This is the home page for Líonra Séimeantach na Gaeilge (the “LSG”, or, in English, the Irish Language Semantic Network), a database consisting of Irish words and the semantic relationships among them. A semantic network of this kind is sometimes called a wordnet, after Princeton's English-language WordNet, which was the first-ever full-scale semantic network (dating back to the mid-1980's). Semantic networks are much richer than traditional thesauri, which generally only record (near) synonyms and sometimes antonyms. The LSG, like most other wordnets, encodes a richer set of relationships, including hypernyms and hyponyms (broader and narrower terms), meronyms and holonyms (part vs. whole), etc.

Semantic networks have many applications in Natural Language Processing. They are used in systems for word sense disambiguation, document summarization and indexing, and information retrieval. When a semantic network in one language contains mappings to a second language (ours is linked to the English WordNet), it can be used in various ways to improve machine translation. In general terms, from an artificial intelligence perspective, a semantic network encodes some of the “real-world knowledge” that is required for computers to understand and process texts in a non-trivial way.

The image to the left is a depiction of the full LSG (click it for a full-size version). In fact, this is a simplification of the true picture — each node in the image represents a whole set of synonymous words which could be added as additional branches. Something like this network, but probably thousands of times more complex, is contained in the brain of every Irish speaker — semantic connections like these are made instantly and intuitively. The 3D graph browser allows you to “fly” through this network and manipulate it in various ways.


Even if you are not interested in developing software for language processing, the database can still be quite useful, and for this reason I am offering access to it in several different ways: