Indiscernibility and similarity in an incomplete information table

Abstract

A new method is proposed for interpreting and constructing relationships between objects in an incomplete information table. An incomplete information table is expressed as a family of complete tables. One can define an indiscernibility relation for each complete table and then get a family of indiscernibility relations of all complete tables. A pair of an indiscernibility and a similarity relation is constructed by the intersection and union of this family of indiscernibility relation. It provides a clear semantic interpretation for relationship between objects of an incomplete information table. In fact, it is a pair of bounds of the actual indiscernibility relation if all values in the incomplete table were known. © 2010 Springer-Verlag Berlin Heidelberg.