Master Thesis - 2011

Recently, as information networks widely spread, the number of contents over networks is increasing. For efficient and flexible information retrieval over such huge networks, agent technology receives much attention. We consider the case where retrieved information has difference of quality according to user’s request and can be scored. We proposed an agent execution control method for time-constrained information retrieval. It achieves to find better results by termination of an agent which has already acquired results with enough high quality or has less probability to improve the quality even if continuing retrieving. In this method, however, it is assumed that each agent has the identical time constraint. It leads a disparity in the obtained score between users who give individual time constraint. In this thesis, we propose a fair and efficient scheduling method based on expectation of improvement of the highest score (EIS). In the proposed method, all CPU resource is allocated to the agent which has the highest EIS. Finally, we evaluate the performance of the proposed method by simulation experiments. It is confirmed that the proposed method decreases the difference of each user’s score and increases the mean highest score of requested results.