Supporting ranked search in parallel search cluster networks
Fang Xiong, Qiong Luo, Dyce Jing Zhao · 2005
Recent work by Cooper et al. proposed the Parallel Search Cluster Network (PSCN) as an efficient P2P network overlay. Organized in clusters, a PSCN allows peers within one cluster to share query workload, and peers across clusters to share indexes. In this paper, we study the problem of supporting ranked keyword search in a PSCN. Because ranking mechanisms, such as TF×IDF, require global information, we investigate how to acquire and distribute the global information in a PSCN. It turns out that this process can be done efficiently by taking advantage of the architectural features of the PSCN. We compare ranked search in a PSCN with that in an unstructured network as well as in a super-peer network, and the results show that our approach is feasible and efficient. 1