Application Domains of Aspect and Sentiment Classification Techniques: A Critical Evaluation Survey

Tracking #: 2061-3274

This paper is currently under review
Jibran Mir
Azhar Mahmood
Shaheen Khatoon

Responsible editor: 
Philipp Cimiano

Submission type: 
Survey Article
Recently, Social media has changed the way how information being produced, transferred and consumed. User generated contents in the form of posts, blogs, comments, feedback, and reviews, has established a new connection between the producers and users of information. Tracking such contents has enabled businesses to collect customer’s feedback to provide better services. For users’ the abundance of information from diverse sources helps them tap into the wisdom of crowds, to aid in making more informed decisions. This raised a question of how to overcome information overload and provide rich and coherent user experience. This question has opened a rich venue for researchers on how to analyze such a huge amount of customer feedback to get actionable insight. Lot of research has conducted in this area, which largely depends on opinion mining, sentiment analysis and text mining algorithms to interpret and make sense out of large amount of textual data. However, due to complexity of natural language every application domain demands a different technique to deal with. Previously a significant progress has made on type of opinion mining techniques such as implicit, explicit and aspect based opinion mining. However, there is still a scarcity of comprehensive literature review to guide the research community on which type of application domain, which technique is suitable. In this study, we have analyzed the literature from application domain point of view by making following contributions: i) provision of comprehensive literature for application of aspect level sentiment analysis on various application domain; ii) identification of aspect based sentiment analysis limitations in specific application domain; and iii) proposed an aspect based sentiment analysis model for IMDB (Internet Movie Database). The comprehensive analysis of existing literature will surely help research community to identify strength and limitations of aspect level sentiment analysis for various application domains.
Full PDF Version: 
Under Review