Hadoop-Based Big Data Sentiment Analysis Using Machine Learning

Abstract

Sentiment Analysis (SA) and Opinion Mining have emerged as critical research areas in the exponential increase in sentiment-rich social media information on the web. SA has changed into a challenging task with the emergence of Big Data. This study presented an effective method to perform SA on large-scale datasets of tweets employing Machine Learning algorithms in the Hadoop ecosystem. In particular, we developed and executed Naive Bayes(NB), Recurrent Neural Networks (RNN), and Support Vector Machines (SVM) classification algorithms for evaluating datasets, and the effectiveness of the strategy was measured in retrieval matrices—Precision, Recall, F-score, and Accuracy. The experimental outcomes show that our method handles vast sentiment datasets with exceptional efficiency, and notably, SVM outperformed other classifiers by achieving an outstanding accuracy of 95% and a ROC of 93%. According to the results of our analysis, SVM stands out as a reliable choice in important sectors where SA plays a crucial role. This work substantially contributes to the area of SA in the Big Data realm by providing a flexible, quick, and scalable method for massively evaluating sentiment-rich social media content. Additionally, it will be more helpful for increasing significant value to businesses, the government, and individuals.