Semi-supervised Multi-view Sentiment Analysis

Semi-supervised learning combines labeled and unlabeled examples in order to find better future predictions. Multi-view learning is another way to improve the prediction by combining training examples from more than one sources of data. In this paper, a semi-supervised multi-view learning approach is proposed for sentiment analysis in the Bulgarian language. Because there is little labeled data in Bulgarian, a second English view is also used. A genetic algorithm is applied for regression function learning. Based on the labeled examples and the agreement among the views on the unlabeled examples the error of the algorithm is optimized, striving after minimal regularized risk. The performance of the algorithm is compared to its supervised equivalent and shows an improvement of the prediction performance.