Concept-Level Sentiment Analysis using Dependency-based Semantic Parsing : A novel approach

Concept-level sentiment analysis is always superior to word-level analysis as it preserves the semantics associated with multi-word expressions. It offers a better understanding of text and helps to significantly increase the performance of sentiment analysis model. Concept extraction from unstructured text is a key step in concept-level sentiment analysis. In this paper, we propose a novel concept parser scheme to extract semantic features that exploit the semantic relationship between words in natural language text. Further, more conceptual information of a concept is obtained using the ConceptNet ontology. Concepts extracted from the text are sent as a query to ConceptNet to extract their semantics. Next, important concepts are selected and redundant concepts are eliminated using Minimum Redundancy and Maximum Relevancy (mRMR) feature selection technique. Further, all the B. Agarwal Department of Computer Science and Engineering MNIT Jaipur, India E-mail: [email protected] S. Poria Department of Computing Science and Mathematics University of Stirling, UK E-mail: [email protected] N. Mittal Department of Computer Science and Engineering MNIT Jaipur, India E-mail: [email protected] A. Gelbukh Centro de Investigacin en Computacin Instituto Politcnico Nacional, Mexico E-mail: [email protected] A. Hussain Department of Computing Science and Mathematics University of Stirling, Stirling FK9 4LA, UK, E-mail: [email protected] concepts are used to make the machine learning model to classify a given document as positive or negative. Proposed extraction approaches are evaluated on benchmark movie review dataset provided by Cornell University, and product review datasets (book, DVD, electronics). Experimental results show the effectiveness of proposed approaches for sentiment analysis over state-of-art methods.