Chinese WSD based on features obtaining with shallow parsing
Yangsen Zhang · Jisuanji gongcheng yu sheji · 2010
To solve the problem of obtaining high-quality features in text,the features obtaining algorithm based on shallow parsing is proposed,and the features obtaining model with chunk analysis and identify as core is constructed.By identifying substantive chunks,analyzing center word and part of speech,and identifying empty word chunks,the disambiguation features of polysemy are obtained.On the basis of Word-Sense Tagging Corpus of Institute of Computational Linguistics,Peking University,there are 44 polysemies selected.On the experiment of trainning and predicting using maximum entropy disambiguation model,the accuracy rate of disambiguation arrives at 78.71%.