The goal of this paper is to discuss how we establish the “Hammer (CAFÉ) Credit System” by applying Gibbs sampling algorithm under the framework of bigdata approach to extract features in depicting bad or illegal behaviors by following the “five step principle” applying international credit rating standards. In particular, we will show that our Hammer (CAFÉ) Credit System is able to resolve three problems of the current credit rating market in China which rate: “1) the rating is falsely high; 2) the differentiation of credit rating grades is insufficient; and 3) the poor performance of predicting early warning and related issues”. In addition the Hammer (CAFÉ) credit is supported by clearly defining the "BBB" as the basic investment level with annualized rate of default probability in accordance with international standards in the practice of financial industries, and the credit transition matrix for “AAA-A” to “CCC-C” credit grades.