This paper mainly studies the rule acquisition and attribute reduction for formal decision context based on two new kinds of decision rules, namely I-decision rules and II-decision rules. The premises of these rules are object-oriented concepts, and the conclusions are formal concept and property-oriented concept respectively. The rule acquisition algorithms for I-decision rules and II-decision rules are presented. Some comparative analysis of these algorithms with the existing algorithms are examined which shows that the algorithms presented in this study behave well. The attribute reduction approaches to preserve I-decision rules and II-decision rules are presented by using discernibility matrix.