Aspect based sentiments analysis using PCA modeling

Abstract

newline The proliferation internet access across the globe and enhancement of web technologies has enabled the mundane man to crisscross ginormous user generated reviews with regards to any product or service. Nevertheless, this massive quantity of data appears in the unstructured form that must be administered by text mining applications instead of human experts due to their enormous size. Sentiment Analysis is one such application that aids in identifying the likes and dislikes of other individuals who are already experiencing that specific entity. These persons keep posting their opinions or comments or discussions on various social media platforms so that other people can benefit from their reviews. This comprehensive framework of sentiment analysis becomes equally significant in both academic and commercial domains. Not only the potential buyers, but also the retailers also find this opinion mining process interesting because this vast pool of opinions enables them to collect feedback, review their products/ services in a better way and acknowledging whether these entities are appreciated by the end users or not. However, the novice internet users may still struggle as the fine-grained sentiments about each aspect of the entity may stay hidden. This is where Aspect Based Sentiment Analysis (ABSA) steps in and it is briefly described as the sentiment analysis s subset which tries to extract more refined user views by disintegration into aspects. Thus, ABSA provides granular opinion mining to determine reviewer s attitude towards specific features of an entity that benefits multiple stakeholders in decision making. This research work provides three imperative contributions in the field of ABSA. First, an effective aspect term mechanism is presented that ensembles POS tagging and dependency rules to competently extracts the major aspects. Second, myriad machine learning classifiers are applied to ascertain efficacy of the vector-based features and Principal Component Analysis (PCA) is utilized for dimensi