Granule description based knowledge discovery from incomplete formal contexts via necessary attribute analysis

Abstract

Incomplete formal contexts are an extension of the classical formal contexts in which the relationship between some objects and attributes is unknown according to the current information. In fact, incomplete formal contexts are frequently encountered in human cognitive activities. This paper discusses knowledge discovery from incomplete formal contexts based on necessary attribute analysis, including granule description, cognitive concept learning and approximate decision rule mining. Specifically, we put forward a novel granule description method for knowledge discovery so that three-way concepts can be formed by attribute enrichment, and approximate decision rules can be mined via the detailed and concise descriptions of granules. Compared to the existing work, the time complexity of mining approximate decision rules is reduced sharply since only the closely related granules are needed in the proposed granule description method.