The Ancient Chinese Word Sense Disambiguation Based on CRF
Hui Li · Microelectronics & Computer · 2009
This paper firstly analyzes the ancient Chinese word sense disposition and the characteristic,and inspects the difficulty of the ancient Chinese word sense disambiguation.Then basing on the existing theory and methods of word sense disambiguation,by choosing contextual words and part of speeches and adding linguistic features,the conditional random fields(CRF) model is used,and six different templates are designed.6 Chinese high frequency words like 将,如,我,信,闻,之 are tested.And the best average F-score achieves 83.04%,which is better than the result of Maximum Entropy and NaiveBayes models.The experiment indicates that CRF model is effective in the ancient Chinese word sense disambiguation.