This paper examines the claim that machine induction can alleviate the current knowledge engineering bottleneck in expert system construction. It presents a case study of the rule induction software tool known as Expert-Ease and proposes a set of criteria which might guide the selection of appropriate domains.

First-order expressive capabilities allow Bayesian networks (BNs) to model problem domains where the number of entities, their attributes, and their relationships can vary significantly between model instantiations. First-order BNs are well-suited for capturing knowledge representation dependencies, but literature on design patterns specific to first-order BNs is few and scattered. To identify useful patterns, we investigated the range of dependency models between combinations of random variables (RVs) that represent unary attributes, functional relationships, and binary predicate relationships. We found eight major patterns, grouped into three categories, that cover a significant number of first-order BN situations. Selection behavior occurs in six patterns, where a relationship/attribute identifies which entities in a second relationship/attribute are applicable. In other cases, certain kinds of embedded dependencies based on semantic meaning are exploited. A significant contribution of our patterns is that they describe various behaviors used to establish the RV’s local probability distribution. Taken together, the patterns form a modeling framework that provides significant insight into first-order expressive BNs and can reduce efforts in developing such models. To the best of our knowledge, there are no comprehensive published accounts of such patterns.

A technology transfer programme, where people from industry have been educated and trained in knowledge engineering on a project basis, has been in operation since 1984 in the Computer and Information Science Department at Linköping University. This review presents the background for the programme, its organization, examples of training projects, educational activities and plans for the future development.

Many real-world domains can benefit from adaptable decentralized task allocation through emergent specialization, especially in large teams of non-communicating agents. We begin with an existing bio-inspired response threshold reinforcement approach for decentralized task allocation and extend it to handle hierarchical task domains. We test the extension on self-deployment of a large team of non-communicating agents to patrolling a hierarchically defined set of areas. Results show near-ideal performance across all areas, while minimizing wasteful task switching through the development of specializations and subsequent respecializations when area demands change. A genetic algorithm is then used to evolve even more adaptable and stable task allocation behavior, by incorporating weight and power coefficients into agents’ response threshold reinforcement action probability calculations.

Artificial Institutions are systems where the regulation defined through norms is based on an interpretation of the concrete world where the agents are situated and interact. Such interpretation can be defined through constitutive rules. The literature proposes independent approaches for the definition and management of both norms and constitutive rules. However, they are usually either not coupled or coupled in an ad hoc and limiting solution. This paper investigates how to make such a coupling. The main contribution of this paper is a formal model basing the regulation provided by the norms on the institutional interpretation of the world provided by constitutive rules. This contribution is based on the Situated Artificial Institutions model that proposes an integrated model of constitutive rules based on status functions.

Lack of sufficient semantic description in the content of Web services makes it difficult to find and compose suitable Web services during analysis, search, and matching processes. Semantic Web Services are Web services that have been enhanced with formal semantic description, which provides well-defined meaning. Due to insertion of semantics, meeting user demands will be made possible through logical deductions achieving resolutions automatically. We have developed an inference-based semantic business process composition agent (SCA) that employs inference techniques. The semantic composition agent system is responsible for the synthesis of new services from existing ones in a semi-automatic fashion. SCA System composes available Web Ontology Language for Web services atomic processes utilizing Revised Armstrong's Axioms (RAAs) in inferring functional dependencies. RAAs are embedded in the knowledge base ontologies of SCA System. Experiments show that the proposed SCA System produces process sequences as a composition plan that satisfies user's requirement for a complex task. The novelty of the SCA System is that for the first time Armstrong's Axioms are revised and used for semantic-based planning and inferencing of Web services.

Psychological models have been used to simulate emotions within agents as part of the decision-making process. The body of this work has focussed on applying the process of decision making using emotions to social dilemmas, notably the Prisoner’s Dilemma. Previous work has focussed on agents which do not move around, with an initial analysis on how mobility and the environment can affect the decisions chosen. Additionally simulated mood has been introduced to the decision-making process. Exploring simulated emotions and mood to inform the decision-making process in multi-agent systems allows us to explore in further detail how outside influences can have an effect on different strategies. We expand and clarify aspects of how agents are affected by environmental differences. We show how emotional characters settle on an outcome without deviation by providing a formal proof. We validate how the addition of mood increases cooperation, while also showing how small groups achieve this quicker than large groups. Once pure defectors are added, to test the resilience of the cooperation achieved, we see that while agents with a low starting mood achieve a payoff closest to the pure defectors, they are reduced in numbers the most by the pure defectors.

Deriving formal specifications from informal requirements is extremely difficult since one has to overcome the conceptual gap between an application domain and the domain of formal specification methods. To reduce this gap we introduce application-specific specification languages, i.e., graphical and textual notations that can be unambiguously mapped to formal specifications in a logic language. We describe a number of realised approaches based on this idea, and evaluate them with respect to their domain specificity vs. generality.

This special issue of The Knowledge Engineering Review includes three papers on the subject of automated knowledge-acquisition tools. Gaines and Shaw (1993) review knowledge acquisition tools based on personal-construct psychology; Birmingham and Klinker (1993) review tools based on predefined models of problem-solving methods; and Eriksson and Musen (1993) review metalevel tools that developers may use to build other knowledge acquisition tools. Although this trio of papers by no means summarizes all recent developments in computer-based assistance for knowledge engineering, these contributions provide examples of three important research frontiers that, in many ways, trace their origins to early expert systems research.

The paper I wrote for the Knowledge Engineering Review was intended primarily to give a balanced review of different techniques developed in (or imported into) Artificial Intelligence to deal with uncertain knowledge.

The First International Conference on Information Processing and the Management of Uncertainty (IPMU) was held in 1986 at a time of great debate about the necessity of modelling uncertainty in intelligent systems (which at that time largely meant rule-based expert systems) and the best way of doing so. Whereas the founders of the Conference on Uncertainty in Artificial Intelligence (UAI) in the United States set out with the aim of promoting the use of probability, the organisers of IPMU chose a diametrically opposed course. Though there were a few papers on probability at IPMU '86, the main focus was on alternative methods, primarily those based upon fuzzy sets. Though subsequent conferences have seen greater mix of papers, IPMU remains largely non-probabilistic with the result that the bulk of the participants come from Europe rather than the United States (despite the large amount of work on uncertainty, and especially probability, that is carried out in the US) making IPMU something of a counterpoint to UAT. The difference in participation is exacerbated by the location—whilst the UAI remains in North America, IPMU alternates between Paris and other cities in Europe, including Urbino in 1988 and Palma in 1992.

The definitions above, separated by ten years, represent two very different conceptions of learning. For Simon learning depends on an internal change in representation, and for Kaelbling it is instead measured in terms of an external change in behaviour. Furthermore, Kaelbling's focus on the situatedness of the learning system being embedded in its environment reflects the recent experience gained by much direct experimentation with physical robots.

It is often claimed that it is easy to write expert systems. This claim was examined by monitoring experienced programmers learning to use the S.I knowledge engineering tool. Their achievements and difficulties were examined using a framework that has emerged from previous research into novices learning to use standard programming languages. Even though the experienced programmers all had several years' experience of programming in more than one standard language, there were similarities between their difficulties in learning to use S.I and the difficulties of complete novices learning to program in standard languages.

The experienced programmers were however able to overcome their initial difficulties fairly quickly, but it is argued that complete novices would not find it so easy to do so. Also the experienced programmers did take time to develop a repertoire of schemeta for representing different kinds of factual, judgemental and procedural knowledge. It was concluded that in S.1, as with other programming languages and softwares tools, it is easy to learn how to do simple things, but difficult, even for experienced programmers to learn how to do more complex things.

No criticism of S.1 is implied. S.1 was found to be a suitable vehicle for introducing non-trivial knowledge engineering concepts, and we believe that similar difficulties would occur in learning to use other knowledge engineering tools.