Performance analysis of different keyword extraction algorithms for emotion recognition from Uyghur text

This paper conducts the comparing research on Uyghur sentence sentiment classification using different keywords extraction methods. Firstly, the keywords expressing happiness and anger are extracted respectively by the methods of TextRank, SAD and SparseSVM, then used to train the sentiment models accordingly. The sentiment text database is built by excerpting two kinds of sentiments including anger and happiness from Uyghur movie transcriptions and novels. Several experiments are undertaken using different classification methods mentioned above. The experimental results show that the classification methods based on keyword extraction used in this paper are effective in Uyghur text sentence emotion recognition. Among them SparseSVM method gifts robustness and higher accuracy in recognition experiments.