Intelligent Inference For Debugging Concurrent Systems
Human Cognition Research Laboratory
The Open University
Milton Keynes, England, MK7 6AA.
Tel: [+44] (0)908 65 5015
FAX: [+44] (0)908 65 3169
email: [email protected]
Abstract: This paper gives an overview of a program visualisation system
which is constructed around a model of information structuring. The? model
consists of three techniques: domain visualisation, user based? information
management and manipulation, and intelligent agent-based inference. ? This
paper will give a brief overview of the model itself, and then concentrate on
(a) novel ways of letting the user use and manipulate trace histories, and (b)
the use of knowledge based techniques to explore the information? present.
The work reported in this paper is part of a system called MRE (Brayshaw, 1991a & b, sub A, sub B). It represents a continuing evolution of program visualisation systems for logic programming (e.g. see Eisenstadt and Brayshaw, 1986, 1988; Brayshaw and Eisenstadt, 1991). In each of these systems a program is run and a trace history generated (non-termination being specially catered for). This trace history can then be reviewed by animation, manipulated according to grainsize of information, or searched. This representational environment allows both novice and experts access a wide body of information, arranged around? the concept of a visual execution model, which portrays the virtual machine as? a concrete visual entity. Such a system for Prolog is now available commercially (Expert Systems Ltd., 1990), freely for Macintoshes? (TPM, 1991), and is used for teaching on our undergraduate and post-graduate courses. In this paper we will introduce the following:
? visually programmed search agents that allow large information areas to be search, either as a result of particular behest of the user or via? continual monitoring.
? agents can be augmented by a simple temporal logic in order to allow users? to reason about time in relation to the occurance of symptomatic clusters? of behaviours.
1This work is currently supported by a MRC/SERC/ESRC UK Joint Research Council Grant #89/CS31, the support of which is gratefully acknowledged.