A Novel Approach for Aspect Level Sentiment Analysis

Brisk development of web technology has upshot the number of customers sharing their reviews about their shopping experience online. Online shopping websites, social media sites and blogs have abundance of such reviews, which are used by other customers for their future decisions of shopping from a particular website. There are many factors that affect the user’s decision to shop online, but the reviews written by users are not specific from a single perspective. Sentiment classification of reviews facilitates in understanding the opinions of the users and identifying which aspect affects the user’s choice of online shopping. There is also a need of effective visual analysis of the user’s opinions as they may impact the business organizations significantly and since the need of understanding the large-scale information is very important for future users and businesses; the need to visualize text has gained importance. Due to unstructured and high dimensional nature of the large text corpora, it is challenging to design effective and self explanatory visual metaphors. In this paper, the author has used word clouds to represent the opinions of the customers, identified from their reviews posted online. The opinion words are classified into positive or negatively opinioned words and further mapped into word cloud to understand which features effect the online shoppers most while using online shopping.