Aspect Based Opinion Mining o n Mobile Product

Abstract

In this era, the web technology is growing quickly. Many people express their feedback related to products, social issues and services. As e-commerce site is becoming more popular, the customer review related to the product grows quickly. Due to this growth it is very difficult for the customer to read huge amount of data and make a decision whether to buy a product or not. It is also very difficult for the manufacturer of the product in-order to manage and focus on customer opinions. In this research we focus on mobile product review which is extracted from Kaggle site. In this experiment we have focused on one particular mobile product review that is Samsung. After data collection we do pre-processing, and further we extract aspect and corresponding opinion using Natural language processing and then categorize whether the extracted opinion is positive or negative by finding polarity for each extracted opinion of words. Finally performance evaluation is done by using two machine learning algorithm i.e. Multinomial Naive Bayesian (MNB) and K-Nearest Neighbour (K-NN) algorithm. This performance evaluation is calculated based on bag of words. Out of two algorithms K-NN gave best accuracy compared to Multinomial Naive Bayesian.