Ant-based document clustering and visualization

Abstract

This paper discusses the document clustering and visualization process: analyzing documents index, clustering document, and visualizing exploration. It focuses on ant-based clustering algorithm and some significant improvements. Clusterings are formed on the plane by ants walking, picking up or dropping down projected document vectors with different probability. It is shown that the similar documents appear in spatial proximity, whereas unrelated documents are clearly separated in visual space. © 2005 by International Federation for Information Processing.