Automatic recognizing of complex sentences with coordinating relation
Chunfeng Yuan · Jisuanji yingyong yanjiu · 2008
This paper extracted nearly 40 000 sentences as training data and testing data from People's Daily newspaper.And then chose some main features such as subject,object and used Hownet to quantify these features.Finally trained these data with support vector machine(SVM) to get the model.The proposed method had about 84 % precision on the test data.The result of experiment indicates that this approach is feasible.