SMASH Project Summary
Starting date: August 1, 1996.
Duration: 5 years
We propose the construction of a general-purpose heterogeneous
rational multi-agent architecture, and the development --on a
computational implementation of this architecture-- of prototype
multi-agent systems with learning capabilities that cooperate in the
solution of complex problems in medical environments. This proposal
involves four working-hypothesis:
Distinctive features of this proposal are:
- Complex rational behaviour, in a software agent, is the result of
the cohesive interaction of basic "rationality aspects"
(ontological, epistemic, motivational, communicational), each one
of them has a formal representation, basic units are assembled
through simple, well-defined means, and a rational software agent
always has elements that are context-specific.
- Those formalizations can be computationally transcribed in an
Object-Oriented environment --where available logic-based tools can
be adapted and encapsulated-- to build general purpose rational
agents that can be tailored to act efficiently in a given context.
- The deployment of rational general-purpose accountable software
agents, that can be tailored to cooperatively solve different
tasks, is a viable, effective and economically sound way of
addressing some of the challenges brought forward by real world
- The medical and hospital-management environments constitute
appropriate "proof of concept" cases for multi-agent based tools;
from a technical standpoint, because of the complexity and richness
of the problems involved, and from a technology-transference
perspective, because of the technological maturity of these
- Formal: the use of a layered reflective first-order framework, based
on a reflective dynamic logic, that integrates into a cohesive
architecture various "aspects of rationality", which, in turn, are
themselves formal systems.
- Technical: (i) the "encapsulation", in an object-oriented
environment, of the computational counterparts of the layered
reflective structure. (ii) Cooperative problem solving through
multiagent systems with federated learning capabilities.
- Technology Transfer: An industry-assessed "proof of concept" in five
real-world test problems of considerable economic significance in
the medical and hospital-management environments.