Hochschulschrift (Dissertation)

VDI-Verlag, Düsseldorf, 2006

ISBN: 3-18-302322-2
ISSN: 1439-958X

FernUniversität in Hagen, Chair of Data Processing Technologie; Jiaotong-Universität Shanghai

URL: http://deposit.fernuni-hagen.de/2195/1/fy-thesis-with-CV.pdf

In order to tackle the “information overload” problem, researchers began to investigate various automated information filtering (IR) techniques that aim to select those information fragments out of large volumes of (dynamically generated) information that are most likely to meet the user’s information requirements. Since then, various methods ranging from content-based filtering (CBF) to collaborative filtering (CF), data mining (DM), and artificial intelligence (AI) have been developed. However, many web applications including e-learning, which lies in the focus of this thesis, exhibit inherent properties such as openness and distribution that are not addressed by existing solutions. They were designed with a centralized architecture in mind and do not scale well. In addition, learning behavior is a very complicated process that requires a more elaborate scheme than exists today to capture relevant user features for the purpose of matching learner interests. To accommodate these needs, this thesis proposes a novel personalized recommendation model and provides an operational framework with a high degree of generality and scalability through research on:

The theoretical results produced in this research have been applied to the real e-learning environment at Shanghai Jiao Tong University to evaluate their effectiveness in monitoring learning communities and recommending personalized resources in large-scale network education.