Aspect Analysis for Opinion Mining of Vietnamese Text

Abstract

Aspect extraction is one of most challenging tasks in opinion mining. Many researches have attempted to solve this problem for English text. For less popular languages such as Vietnamese, their complex structure causes difficulties in management or semantic analysis tasks. In this paper, we propose an approach to extracting and classifying aspect-terms for Vietnamese language. The semi-supervised learning GK-LDA is proved to have better performance than the traditional topic modeling LDA. In the aspect inference, we use dictionary-based method which can extract noun-phrases for obtaining better performance than just extract word seeds or use a complete sentence to infer aspects. Our experimental results show that our proposed method can effectively perform the aspect extraction and classification task. Even though our approach is initially proposed for handling Vietnamese text, we believe that it is also applicable to other languages.