Bi-view semi-supervised active learning for cross-lingual sentiment classification
作者：
Highlights：
• We combine active and semi-supervised learning for cross-lingual sentiment classification.
• Density analysis of unlabeled data is used in active learning.
• We test our proposed model on three different languages.
• This model reduce manual labelling efforts in cross-lingual sentiment classification.
• Results show that incorporating density analysis can speed up learning process.
摘要
•We combine active and semi-supervised learning for cross-lingual sentiment classification.•Density analysis of unlabeled data is used in active learning.•We test our proposed model on three different languages.•This model reduce manual labelling efforts in cross-lingual sentiment classification.•Results show that incorporating density analysis can speed up learning process.
论文关键词：Cross-lingual,Sentiment classification,Co-training,Active learning,Density measure
论文评审过程：Received 7 June 2013, Revised 14 March 2014, Accepted 17 March 2014, Available online 24 April 2014.
论文官网地址：https://doi.org/10.1016/j.ipm.2014.03.005
