When building a Bayesian belief network, usually a large number of probabilities have to be assessed by experts in the domain of application. Experience shows that experts are often reluctant to assess all probabilities required, feeling that they are unable to give assessments with a high level of accuracy. We argue that the elicitation of probabilities from experts can be supported to a large extent by iteratively performing sensitivity analyses of the belief network in the making, starting with rough, initial assessments. Since it gives insight into which probabilities require a high level of accuracy and which do not, performing a sensitivity analysis allows for focusing further elicitation efforts. We propose an elicitation procedure in which, alternately, sensitivity analyses are performed and probability assessments refined, until satisfactory behaviour of the belief network is obtained, until the costs of further elicitation outweigh the benefits of higher accuracy or until higher accuracy can no longer be attained due to lack of knowledge.

We propose an approach to knowledge-based decision-model construction. The domain knowledge is represented in the form of multilevel influence diagrams. At each level, a regular influence diagram representing a decision model may be generated and the influence diagrams between the levels are related through a series of operations. We defined eight such operations and show that these are sufficient at the structural level to construct any target influence diagram in any practical situation. We provide a series of examples drawn from different domains on the use of multilevel influence diagrams for model construction. We also propose suggestions on providing users with guidance that can direct the model-construction procedure towards the best target model.

Bayesian networks (BNs) model problems that involve uncertainty. A BN is a directed graph, whose nodes are the uncertain variables and whose edges are the causal or influential links between the variables. Associated with each node is a set of conditional probability functions that model the uncertain relationship between the node and its parents. The benefits of using BNs to model uncertain domains are well known, especially since the recent breakthroughs in algorithms and tools to implement them. However, there have been serious problems for practitioners trying to use BNs to solve realistic problems. This is because, although the tools make it possible to execute large-scale BNs efficiently, there have been no guidelines on building BNs. Specifically, practitioners face two significant barriers. The first barrier is that of specifying the graph structure such that it is a sensible model of the types of reasoning being applied. The second barrier is that of eliciting the conditional probability values. In this paper we concentrate on this first problem. Our solution is based on the notion of generally applicable “building blocks”, called idioms, which serve solution patterns. These can then in turn be combined into larger BNs, using simple combination rules and by exploiting recent ideas on modular and object oriented BNs (OOBNs). This approach, which has been implemented in a BN tool, can be applied in many problem domains. We use examples to illustrate how it has been applied to build large-scale BNs for predicting software safety. In the paper we review related research from the knowledge and software engineering literature. This provides some context to the work and supports our argument that BN knowledge engineers require the same types of processes, methods and strategies enjoyed by systems and software engineers if they are to succeed in producing timely, quality and cost-effective BN decision support solutions.

Continuing the series of workshops begun in 1996 (Luck, 1997; Doran et al., 1997; d'Inverno et al., 1997; Fisher et al., 1997) and held in each of the two years since (Luck et al., 1998; Aylett et al., 1998; Binmore et al., 1998; Decker et al., 1999; Beer et al., 1999), the 1999 workshop of the UK Special Interest Group on Multi-Agent Systems (UKMAS'99) took place in Bristol in December. Chaired and organised by Chris Preist of Hewlett Packard Laboratories, with support from both HP and BT Laboratories, the workshop brought together a diverse range of participants, from the agent community in both the UK and abroad, to discuss and present work spanning all areas of agent research. Although dominated by computer scientists, also present at the meeting were electronic engineers, computational biologists, philosophers, sociologists, statisticians, game-theorists, economists and behavioural scientists, with both academia and industry well represented. Indeed, numbers attending these workshops continue to grow, reflecting the continued and rising interest in agent-based systems. The meeting truly demonstrated the wider view of what the term “agency” implied to research in other disciplines and the questions raised at the end of presentations were a pertinent reminder of the diversity of the audience.

The theory, principles and practice of multi-agent systems is typically characterised as a computational and engineering discipline, since it is through the medium of computational systems that artificial agent systems are most commonly expressed. However, most definitions of agency draw directly on non-computational disciplines for inspiration. During the 1999 UK workshop on multi-agent systems, UKMAS'99, we invited four speakers to address the conceptualisation of multi-agent systems from their perspective as non-computer scientists. This paper presents their arguments and summarises some of the key points of discussion during the panel.