sensefor

Sensefor is a software framework and API designed to provide word-sense disambiguation and semantic tagging for natural language text. It maps words and phrases to defined senses in a structured ontology, enabling consistent interpretation across applications such as search, annotation, and knowledge graphs.

Its core features include multilingual support, compatibility with standard lexical resources (WordNet, FrameNet), and an extensible pipeline that combines rule-based heuristics with neural context models. Developers can annotate text automatically, refine sense inventories with custom domains, and query the sense repository via REST or gRPC endpoints. It supports batch processing and streaming input.

Sensefor's architecture consists of a sense annotator, a disambiguation engine, a sense repository, and an ontology mapper. The annotator tokenizes input, tags parts of speech, and identifies candidate senses; the disambiguation engine selects the most likely sense using contextual features and, optionally, user feedback. The repository stores sense definitions and links to related concepts in an ontology or knowledge graph.

Common use cases include content tagging for search and recommendation systems, organization of large document collections, and population of knowledge graphs with sense-level edges. It is used in research settings to study lexical ambiguity and in industry projects that require precise semantic annotation.

Sensefor began as a collaborative project among researchers and developers and has since evolved through multiple releases and community contributions. It is maintained as open-source software with active documentation and contributor guidelines.