SEMANTIC AUGMENTATION
	 A bridge between models and realities
	---------------------------------------

		Wei-Min Shen
		Darrell Woelk
		Mike Huhns
		Phil Cannata

		Carnot, MCC, Austin, TX, USA


This paper describes a machanism called Semantic Augmentation, developed in
the Carnot project in MCC, for solving the binding problems between models
and applications in the context of Enterprise Integration. The main ideas
are (1) using Articulation Axioms and Value Mappings to represent the
products of Enterprise Modeling, and (2) using these axioms and mappings to
dynamically augment the internal, distributed query representations so that
all the applications within a "semantic closure" can come to play. With the
current implemantation and experiments, this machanism has demonstrated a
wide range of features for supporting the full connections between models
and applications. These features include: efficient representation for
Enterprise Modeling (requires only O(n) linkages), clear and declarative
boundaries between models and applications for flexible improvements and
modifictions, fully support for heterogreneous and distributed applications
with the option of leaving the existing applications intact, powerful
expressiveness for integrating heterogreneous schemas and different value
formats with the option of "value tagging," easiness of defining and
modifying policies for distribution and security, and potential for
optimizing queries that are semantically distributed in a large number of
heterogeneous platforms. Finally, experiments have shown that the mechanism
is indeed practical in the Carnot's heterogenous execution environments.