Sentiment Analysis (SA) utilizes text contextual mining to identify and extract core subjective information from the source material. Businesses can use SA to monitor online comments or reviews about their brand, product, or service to understand how people feel about their organization. Currently, available classifiers are developed to classify sentiments only in a specific domain, and if it is applied in different domains, it will never give its better performance. General machine learning-based algorithms cannot give the best performance when applied in big data. This paper proposes a Resilience Hidden Markov Model based Support Vector Machine (RHMM-SVM) to classify sentiments in different big product review datasets. RHMM - SVM makes use of a forward - backward strategy to attain better classification accuracy in different big product review datasets. RHMM - SVM is compared and analyzed against existing classifiers with benchmark metrics, namely Precision, Matthew Correlation Coefficient, F1 - Score and Classification Accuracy. Results make a clear indication that RHMM - SVM has better performance than previous classifiers.