Multiword Lexical Unit Extraction

摘要

The acquisition of multiword lexical units has long been a significant problem in natural language porcessing, being relegated to the borders of lexicographic treatment. The access to large-scale text corpora in machine-readable formats has recently originated a new interest in phraseology allowing the testing of assumptions made about syntactical regularities and flexibility constraints. In this paper, we porpose a system based exclusively on a statistical methodology that retrieves from naturally occurring text, contiguous and non-contiguous multiword lexical units.