Redescription Mining With Three Primary Data Mining Functionalities

Abstract

Describing an object in two ways or shifting the vocabulary of the same concept is Redescription. Not a new problem, Redescription Mining premise had resulted the subsets of objects that afford multiple definitions, in a given Universal set of the same, and a collection of features to describe them. Now-a-days, huge amounts of data available either to classify or to categorize leads us to ambiguous state as it is accomplished with complementary and contradictory ways. Hence data has to be reduced. This involves cataloging, classification, identifying rules among the data, segmentation or partitioning of the data. The Learning algorithms of data mining techniques on this data can often be viewed as a further form of data reduction. This Sine-qua-non data has been characterized by the multitude of descriptors. In a way, these descriptors are also made equivalent and hence reduced. The methodology of redescriptions can be obtained in scores of data mining techniques. In this paper we overview how data mining functionalities like classification, clustering and Association rule mining achieve the goal of redecsriptions.