This paper presents the enhanced emotion recognition in text and its classification. Emotion detection plays an important role and can be used in wide range of health, business and security applications. The recent research in text mining is mostly based on bipolar approach wherein the emotions are being classified as positive (happy) or negative (angry). In this paper, keyword based approach has been adopted to classify the text into further four emotional categories, i.e., happiness, sadness, anger and other. The proposed approach processes the special linguistic cases like conjunctions and contradictory conjunctions or contrast sentences or its parts to enhance the efficiency of emotion classification. This approach results in twofold advantage: firstly, by considerably enhancing the accuracy of emotion classification and secondly, providing a complete road map to this new area of research. The proposed technique removes emotion stimulus in the text and adjusts the sense of the sentence which improves the accuracy of the classifier.