A Study On The Automatic Acquisition Of Verb-Object Collocation For Chinese Language

Collocation describes how a word is used in relation to others in a language structure to form a specific meaning. Extracting the collocations of words is a key part of language use and understanding. Verb-Object collocations play an important role in parsing because of its long spread and strong combination competition.In this paper, we presented an efficient Chinese verb-object collocation acquisition algorithm. Co-occurrence, Arbitrariness, Semantic restriction, Spread between Verb and Object, and the rhythm feature are important properties of collocations. The approach we proposed is to combine four factors acquired from annotated corpus to derive useful models, including (V, N) structure conditional probability, semantic combined probability, span combined probability and rhythm combined probability, and then apply the statistical data acquired with proper weights in determining whether a Verb and a noun after it form a collocation. The experiments results demonstrated that high precision can be achieved with our method.The error analysis shows that the performance can be improved by applying shallow parsing technique as a preprocessor and using a larger scaled and more appropriate training corpus.Wang Xia (Language Information Processing) Directed by Dr. Sun Honglin...