Knowledge – Miner: The Conceptual and Architectural Design of an Efficient Data Mining Tool

The purpose of this paper is to present the conceptual and architectural design of an efficient data mining tool/system that supports the implementation of association rule mining algorithms. It has been observed that most of the existing knowledge discovery systems/data mining tools use fixed algorithms for mining knowledge irrespective of characteristics of the database to be mined. Literature reveals that the performance of algorithms also depend on the characteristics of the database to be mined, this has also been realised by the author during the conduction of experiments. These systems also inherit the shortcomings of the algorithms on which they are based. Moreover, such systems are not able to guide the user/decision makers (who may be the non technical person) regarding suitability of algorithms with respect to targeted database. In this research, a new mining system addressing the above issues has been developed. The system is intelligent in the sense that it can automatically select a suitable algorithm on the basis of dataset characteristics. System provides an interactive environment which is convenient, efficient, flexible, user friendly and ultimately enhances ease to its users. Furthermore, system permits the user to flexibly refine the knowledge in order to discover the interesting knowledge and enables the user to use different components independently as per the requirement. System also offers an integrated application based on real-life Mushroom database. Developed application is useful for the identification of associations that exist between the physical properties of Mushroom.