Abstracts SIKS-dag 2001


Intermediate-level cognition

Dr. E.O. Postma

Cognition develops from the interaction of a biological system with its natural environment. Therefore, experiments with situated systems (e.g., robots) are crucial to further our understanding of the nature of knowledge and intelligence. In the lecture, "intermediate-level cognition" is argued to bridge the gap between low-level behaviour (e.g., obstacle-avoidance and navigation) and high-level cognition (e.g., reasoning and thinking). The study of intermediate-level cognition is argued to lead to novel insights into the manifestation and nature of knowledge representations. I will address the following three questions:
  1. How do intermediate-level cognitive structures emerge from visual (auditory, haptic, olfactory) interaction with the environment?
  2. What is the nature of knowledge representations in situated models of intermediate-level cognition and in what way do they differ from the traditional knowledge representations?
  3. Is it possible to apply situated knowledge representations in a non-situated context?
The preliminary answers to these questions underscore the importance of intermediate-level cognition for future artificial-intelligence research.

From business architecture to software architecture

Prof dr. J.L.G. Dietz

A major problem in designing information systems that support business systems concerns the development phase requirements engineering (RE). The purpose of RE is to provide a complete and precisely specified set of requirements that must be fullfilled by the information system. I will confine myself to the most problematic kind of requirements, the so-called functional requirements. Three propositions are presented and discussed.

The first one is that information systems basically support the operation of business systems and that therefore constructional knowledge of these business systems is needed. The notion of constructional knowledge will be elaborated and implications of this proposition for RE methods and techniques will be discussed.

The second proposition is that the language action perspective (LAP) provides the most appropriate theoretical foundation for acquiring constructional knowledge of business systems. I will particularly present and discuss my own CAP-theory that a.o. is the basis of the methodology DEMO, which is successfully applied in practice and at a growing scale.

The third proposition is that the notion of architecture is very helpful in finding the right level of abstraction in the acquired constructional knowledge and in deriving from it the functional specifications of information systems. The notion of architecture will be elaborated, both for business systems and for information systems.

Verification by model checking of feature interactions in software systems

Dr. M. Ryan

A feature is a piece of functionality. For example, possible features of a mobile phone system include: ability to copy numbers from the SIM card to the handset; ability to take voice mail when user unavailable; ability to selectively automatically forward text messages based on sender.

Often, new systems are developed by adding features to old ones, a practice which we call "feature integration". When several features are integrated into the same system, they may interfere with each other in undesirable ways; this is called the "feature interaction problem".

I will review some known examples of feature interaction which have arisen in telecommunications, and describe an approach to interaction detection based on verification by model checking a feature construct.

The feature construct allows the programmer to override behaviour of the base system. I will describe some of its properties, and (if time) examine feature integration in a more abstract setting.