Corpus-based Techniques for Word Sense Disambiguation
Gina Levow · DSpace@MIT (Massachusetts Institute of Technology) · 1997
1.2 Roadmap In the course of this paper, I will describe a set of three techniques which claim to use corpus-based statistical methods to try to solve the problem of word sense disambiguation. As an introduction I will describe briefly some of the range of approaches which have been applied to this task. I will further indicate the limitations on scalability of these techniques which motivated the shift to automatic machine-learning techniques trained of large corpora for this task. Then I will embark on a careful assessment of the three techniques described in the examination papers addressing the issues below: