Sentiment Analysis Based on MapReduce

Sentiment analysis is the process of analyzing people's sentiments, opinions, evaluations and emotions by studying their written text. It attracts the interest of many researchers, since it is useful for many applications, ranging from decision making to product evaluation to mention a few. Sentiment analysis can be conducted using machine-learning techniques, lexicon-based techniques or hybrid techniques that combines both. As people are more reliant on social networks such as Twitter, this has become a valuable source for sentiment analysis. However, the existence of big data frameworks require adaptation of these techniques to run within such frameworks. This paper reviews sentiment analysis techniques, focusing on the MapReduce-based analysis techniques. We found that the Naïve Bayes algorithm was the most used machine learning technique for extracting sentiments from big datasets because of its high accuracy rates. However, the dictionary-based techniques achieved better results in terms of execution time.