Multilabel Aspect-Based Sentiment Classification for Abilify Drug User Review

Abstract

Multilabel text classification plays an important role in text mining applications such as sentiment analysis and health informatics. In this paper, we propose a multilabel aspect-based sentiment classification model for Abilify drug user reviews. First, we employ preprocessing techniques to obtain the quality of data. Second, the term frequency-inverse document frequency (TF-IDF) features are extracted with Bag of words (BoWs). Third, a joint feature selection (JFS) method with Information Gain (IG) is applied to select label specific features and label sharing features. Moreover, multilabel classification task can be solved using the problem transformation approaches, adapted algorithm approaches, and ensemble approaches. Finally, we study the problem transformation approaches, binary relevance (BR), classifier chains (CC), and label Powerset (LP) to classify Abilify user reviews into a set of aspect term sentiment (ATS). The baseline classifiers Naïve Bayes (NB), decision tree (DT), and support vector machine (SVM) is employed on both feature sets. The proposed method evaluated on multilabel metrics such as accuracy, Hamming Loss, F1-micro averaged, and accuracy per Label. The empirical results show that the support vector machine outperforms.