disambiguation

Disambiguation is the process of resolving ambiguity when a term, symbol, or statement has multiple possible meanings. It appears in language, data, and information systems, with the aim of identifying the intended sense within a given context.

In linguistics, disambiguation includes word sense disambiguation, which selects the specific meaning of a word in context, and referential or anaphora resolution, which links pronouns and references to the correct antecedents. In information retrieval and natural language processing, named entity disambiguation or entity linking connects mentions to precise entries in a knowledge base. Polysemy and homonymy are the linguistic phenomena that disambiguation seeks to distinguish.

In reference works and databases, disambiguation pages or qualifiers help readers distinguish topics with the same name. For example, a page titled "Mercury" may direct readers to articles about the planet, the chemical element, the Roman god, or a car model, often using qualifiers like "(planet)" or "(element)."

Methods range from simple heuristic rules based on surrounding text to advanced machine learning models trained on annotated data. Approaches often combine contextual features, prior knowledge from knowledge bases, and cross-language signals. In practice, disambiguation is essential for accurate search, question answering, summarization, and data integration.

Challenges include subtle sense distinctions, cross-domain ambiguity, multilingual contexts, and resource limitations. Ongoing work aims to improve accuracy with larger knowledge bases, better representations, and robust evaluation benchmarks.