Multiword Expressions and Idiomaticity: How Much of the Sailing Has Been Plain?

摘要

Much progress has been made in designing accurate word representations [2—4], with improvements for language technology applications like machine translation and text simplification. Precise natural language understanding requires adequate treatments both of single words and of larger units. In particular, one commonly held assumption for constructing the representation for larger units like expressions, phrases and sentences, is that the meaning of the unit can be constructed from the meanings of its parts, in what is known as the Compositionality Principle. While it allows an interpretation to be generated even for unseen combinations of known words, it may not be adequate for expressions like idioms, verb-particle constructions and compound nouns as they often display idiomaticity. For instance, this is the case of loan shark with the meaning of a person who lends money at extremely high interest rates (rather than a fish that can be borrowed). Therefore it is important to identify which words in a sentence form an expression [5], and whether an expression is idiomatic [1,6] and should be treated as a unit, as this determines if it can be interpreted from a combination of the meanings of their component words or not. In this talk I discuss advances on the identification and treatment of multiword expressions in texts, focusing in particular on techniques for modelling idiomaticity.