Sentence-Level Classification Using Parallel Fuzzy Deep Learning Classifier

At present, with the growing number of Web 2.0 platforms such as Instagram, Facebook, and Twitter, users honestly communicate their opinions ideas about events, services, products. Owing to this rise in social extensive use by people, enormous amounts data are produced hourly. However, sentiment analysis or opinion mining is considered a useful tool that aims extract emotion attitude from user-posted on media using different computational methods linguistic terms various Natural Language Processing (NLP). Therefore, enhancing text classification accuracy has become feasible, an interesting research area for many community researchers. In study, new Fuzzy Deep Learning Classifier (FDLC) suggested improving performance data-sentiment classification. Our proposed FDLC integrates Convolutional Neural Network (CNN) build effective automatic process extracting features collected unstructured Feedforward (FFNN) compute both positive negative sentimental scores. Then, we used Mamdani System (MFS) fuzzy classifier classify outcomes two deep (CNN+FFNN) learning models three classes, which are: Neutral, Negative, Positive. Also, prevent long execution time taking our hybrid FDLC, have implemented proposal under Hadoop cluster. An experimental comparative study between some other suggestions literature performed demonstrate offered classifier's effectiveness. The empirical result proved performs better than classifiers true rate, false error precision, kappa statistic, F1-score consumption, complexity, convergence, stability.