A Model for Tracking Sentiment in Reviews at Aspect Level Using Support Vector Machines

Abstract

Abstract

In this era of internet and online services, opinionated data are generated at an incredible amount. Wading through these huge amounts of opinions to glean relevant insight or information for decision making is a daunting task. Sentiment Analysis is a vital tool in Natural Language Processing (NLP) that has made it possible to understand people’s opinions on different topics. In this study, a level of sentiment analysis known as Aspect Level Sentiment Analysis was used to identify what exactly people are talking about in reviews (such as aspects or features of a subject). A model for Aspect Level Sentiment Analysis using Support Vector Machines was developed. Datasets from a customer review area (Laptop products) was used to train and evaluate the developed model. Results obtained gave an appreciable performance showing concise reviews and accuracy in categorizing opinions on polarity scale (positive or negative).