EverMiner: consideration on knowledge driven permanent data mining process

A data mining task is usually solved in six main steps described in the CRISP-DM methodology. The paper introduces another approach to data mining. Data mining is understood as a permanent knowledge driven process. It is assumed that there is a well organised knowledge repository containing both relevant domain knowledge and new knowledge found in the analysed data. It is also assumed that there are several tools that formulate reasonable data mining tasks, search in the analysed data for true patterns relevant to the formulated tasks, filter out found patterns that can be understood as the consequences of items of knowledge stored in the repository, synthesise new items of knowledge from the remaining patterns and store items of new knowledge in the knowledge repository. It is argued that a system consisting of such tools can be built from already existing tools based on the GUHA method and observational calculi.