The ability to combine domain specific knowledge and special knowledge about using mathematical-statistical methods for analyzing big data bases at present time is not wide-spread in science and business.For the near future, an increase in data mining applications can be expected. So, one needs instruments to support non-specialists in using specific knowledege about data mining.In this paper a data mining architecture is introduced. Its main advantage is to offer a systematical scheme for data mining methods. These methods are structured with reference to applications.The data mining application architecture is a decision and structuring support for data mining problems to users, scientists and students.