Customer Opinion Mining by Comments Classification using Machine Learning

Abstract

In this era of digital and competitive market, every business entity is trying to adopt a digital marketing strategy to get global business benefits. To get such competitive advantages, it is necessary for E-commerce business organizations to understand the feelings, thinking and seasons of their customers regarding their products and services. The major objective of this study is to investigate customers' buying behavior and consumer behavior to enable the customer to evaluate an online available product in various perspectives like variety, convenience, trust and time. It performs data analysis on the Ecommerce customer data which is collected through intelligent agents (automated scripts) or web scrapping techniques to enable the customers to quickly understand the product in given perspectives through other customers' opinion at a glance. This is qualitative and quantitative e-commerce content analysis in using various methods like data crawling, manual annotation, text processing, feature engineering and text classification. We have employed got manually annotated data from e-commerce experts and employed BOW and N-Gram techniques for Feature Engineering and KNN, Naïve Bays and VSM classifiers with different features extraction combinations are applied to get better results. This study also incorporates data mining and data analytics results evaluation and validation techniques like precision, recall and F1-score.