IMPLEMENTATION OF A NOVEL SENTIMENT ANALYSIS TECHNIQUE FOR TWITTER DATA USING RHADOOP

Thasnim K M, M Sudheep Elayidom

Abstract: Sentiment analysis is a method of mining knowledge from sentiments or opinions of others about some specific domain or a problem. It is widely applied in product reviews, movie reviews, Twitter, Micro blog etc. Sentiment analysis or in other words, opinion mining of the tweets available on a micro blogging site such as Twitter offers insights to identify products and services that are relevant to users. Before choosing a product, consumers usually go for others opinion about a particular product or a brand. Our specific objective is to identify the popularity of different smart phone brands using RHadoop . This is done by identifying the polarity of tweets derived and assign sentiment scores accordingly to each brand. The comprehensive evaluation of users response is found more challenging for some brands compared to others since the availability of tweets varies significantly for various brands over time. This paper uses a much advanced technology that integrates the Big Data framework-Hadoop, and a very popular visualization tool called R using the RHadoop package. From the literature it is being observed that very few works are done using this integrated framework. The problem with existing technology is that it uses only Hadoop but my work overcomes the problem of scalability and speed along with providing visualization techniques using RHadoop

Keywords: Product Review, Sentiment Analysis, Opinion Mining, Smart Phone Brands, Polarity of Tweets, RHadoop

DOI: https://doi.org/10.15623/ijret.2016.0505076