Real-Time Sentiment Prediction on Streaming Social Network Data Using In-Memory Processing
V. Jude Nirmal, D. I. George Amalarethinam · 2017
Sentiment analysis, also called opinion mining is a technique that has been adopted by most organizations for market research, product analysis and customer feedback analysis. Social networking data provides a huge promise by providing the most accurate of the reviews. However, it was identified that the huge nature of the data and the velocity at which the data is generated is on the downside. This is due to the unavailability of techniques used to process them. This paper presents a Spark based sentiment analysis technique that operates on streaming data to provide fast and most accurate results. A comparison of accuracy of the proposed technique was also carried out and it was identified that the proposed parallelized Naïve Bayes technique provides both faster and more accurate results compared to sequential Naïve Bayes technique and SVM technique.