Transfer Learning Using Twitter Data for Improving Sentiment Classification of Turkish Political News

摘要: In this paper, we aim to determine the overall sentiment classification of Turkish political columns. That is, our goal is whether whole document has positive or negative opinion regardless its subject. order enhance performance classification, transfer learning applied from unlabeled Twitter data labeled A variation self-taught been proposed, and implemented for classification. Different machine techniques, including support vector machine, maximum entropy Naive-Bayes used supervised phase. experiments have obtained up 26 % increase in accuracy with inclusion into columns using learning.