Various semi-supervised learning methods have been proposed recently to solve the long-standing shortage problem of manually labeled data in sentiment classification. However, mos...
The lack of Chinese sentiment corpora limits the research progress on Chinese sentiment classification. However, there are many freely available English sentiment corpora on the W...
Automated identification of diverse sentiment types can be beneficial for many NLP systems such as review summarization and public media analysis. In some of these systems there i...
In this paper we discuss problems of constructing classifiers from imbalanced data. We describe a new approach to selective preprocessing of imbalanced data which combines local ov...
Motivated by the numerous applications of analysing opinions in multi-domain scenarios, this paper studies the potential of a still rarely considered approach to the problem of mu...