A Novel Large-scale Model for Real Time Sentiment Analysis Using Online Shop Reviews and Comments

Abstract

Sentiment analysis concept is referring to natural language processing in lots of domains in order to find which kind of subjective data or information it expresses. Sentiment analysis recently used as a method in online shops to identify about type of buyers review and comment and an impression about services and products. In this research, we propose an adjustable sentiment analysis algorithm for real time analysis on user generated data on products in online shops with a UI that run on local shops as a friendly tool. The proposed model builds a dynamic dictionary from buyers comment and reviews gathered from online shops firstly using selected set of admin-based features extracted from a specific product (or top of a products in a category), then classifying these preprocessed data under predefined classes. According to best knowledge of authors the proposed method introduces new features vectors that strongly increase accuracy and trustworthiness in analyzing online shop reviews and comments with a low time overhead. Our extensive simulation result especially combination with an online shop for real data shows the improved accuracy and fine tuning of the polarity rank for online shops manager.