Behavior and Sentiment Analysis of Smart Digital Societies Using Deep Machine Learning Technologies

Abstract

Prediction of behavior and sentiment cues plays a pervasive role in smart digital society including the domains of healthcare and social behavior and sentiment research. The social behavior and sentiment analysis can be performed from the individuals’ offline or online mutual communications and interactions. The corpus of these communications and interactions is considered important for visualizing the behavior and sentiment of a smart digital society. This chapter delineates how these features could be implemented or subsequently fused with other pertinent behavioral, sentimental, and physiological traits to enable different machine learning (ML) techniques to perform the classification problem. This chapter also entails the various challenges and application domains associated with the task of behavior and sentiment analysis. Perhaps, the accuracy of prediction along with the efficacy to imbibe the role of individuals’ behavior and sentiment determinants and to render the explicit explanations to the behavior and sentiment predicted are also indispensable. This is because these aspects can enhance the trust, liability, and reliability of the system. However, the majority of behavior and sentiment analysis approaches do not reflect the logical inferences and explanations for the pursuit of behavior and sentiment prediction. Finally, this chapter describes the deep learning method that has been emerged forth as a pragmatic ML method. Deep learning (DL) possesses the characteristic of enhancing learning related to the data and outputted the superior quality of results. Different DL-based methods have been implemented in different problem domains successfully and researchers also implanted deep learning techniques within the purview of behavior and sentiment prediction. This chapter also incorporates the implementation details of deep learning-based approaches for this purpose.