Systematic Review of Word Sense Disambiguation Systems: State of the Art Techniques and Challenges for Low-Resource Languages

ACL ARR 2025 May Submission1411 Authors

Abstract: Recent advances in Natural Language Processing (NLP) have been driven by the widespread adoption of Large Language Models (LLMs). Despite these improvements, state-of-the-art NLP models still struggle with ambiguous words, often failing to recognise the intended meaning of less commonly used terms in a sentence. This ambiguity problem impacts various linguistic tasks including machine translation and information retrieval, underscoring the importance of Word Sense Disambiguation (WSD). While significant progress has been made in WSD for high-resource languages like English, a notable research gap exists in understanding how current methods perform across multilingual and low-resource settings. Moreover, the impact and potential of LLMs in advancing WSD remain underexplored. This work presents a critical analysis of computational approaches to WSD, evaluating their effectiveness across English, multilingual, and low-resource contexts. We highlight current challenges for state-of-the-art systems and propose future directions in this evolving field.

Paper Type: Long

Research Area: Semantics: Lexical and Sentence-Level

Research Area Keywords: Survey;Polysemy; lexical relationships; textual entailment; compositionality; multi-word expressions

Contribution Types: Surveys

Languages Studied: English and low resourced languages

Submission Number: 1411