Oskar Dressler

An extended Basic ATMS

Classical truth maintenance systems keep one consistent database at a time and allow non-monotonic justifications. De Kleers ATMS [de Kleer 86a] processes multiple contexts simultaneously, but allows monotonic justifications only. The attempt to introduce non-monotonic justifications and defaults at higher levels of the system [de Kleer 86b] failed.

Putting the Problem Solver Back in the Driver's Seat: Contextual Control of the AMTS

The Atms is a powerful tool for automated problem solvers and has been used to support several model-based reasoning tasks such as prediction and diagnosis. It provides an efficient mechanism for maintaining consistent sets of beliefs and recording the assumptions underlying them. This enables the problem solver to switch rapidly between contexts and compare them. Such capabilities are central to diagnostic systems, and are also valuable to design and planning systems. Applications to larger problems have been hampered, however, by the problem solvers inability to maintain control over the Atms.

Assumption-Based Truth Maintenance

Truth Maintenance Systems (TMS) have been in use for quite a while. They forced the view of dividing the overall problem solver into a component solely concerned with the domain rules and a component solely concerned with recording the state of the search. These systems seemed to be exclusively devoted to keeping a consistent database. Assumption-based Truth Maintenance (ATMS) extended this view and introduced the capability to handle multiple contexts simultaneously. This paper shortly reviews the motivations for TMS. Then the ATMS terminology is introduced and the Basic ATMS is described. Finally, some extensions to the Basic ATMS are given.

Problem Solver Control Over the ATMS

The ATMS is a powerful tool for automated problem solvers. It provides an efficient mechanism for maintaining consistent sets of beliefs and recording the assumptions underlying them. This enables the problem solver to switch rapidly between contexts and compare them. Applications to larger problems have been hampered, however, by the problem solver’s inability to maintain control over theATMS.

A framework for controlling model-based diagnosis systems with multiple actions

In recent years reasoning about structure and function of physical systems for the purpose of diagnosis has seen a dramatic increase in activities. New exciting results concerning modelling issues, diagnostic inference patterns and inferential power have emerged. A state of the art diagnosis agent now has a considerable toolset at hand. A main obstacle for building large diagnosis systems, however, remains. How can we controlwhen to usewhich inference pattern or representation? We argue that the actions available to a diagnosis agent can be understood in terms of change ofworking hypotheses. The control problem then becomes a belief revision problem: when to adopt or drop beliefs. Our approach proceeds in two steps. First, we adopt the principle of informational economy from Gärdenfors, Knowledge in Flux (MIT Press, 1988) as kind of a law of inertia for diagnostic processes, that helps us identify candidates for revised belief states. In a second step we employ specificdiagnostic knowledge to actually choose the next belief state. We demonstrate the use of our concepts on an example in the domain of ballast tank systems as e.g. used in offshore plants.

Propagation of Temporally Indexed Values in Multiple Contexts

Modelling multiple views of a dynamic system requires both an efficient technique for dealing with multiple contexts and a mechanism for explicit reasoning about time. Assumption-based truth maintenance systems simultaneously handle multiple, possibly contradictory views, which may be characterized by different sets of assumptions about correctness of components, particular operating modes, etc.. This is mainly achieved by recording inference steps in the form of explicit dependencies and propagating sets of assumptions along these dependencies. Constraint systems explicitly reasoning about time use value-interval pairs, so-called episodes, as parameter values.

Erweiterungen des Basic ATMS

Truth Maintenance Systeme halten zu jeder Zeit eine konsistente Datenbasis und verarbeiten nicht-monotone Justifications. De Kleers ATMS [de Kleer 86a] berechnet multiple Kontexte simultan, verarbeitet aber nur monotone Justifications. Erst auf hoheren Ebenen lassen sich nicht-monotone Justifications und die darauf aufbauenden normal- und non-normal Defaults sowie Negation einfuhren [de Kleer 86b].