The book presents rough set formalisms and methods of modeling and handling incomplete information and motivates their applicability to knowledge representation, knowledge discovery and machine learning. The book focuses on providing representational and inference mechanisms for dealing with two particular aspects of incompleteness, namely indiscernibility and similarity. Those manifestations of particular aspects of incompleteness are inherent in any data structure and any cognitive unit. Knowledge discovered from such an information is uncertain in that it can only be asserted with a tolerance. The methods developed in the book are capable of exposing the limits of that tolerance and of making reliable inferences in the environments where complete information is not available. The framework presented in the book is general and unrestrictive, and yet at the same time captures the relevant features of a great variety of the user's data.