Parallel and Distributed Intelligent Systems

点击翻译按钮获取中文摘要

PARALLEL AND DISTRIBUTED INTELLIGENT SYSTEMS The long-term objective of our work has been to develop efficient models, algorithms and implementations of parallel and distributed intelligent systems. Our goal has been not only to develop and experiment with new algorithms, but to develop operational software systems that can be employed in the real world as well. Whenever opportunities arise, the outcomes of our research will be commercialized and integrated into products. We propose to generalize the concept of the weighted tree similarity proposed earlier by us, to the weighted graph similarity and exploit it to carry out semantic matching, matchmaking and semantic search. Graph algorithms are more complex than tree algorithms. Therefore, in order to handle large data sets, we will develop parallel and distributed algorithms suitable for multi-core processors, clusters, grids and clouds to drastically reduce computing time. Our proposed work, which generalizes our previous work to provide more precise matching, has even greater potential for real world applications. The applications in e-Business would result in much more precise matching than simple keyword matches used in Kijiji and similar systems. In e-Health domain, the proposed approach promises better social networking among patients and more precise results from querying patient record histories. Finally, the proposed approach will give highly relevant results to user queries compared to simple keyword-based searches used in search engines like Google.