Degree Name

Master of Science (Hons.)


School of Information Technology and Computer Science


In this thesis, we introduce a new approach, agent-based intelligent information source selection, which is an alternative way for overcoming the problem of information selection from distributed information sources by using three artificial intelligence techniques, including query expansion with a Naive Bayes text classifier, intelligent information selection with case-based reasoning and adaptation to the dynamic web environment with reinforcement learning. My contribution to this research is to propose an intelligent environment where the Analysis Agent, Case-Matching Agent and Learning Agent, these three major agents iteratively work together to locate the most appropriate information sources to search so as to effectively and efficiently satisfy the user's expectation. We have finished the implementation of the first component - Analysis Agent. The experimental results show that it is possible to improve the effectiveness on both selection and retrieval stages in a distributed searching environment by using query expansion with a Naive Bayes text classifier.