Objective
The MEDLAR Action aimed to produce a general framework for deduction in the logics of practical reasoning by internal analysis of different logics, so that various logics and deduction calculi can be mixed. Benchmarks were provided for the capacity of a class of logics to characterise central problems in achieving better human/computer interaction and robot action planning.
The project developed the community's capacity for mechanized reasoning about time and action, the beliefs and knowledge of agents, their intentions and their obligations. These tasks arise naturally in interactive environments, for instance in seeking to provide helpful answers in a travel reservation system or the actions of a robot. Logics for reasoning about such situations are aptly called the logics of practical reasoning by philosophers.
The initial approach involved 3 activities:
case studies in human/computer interaction and robot planning;
evolution of a (potentially multimodal) class of logics capable of practical reasoning with time, action and intent;
development of basic techniques for mechanized deduction with the logics of practical reasoning.
Later activities sought to consolidate this earlier research by planning future exploitation and developing general techniques for interfacting logics and deduction calculi, applying to selected case studies for evaluation.
A series of technical workshops were used to manage the gathering and dissemination of information throughout the consortium and within the wider scientific community.
The interim results of the first 3 activities were presented as 3 collections of initial scientific papers for the first milestone review and subsequently a public workshop at Silwood Park, United Kingdom. These collections are on human computer interaction (HCI) and robotic problems, logics of practical reasoning and the deduction calculi. It is intended that the revised papers for these workpackages will appear in an edited volume.
This study involves seeking an integrated framework for mechanizing nonclassical logics. The particular logics and calculi are structured from the point of view of applications. As a first example for testing the prototype of the general framework, a generalized interpretation of modal logics is investigated. Next, the methodology of labelled deductive systems is introduced, demonstrating why this approach for a general framework is adequate to integrate various logical systems via a unified methodology. Finally, the need for different operational methods of solving problems in formal logic has been assessed in the context of an ambitious example suggested from MEDLAR case studies.
APPROACH AND METHODS
The initial approach involved three activities:
-case-studies in human/computer interaction and robot planning
-evolution of a (potentially multi-modal) class of logics capable of practical reasoning with time, action and intent
-development of basic techniques for mechanised deduction with the logics of practical reasoning.
Later activities sought to consolidate this earlier research by:
-planning future exploitation
-developing general techniques for interfacing Logics and Deduction calculi, applying to selected case studies for evaluation.
A series of technical workshops were used to manage the gathering and dissemination of information throughout the consortium and within the wider scientific community.
PROGRESS AND RESULTS
The interim results of the first three activities were presented as three collections of initial scientific papers for the first milestone review and subsequently a public workshop at Silwood Park, UK. These collections are on HCI and Robotic Problems, Logics of Practical Reasoning, and on the Deduction Calculi. It is intended that the revised papers for these workpackages will appear in an edited volume.
The Action has held scientific workshops attended by workers both from other ESPRIT projects and from non-EC countries, as well as supporting other closely related meetings.
The outline of a general labelled deduction system framework for mixing various logics and calculi has emerged.
The work of this Actions will be continued in the MEDLAR II Basic Research project (6471), where the emphasis will be on developing the concept of a practical reasoning agent, capable of acting autonomously and interacting flexibly with its real world environment.
POTENTIAL
This Action has:
-thrown light on the design of a future deduction workstation in which special-purpose problem solvers and planners cooperate
-provided know-how for the design of specific automated reasoning systems, like robot action planners and interrogatable databases
-advanced the science of automated reasoning
Future work in the MEDLAR II project includes the design of a Medlar Practical Reasoner and scientific collaboration with more applied projects.
Fields of science
Topic(s)
Data not availableCall for proposal
Data not availableFunding Scheme
Data not availableCoordinator
SW7 2BZ LONDON
United Kingdom