A disambiguate method to covering ambiguity based on the collocation information
Su-qin Feng, Lijuan Jiao · 2010 Sixth International Conference on Natural Computation · 2010
Covering ambiguity is a vital issue in Chinese word segmentation. The paper presents the disambiguation strategies based on the collocation information. Firstly, it gets the word that is combinatorial ambiguities from a larger scaled corpus, then counts up it's collocation information. Lastly it uses multi maximal log algorithm for disambiguation. Further the paper uses disambiguated corpus to strengthen and stabilize collocations It is proved to be an easy and effective way in the experiments.