@PhdThesis{Fernandes-95,
  author = 	 "Alvaro Adolfo Antunes Fernandes",
  title = 	 "{An Axiomatic Approach to Deductive Object-Oriented
		  Databases}",
  school = 	 "Department of Computing and Electrical Engineering,
		  Heriot-Watt University",
  year = 	 1995,
  address =	 "Edinburgh, Scotland, UK.",
  month =	 "September",
  note =	 "Available via URL:
		  \verb+http://www.cee.ac.uk/~alvaro/alvaro.html+"
}
% Abstract
% -------------------------------------------------------------------
%
%
% Deductive object-oriented databases  (DOODs) have the potential to take
% the applicability and usability of database  technology well beyond the
% current state-of-the-art, insofar as they  unify two of the best  known
% paradigms in  the cutting-edge  of  research on data  management, viz.,
% deductive databases  (DDBs) and  object-oriented databases (OODBs).  If
% this potential is to be fully realized,  the problem of providing DOODs
% with formal foundations cannot be avoided. In particular, a solution to
% the characterization  problem    of DOODs must allow     the principled
% development not just of theoretical  research  but, as importantly,  of
% robust,    fully-functional, industrial-strength implemented systems as
% well.  The great interest in this characterization problem demonstrates
% that there is widespread awareness of such needs.
%
% This dissertation contributes a novel,  {\em axiomatic} approach to the
% formal characterization of   DOODs.  The approach   used  is unique  in
% employing a very  rich axiom set to provide  a logical  account of most
% aspects   of  object-orientation,   including    those essential    for
% behaviour-based modelling. This strategy allows a DOOD to be defined as
% an axiomatically-presented OODB whose metalogic is predominantly shared
% with deductive relational databases (DRDBs).
%
% This result  is relevant  in  removing most  of the  need  to develop a
% metalogic for DOODs  from scratch, since that  of DRDBs can be  reused.
% This,  in  turn, has  the    desirable practical consequence  that  the
% techniques and methods used to implement DRDBs  can be similarly reused
% to  implement DOODs.  Finally,  DOODs   characterized by the  axiomatic
% approach are demonstrably flexible, and easily allow extensions.
%
% The main contributions of this dissertation  are: a formalization of an
% object-oriented data model as a species of mathematical structures, its
% reconstruction  as an    applied first-order   predicate calculus,  the
% conversion  of this  applied  calculus into  a  class of logic programs
% without function symbols, and a logical analysis of the dynamic aspects
% of object-orientation based on a simple and  intuitive extension to the
% standard inference engine of DDBs.
%
% The dissertation also   contains a brief  overview of  ROCK \& ROLL,  a
% fully-operational DOOD system based  on the axiomatic approach, thereby
% substantiating the  claim that the approach  allows the reuse for DOODs
% of   implementation strategies   developed  for  DRDBs.   Finally,  the
% flexibility   conferred by the     approach  is  exemplified   by   the
% strengthening  of the data   model with   additional domain-independent
% constraints, by the incorporation  of a calculus  of state  change into
% the framework, and by the contribution of a logic-based model of active
% databases in which an independent inference  engine for active rules is
% reconciled with the one used to drive deduction.