Advanced Data Mining Enabled Robust Sentiment Analysis on E-Commerce Product Reviews and Recommendation Model

Abstract

Currently, digital review plays a crucial role in influencing consumer buying patterns and improving global communications amongst consumers. E-commerce giants such as Flipkart, Amazon, and so on, provide real insights about the performance of the product to future buyers and provide a platform for the consumer for sharing their experiences. In order to extract valuable insight from a huge set of reviews, classification of reviews into negative and positive sentiments is needed. Sentiment Analysis (SA) is a computational study for extracting subjective data from text. This article focuses on the design of Advanced Data Mining Enabled Robust Sentiment Analysis on E-Commerce Product Reviews (ADMRSA-EPR) model. The presented ADMRSA-EPR technique mainly relies on the differentiation of the sentiments exist in the online product reviews. In the presented ADMRSA-EPR technique, the first step is to analyze the raw product reviews into useful format and word embedding process takes place. To analyze sentiments exist in product reviews, stacked auto encoder (SAE) model is applied. At the final stage, the parameters related to the SAE model get optimally adjusted using the manta ray foraging optimization (MRFO) algorithm. The experimental result analysis of the ADMRSA-EPR technique on distinct datasets reports a promising performance over the other existing models.