In this present work the aim is to utilizing the global text features or identifier for classification of social media text in place of normal text classification. The twitter dataset is used for initial input . Here two text feature extractors are used . Additionally the NLP (Natural Language Processing) based POS (Part of Speech tagger) is used for extraction of text feature. Both text features are used with similar classifier namely the BPN (Back Propagation Neural Network ) and classification is performed . The outcome of the BPN is used for comparing the performance of both the feature extraction potential.