Assessing naive Bayes and support vector machine performance in sentiment classification on a big data platform

<p><span lang="EN-US">Nowadays, mining user reviews becomes a very useful mean for decision making in several areas. Traditionally, machine learning algorithms have been widely and effectively used to analyze user’s opinions on limited volume of data. In the case massive data, powerful hardware resources (CPU, memory, storage) are essential dealing with whole data processing phases including, collection, pre-processing, an optimal time. Several big technologies emerged efficiently process like Apache Spark, which is distributed framework that provides libraries implementing algorithms. order evaluate performance Spark's library (MLlib) large classification accuracies time two implemented spark: naive </span><span>B</span><span lang="EN-US">ayes support vector (SVM) compared achieved by standard implementation these different size datasets built from movie reviews. The results our experiment show classifiers running under spark higher than traditional ones reaches F-measure greater 84%. At same time, we found framework, relatively low.</span></p>