optimization algorithm. The results of two experiments involving automatically locating boundaries between a series of concatenated documents are presented. Areas of application and future directions for this work are also outlined.

3 pages, in student session of ACL 1994