Applying Sentiment-oriented Sentence Filtering to Multilingual Review Classification
Takashi Inui, Mikio Yamamoto · 2011
A method for multilingual review classification is described. In this classification task, machine translation techniques are used to remove language gaps in the dataset, but many translation errors occur as a side-effect. These errors cause a decrease in the review classification performance. To resolve this problem, we introduce a sentiment-oriented sentence filtering module to the process of multilingual review classification. Experimental results showed that the proposed method achieved 81.7 % classification accuracy for the evaluation data. 1