Sarcasm is a form of language in which individual convey their message in an implicit way i.e. the opposite of what is implied. Sarcasm detection is the task of predicting sarcasm in text. This is the crucial step in sentiment analysis due to inherently ambiguous nature of sarcasm. With this ambiguity, sarcasm detection has always been a difficult task, even for humans. Therefore sarcasm detection has gained importance in many Natural Language Processing applications. In this paper, we describe approaches, issues, challenges and future scopes in sarcasm detection. © 2017 IEEE.