"Foundations of Knowledge Systems with Applications to Databases and Agents"
by Gerd Wagner, Kluwer, 1998
New book review by Michael Schroeder
"Foundations of Knowledge Systems with Applications to Databases and Agents" by Gerd Wagner builds upon well-established results for relational database systems and extends them towards the broader notion of knowledge systems that cover advanced topics such as deduction, abduction, diagnosis, planning, and distribution.
Relational databases are without doubt one of the best examples for theoretically founded research results with tremendous impact in practice. Wagner not only summarises these results in the first part of his book, but addresses already current research issues such as agent-object-relationship modelling and object-relational databases which are according to Stonebraker the databases of the future.
In the following four parts of the book, Wagner extends the initially introduced conceptual framework to incorporate rules and positive and negative knowledge thus unfolding step by step the full concept of knowledge systems.
Part II is dedicated to rules and Wagner distinguishes two important classes of rules: deduction rules as found in the literature ofdeductive databases and reaction rules as introduced in the literature on active databases. Two strengths of Wagner's book are the seamless and coherent incorporation of the two classes of rules into his framework and the balance between rigorously defined formal semantics and practical examples feeding intuition.
Part III and IV deal with positive and negative knowledge systems, respectively. Besides temporal knowledge systems, Wagner introduces fuzzy knowledge systems and conceptualises multi-level security databases. Both of which are novel and highlight a main contribution of the book: The integration of very diverse but closely related database and knowledge system issues.
Next, Wagner addresses negative information and clearly discusses the problems arising and their solutions. In particular, the chapter on negation for fuzzy databases is very interesting because it applies much of the knowledge for classical logic programming semantics to soft-computing. Finally, Wagner addresses advanced topics such as cooperative knowledge bases, diagnosis, and planning. All of them form the basis for the use of knowledge systems as core of intelligent agents.
Didactically, the book is well-written. Moving smoothly from basic to advanced topics, the book develops a coherent framework integrating diverse concepts many of which addressed for the first time. I would recommend the book to anybody - be it theoretically or practically motivated - interested in databases, knowledge representation and logic programming. Covering many classical results the book is a suitable textbook for an advanced course on databases. And last, not least, I really enjoyed reading it.