Exploiting contextual knowledge for information retrieval in a multiagent-based digital library service system

摘要

In this paper, we propose a generic design of a multi-agent digital library information service system that provides an interactive and guided search process to facilitate information retrieval. The multi-agent architecture allows modular functions and flexible interactions to develop increasingly sophisticated services. We built a prototype of the multi-agent information service system on the retrieval of offical gazettes. We have also evaluated the performance of the system on many real user queries and found encouraging results. Recently, we are extending the system design to a multimedia information retrieval system based on the historical multimedia domain of The First Emperor of China as a starting point in the IDLP project as a collaborator of a major US NSF International Digital Library Project, Chinese Memory Net (CMNet). Our goal is to develop an agent-mediated retrieval system for historic literature and media.