TreeMatch: A Fully Unsupervised WSD System Using Dependency Knowledge on a Specific Domain

2010-07-15

International Workshop on Semantic Evaluation

AI Summary

This work addresses domain-specific Word Sense Disambiguation (WSD), proposing the first fully unsupervised, dependency-knowledge-driven framework. Unlike conventional supervised or semi-supervised approaches that rely on annotated corpora or general-purpose lexical resources (e.g., WordNet), our method leverages only domain-customized dependency relations extracted from a domain-specific knowledge base to model word semantics. It performs unsupervised semantic similarity computation guided by dependency structures and conducts coarse-grained sense matching—both without any human-annotated training data or external dictionaries. The key contribution lies in the explicit integration of domain-specific dependency knowledge into the WSD pipeline, enabling effective sense discrimination in a purely unsupervised setting. Evaluated on the SemEval 2010 Task 17 benchmark, our approach substantially outperforms the First Sense Baseline, demonstrating both the efficacy and transferability of domain dependency knowledge for unsupervised WSD.

Technology Category

• Natural Language Processing: Lexical Semantics and Morphology
• Data Mining & Knowledge Management: Semantic Web
• Knowledge Representation and Reasoning: Common-Sense Reasoning

Application Category

• Semantics and Knowledge: Methods, algorithms and applications for the development of semantic models, knowledge graphs and other forms of structured data models with machine-interpretable semantics
• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Word sense disambiguation (WSD) is one of the main challenges in Computational Linguistics. TreeMatch is a WSD system originally developed using data from SemEval 2007 Task 7 (Coarse-grained English All-words Task) that has been adapted for use in SemEval 2010 Task 17 (All-words Word Sense Disambiguation on a Specific Domain). The system is based on a fully unsupervised method using dependency knowledge drawn from a domain specific knowledge base that was built for this task. When evaluated on the task, the system precision performs above the First Sense Baseline.

Problem

• Automatic Word Sense Disambiguation
• Computer Understanding
• Specific Domains

Innovation

• Automatic Semantic Disambiguation
• Unsupervised Learning
• TreeMatch System