Discussions and Summary

While some research work has been devoted to big data analytics recently, very few studies about big data stream analytics are reported in the literature. The main theoretical contributions of our research include the design and development of a novel big data stream analytics framework, named BDSASA for the near real-time analysis of consumer sentiments. Another main contribution of this paper is the illustration of a probabilistic inferential language model for analyzing the sentiments embedded in an evolving big data stream generated from online social media. The business implication of our research is that business managers and product designers can apply the proposed big data stream analytics framework to more effectively analyze and predict consumers' preferences about products and services. Accordingly, they can take proactive business strategies to streamline the marketing or product design operations.

One limitation of our current work is that the proposed framework has not been tested under an empirical setting. We will devote our future effort to evaluating the effectiveness and efficiency of the BDSASA framework based on realistic consumer reviews and social media messages collected from the Web. On the other hand, we will continue to refine the proposed inferential language model for better sentiment polarity prediction. For instance, a consumer may connect to other consumers via a social network. We may incorporate such connection features in the inferential language model when the sentiment polarity of a review is analyzed. Moreover, the prediction thresholds for probabilistic opinion scoring will be fine-tuned using the proposed PCGA. Finally, we will conduct a usability study for the proposed big data stream analytics service in a real-world e-Business environment.