Emotions are strong feeling deriving from one's circumstances, mood, or relationships with others. It involves feelings, experience, physiology, behavior, cognitions and conceptualization. Recognizing user's emotions is a major challenge for both humans and machines. Emotion detection task is very important as it can be used in domains such as software engineering, website customization, education, and gaming. Emotion detection task can be categorized into machine learning, lexical and hybrid approaches. Lexical method uses anonym and synonym of words that are directly describing basic Emotion. If any input text is applied then the word of text will be thoroughly scanned and are matched from the set of word present in the database. Machine Learning-based methods try to recognize emotions based on a previous trained classifier, which map with various machine learning classifiers to detect which emotion category, the input text belongs. In Hybrid methods, emotions are detected by using the combination of emotional keyword and learning patterns collected from training sets in addition to the information of different sciences like Human Psychology. This paper discusses the Machine Learning, lexical and Hybrid Based techniques, which are being
used for emotion detection .