英文摘要 |
In this paper, we describe a new method for extracting monolingual collocations. The method is based on statistical methods extracts. VN collocations from large textual corpora. Being able to extract a large number of collocations is very critical to machine translation and many other application. The method has an element of snowballing in it. Initially, one identifies a pattern that will produce a large portion of VN collocations. We experimented with an implementation of the proposed method on a large corpus with satisfactory results. The patterns are further refined to improve on the precision ration. |