Text mining is a classical domain of research and development, in this technique the text data is used for preparing the data models for classifying similar patterns of text documents. In order to perform this classification and categorization algorithms are implemented to perform the required task. But the text patterns in different social sites and blogs are not only categorized in their subjective similarities. These texts can also be classified on the basis of author’s sentiments or the emotional aspects. Therefore the identification of emotional components and features from the social site text and classify the posts on the basis of their sentiments is respectively new domain of research and development. In this presented work the micro-blog text is classified according to the text sentiments and emotional features. Therefore the twitter micro text is used for training and testing of supervised data models. In this context a supervised hybrid classification technique is developed using the Bay’s classifier and the Back propagation neural network. The key role of the Bay’s classifier is to find the emotional components in terms of positive wordlist and negative word list. Additionally using these components the message is encoded in numerical strings. These numerical strings are further used with the BPN algorithm for performing the training and testing operations. In addition of that to improve the classification performance of text the abbreviations and similes are also recovered as the emotional components. The implementation of the proposed technique is performed using JAVA technology and for classifiers WEKA library is used. After the implementation the performance of the proposed technique is evaluated and compared with the similar classification technique. In comparative performance study the proposed model found efficient and accurate as compared to traditionally used technique.