Investigating Approaches to Semantic Category Disambiguation Using Large Lexical Resources and Approximate String Matching

概要

論文の詳細を見る
This paper proposes and investigates several improvements for an existing machine learning-based system for the task of semantic category disambiguation. For querying large scale lexical resources with millions of lexical entries using approximate string matching we investigate the application of a semantically motivated distance measure, using start/end markers for the query, selecting the most beneficial lexical resources out of a set and the effect using similarity thresholds. These approaches are evaluated using six datasets from the domain of BioNLP and while some modest improvements are observed we fail to establish a consistent benefit for any of the suggested methods for all datasets. The introduced system and all related resources are freely available for research purposes at: https://github.com/ninjin/simsem
2011-11-14