Treating Incomplete Knowledge in Formal Concept Analysis

Abstract

Whenever human knowledge is considered, one has to take into account that such knowledge may be incomplete. Since Formal Concept Analysis (FCA for short) deals with the representation and investigation of such knowledge which can be connected to data tables – in FCA these are represented as formal one- or many-valued contexts –, it seems to be useful to have ways of representing (and dealing with) situations, where one knows about the incompleteness of parts of the represented knowledge. In this note we try to give a survey of what has been done so far in connection with the treatment of incomplete knowledge in FCA. In particular, we shall compare different algorithms developed so far in connection with the knowledge acquisition tool of “attribute exploration”, which also treat incomplete knowledge – cf. in particular [B91], [G99] and [H01]. Moreover, at the end, different treatments of incomplete knowledge in many-valued contexts and databases are discussed.