Note for phd students: We can store a copy of your thesis here, so that it remains available after you leave your institute!
"...the presented work describes a step towards the industrial and commercial acceptance of intelligent user support for application systems with graphical interactive user interfaces. It allows of an inexpensive integration with an established application system, and it supports the optional sophistication of the provided user support through the development of more application-tailored support in an experimental setting. Appropriate findings can then be realized in a subsequent release of the particular application system. This way, users experienced in application systems can become used to the new help opportunities without being disturbed in their daily routine, while new users are already able to exploit the advantages of extended and intelligent user support in order to become familiar with the application system. "
Described within this thesis is an approach to educational hypermedia that utilises symbolic AI and connectionist AI to provide generic student modelling. The needs for generic tutoring systems are discussed, in terms of a system that is applicable to a multitude of teaching domains, whilst maintaining diagnostic facilities of the student. The first contribution to knowledge is an approach to educational hypermedia that employs a semantic network allowing the use of automatic reasoning, using symbolic AI, to produce weighted links, the weight being the relevance to the current node according to the semantic network. The novice student is aided by weighted links in that complexity can be reduced by removing the links with the lowest weight and hence relevance. The linking system is further enhanced by the second contribution to knowledge. The second contribution to knowledge concerns a type of student model that is employed to record information about the student so that the weighted links can be tailored towards the student's interests. The student model contains sub-systems to measure and record domain-independent information about the student, which includes their interests, as derived from the semantic network, their ability to complete tutorials and their browsing behaviour. A neural network is used to grade the student into an ability level based upon
This thesis proposes the concept and realization of an adaptation component for an open, adaptive hypermedia system which implements advanced teaching strategies and enables integration and adaptation of learning material found in the WWW. The adaptation component uses an indexing concept for describing the content of the various information resources. This indexing concept is also taken as a base for constructing a model of the application domain. A Bayesian inference mechanism calculates estimations about the user's knowledge on top of this domain model.
In this dissertation, we address the problem of information filtering & sharing on the Internet from an artificial intelligence (AI) perspective. Taking advantage of recent advances in machine learning, knowledge representation, information retrieval and multi-agent systems, we define a model for Agent-based Information Filtering and Recommender Systems on the Internet, validating it though several prototypes and theories.
The main goal of this dissertation is to improve the accuracy and efficiency of the diagnosis process in an ITS. To this end, we have explored the possibility of using Approximate Reasoning techniques, with special emphasis on simplifying their application as much as possible so their use is not an unaffordable load of additional work to the considerable task of developing an ITS. The proposed solution is substantiated in the definition of a new integrated student model based on Bayesian Networks (BN), and in the application of Computer Adaptive Tests (CAT) theory to improve the efficiency and accuracy of the diagnosis process. This student model allows substantial simplifications when defining the BN (nodes, links and parameters) used to represent the student model, and its integration with a CAT diagnostic algorithm (in which each question is selected adaptively according to the current estimation of students knowledge level) has shown an excellent performance in the evaluations we have realized with simulated students.
This dissertation presents an engineering approach to adaptive hypermedia and personalised Web applications. The approach includes a development process that supports the entire lifecycle of adaptive hypermedia applications from feasibility study to maintenance including project management, software development and quality management activities. The development process is object-oriented, incremental and iterative.
The main focus of the work is the description of a systematic methodology for the analysis and design of adaptive hypermedia applications. A UML profile is specified. The aim is to provide an adequate notation that allows for an easy construction of navigation, presentation and adaptation models, which are part of the proposed methodology. In addition, a Dexter-based reference model for adaptive hypermedia systems is defined. It is an object-oriented approach specified by a combination of a visual modeling language (UML) and a formal specification language (OCL).