Burmese Sentiment Analysis Based on Transfer Learning

Using a rich resource language to classify sentiments in a language with few resources is a popular subject ofresearch in natural language processing. Burmese is a low-resource language. In light of the scarcity of labeledtraining data for sentiment classification in Burmese, in this study, we propose a method of transfer learningfor sentiment analysis of a language that uses the feature transfer technique on sentiments in English. Thismethod generates a cross-language word-embedding representation of Burmese vocabulary to map Burmesetext to the semantic space of English text. A model to classify sentiments in English is then pre-trained using aconvolutional neural network and an attention mechanism, where the network shares the model for sentimentanalysis of English. The parameters of the network layer are used to learn the cross-language features of thesentiments, which are then transferred to the model to classify sentiments in Burmese. Finally, the model wastuned using the labeled Burmese data. The results of the experiments show that the proposed method cansignificantly improve the classification of sentiments in Burmese compared to a model trained using only aBurmese corpus. KCI Citation Count: 0