Projects

The projects of the research group can be grouped in four main areas:

	Tutorial
	Efficient Methods for User Modelling
	Information Processing Mechanisms
	User Models for System Assessment and Design
	Online Application of User Models

Tutorial

Half-day tutorial for the ICCM 2006 (5. - 8. April 2006, Trieste)

AGImap is a tool chain that supports cognitive modelling in an effective and efficient way at the interaction level (i.e., perception and execution) and is implemented for ACT-R 6.0. ACT-R models are derived from a graphical description of the interface or an intermediate high-level XML representation. The tool chain provides a graphical editor to create an XML model of the human-machine interface. XLST is used to transform this representation to genuine model (Lisp) code. Currently, the tool chain supports code creation for ACT-R's perceptual-motor subsystem and its AGI interface. To connect the AGI interface elements with a simulated external task environment or even a real process control system the AGImap-extension for cognitive architectures is used. In typical applications the amount of code generated by agimap can take up to 50% of the model's code. The approach allows efficiently dividing the physical and cognitive modelling tasks by using and connecting engineering simulation tools for the "machine" part and cognitive architectures for the cognition part what speeds up the modelling process furthermore. The development of AGImap was supported by DFG and VolkswagenStiftung.
[ more ... ]

Efficient Methods for User Modelling

The use of cognitive architectures to support the design and implementation of complex and dynamic human-machine-systems promises reduced cost and better functionality compared to building individual micro-theories. Current cognitive architectures like ACT-R or SOAR are not sufficiently equipped to support engineering complex and dynamic human-machine-systems. The goal of this project is to identify requirements and propose modifications to current cognitve architectures.
[ more ... ]

ACT-R/PM is a production system and a theory of the cognitive architecture. We apply ACT-R/PM for human-machine system modelling. Practical work turned out that ACT-R/PM and its development environment can be improved for software engineering purposes. We developed a visualization of ACT-R/PM models based on the state chart notation. A novel algorithm was developed to detect and generalize different working memory configurations. It was implemented as a plug-in for the integrated development environment eclipse. The purpose is to make development of complex models easier and less erroneous. It also helps understanding and communicating about cognitive models between modelers.
[more ...]

Information Processing Mechanisms

Temporal reasoning is part of system state diagnosis in supervisory control. As the estimation of durations is supposed to be a prerequisite of this kind of diagnostic reasoning, temporal errors are a main issue of human-machine reliability. Different hypothesis about the cognitive nature of temporal errors are tested in several simualtion experiments. The results provide clear evidence that temporal errors influence operators' picture of the system state and their decisions.
[ more ... ]

Operators in human performance models beyond doubt behave differently. Nevertheless a wide range of psychological theories claims to capture most people`s behavior with the help of normative models not including aspects like vigilance or working memory capacity. Do we thus need theories considering these components, or does one size really fits it all? The main core of this work aims to provide answers to these questions by illustrating the necessity of a theory of interindividual differences.
[ more ... ]

Trend literacy, identified as pattern identification and categorization, is one of central skills expert operators employ to maintain awareness of what's going on in the process. But how do operators read trend views? And how can we utilize this knowledge in situation awareness trainings?
[ more ... ]

User Model Aided System Assessment and Design

A new GOMS based modeling approach is developed and applied to the assesment of infotainment systems in cars. A modified version of the D2 test of attention is used in a car simulator experiment to validate the multitasking-GOMS parametrization. Modeling tools for GOMS like task analysis and new GOMS simulator are implemented.
[ more ... ]

Different approaches, methods and tools in the development process of human-machine systems are used by engineers at the present state of research. These components are well suitable for their own specific tasks but fail in the whole development process. Thus the focus of this research project mainly concentrates on the conceptual development and the prototypical implementation of a simulation environment in order to integrate a wide range of tools and methods into one unified system architecture.
[ more ... ]

The data gained during system engineering-processes from simulation experiments must be consolidated and evaluated and be integrated continuously into the development process. The main focus of this PhD-project is to find methods which allow the adjustment of technical systems on the abilities and characteristics of human during the engineering-process. Quantitative criteria for components of the technical systems are derived (for example control reliability, authority favourableness) which form the basis for algorithms. These algorithms make it possible to draw conclusions and publish recommendations for the engineering of human-machine-interaction in the context of simulation experiments. This will form a way to improve the design- and engineering-process of human-machine-interaction.
[ more ... ]

Online Application of User Models

Socialware is software for user adaptive collaboration support. The UM Testbed project is an attempt to design a reference architecture for socialware systems. The implementation has a interface for different user modelling algorithms. The user model is used to deduce the current relevance of information objects within the socialwares database for a given user. The testbed is based on Java 2 Enterprise Edition Technology deploying the Model View Controller scheme Model 2 (MVC Model 2). It has been used to implement two different web portals for scientific online communities (useworld.net and eupace.net).
[ more ... ]

Current implementations of trend displays in process control systems provide two forms of navigation: panning and zooming. Based on the findings of the field studies on trend literacy we develop new forms of interaction that allow to enhance trend displays with query by example capabilities. This supports operator to get answers for actual questions like "I know this pattern. When did it occur last time?" or "This pattern is new to me. Did it occur some times before?"
[ more ... ]

One of the key-skills of operators in many human-machine-systems is building and maintaining situation awareness i.e. to know the states and elements of the system, especially of important ones, their relevance to the current situation, and to know their future states. We propose a general strategy for training situation awareness skills for human-machine-systems that is based on the application of cognitive models.
[ more ... ]