Abstract

A partition-based framework is presented for a formal study of consistent classification problems. An information table is used as knowledge representation. Solutions to, and solution space of, classification problems are formulated in terms of partitions. Algorithms for finding solutions are modeled as searching in a space of partitions under a refinement order relation. We focus on a particular type of solutions called conjunctively definable partitions. Two level construction methods for decision trees are investigated. Experimental results are reported to compare the two level construction methods.

Citations