A Wide-Coverage Context-Free Grammar for Icelandic and an Accompanying Parsing System

摘要: We present an open-source, wide-coverage context-free grammar (CFG) for Icelandic, and accompanying parsing system. The has over 5,600 nonterminals, 4,600 terminals 19,000 productions in fully expanded form, with feature agreement constraints case, gender, number person. system consists of enhanced Earley-based parser a mechanism to select best-scoring parse trees from shared packed forests. Our is able about 90% all sentences articles published on the main Icelandic news websites. Preliminary evaluation evalb shows F-measure 70.72% parsed sentences. demonstrates that morphologically rich language using CFG can be practical.