Omiotis: A thesaurus-based measure of text relatedness

Abstract

In this paper we present a new approach for measuring the relatedness between text segments, based on implicit semantic links between their words, as offered by a word thesaurus, namely WordNet. The approach does not require any type of training, since it exploits only WordNet to devise the implicit semantic links between text words. The paper presents a prototype on-line demo of the measure, that can provide word-to-word relatedness values, even for words of different part of speech. In addition the demo allows for the computation of relatedness between text segments. © 2009 Springer Berlin Heidelberg.