WSD

WSD stands for word sense disambiguation, a task in natural language processing and computational linguistics that seeks to determine which sense of a polysemous word is activated by its context. For example, in the sentence “I went to the bank,” surrounding words help decide whether bank refers to a financial institution or a riverbank. WSD typically relies on sense inventories such as WordNet that organize meanings into distinct senses.

Approaches to WSD fall into several broad categories. Knowledge-based methods use dictionaries, thesauri, or semantic networks to measure similarity between the surrounding context and candidate senses. Supervised methods train classifiers on labeled corpora where word occurrences are annotated with their correct senses. Unsupervised methods cluster usages or infer senses from large unlabeled data without explicit annotations. Recently, contextualized word representations from neural models (for instance, BERT-based approaches) are widely used to derive sense-aware representations from context.

Evaluation of WSD systems commonly employs standard datasets from shared tasks, such as Senseval or SemEval, and reports metrics like precision, recall, and F1-score. Challenges include determining appropriate sense granularity, handling domain shifts where sense distributions differ, and providing robust performance for low-resource languages or domains with limited annotation.

Applications of WSD include improving information retrieval, machine translation, question answering, and text understanding by reducing ambiguity and aligning meanings across linguistic tasks. Despite progress, high-precision WSD remains difficult, particularly for fine-grained sense inventories and specialized domains.

See also: WordNet, natural language processing, semantic disambiguation, context representation.