Capturing the semiotic relationship between terms

Download

PDF - Published Version Download (1020Kb)

Description/Abstract

Tags describing objects on the web are often treated as facts about a resource, whereas it is quite possible that they represent more subjective observations. Existing methods of term expansion expand terms based on dictionary de�nitions or statistical information on term occurrence. Here we propose the use of a thematic model for term expansion based on semiotic relationships between terms, this has been shown to improve a system's thematic understanding of content and tags and to tease out the more subjective implications of those tags. Such a system relies on a thematic model that must be made by hand. In this article we explore a method to capture a semiotic understanding of particular terms using a rule-based guide to authoring a thematic model. Experimentation shows that it is possible to capture valid de�nitions that can be used for semiotic term expansion but that the guide itself may not be su�cient to support this on a large scale. We argue that whilst the formation of super de�nitions will mitigate some of these problems, the development of an authoring support tool may be necessary to solve others.