This paper explains an application of big data analytics to detect illegitimate transactions performed by fraudulent communities of people who are engaged in a notorious tax evasion practice called circular trading. We designed and implemented this technique for the commercial taxes department, government of Telangana, India. This problem is solved in two steps. In step one, the problem is formulated as detecting fraudulent communities in a social network, where the vertices correspond to dealers and edges correspond to sales transactions. In step two, specific type of cycles are removed from each fraudulent community, which were identified in step one, to detect the illegitimate transactions. We used RHadoop framework for implementing this technique. © Springer Nature Switzerland AG 2019.