Experiment in the Syntactical Analysis of English Noun Phrases.

摘要

The note describes an exercise in the syntactical analysis of phrases from a small corpus of titles of scientific publications. The analysis method is essentially dependency analysis. It is formalized in a manner which deviates from the normal presentation of dependency syntax. Its analytic power is compared to that of the well-known regular and context-free formalisms. Besides the introduction and comparison of formalisms, the author discusses the creation of a lexicon from a corpus, and how to measure the quality of such a lexicon. The discussion is illustrated by an approach towards the analysis of phrases in the corpus. Finally, he poses a number of problems. Some are of a technical nature, e.g. concerning the efficiency of analysis algorithms. Others are more linguistic. An important one is how to characterize the various relations that can exist between words in a phrase.