Efficient top-k processing in large-scaled distributed environments

The rapid development of networking technologies has made it possible to construct a distributed database that involves a huge number of sites. Query processing in such a large-scaled system poses serious challenges beyond the scope of traditional distributed algorithms. In this paper, we propose a new algorithm BRANCA for performing top-k retrieval in these environments. Integrating two orthogonal methodologies “semantic caching” and “routing indexes”, BRANCA is able to solve a query by accessing only a small number of servers. Our algorithmic findings are accompanied with a solid theoretical analysis, which rigorously proves the effectiveness of BRANCA. Extensive experiments verify that our technique outperforms the existing methods significantly.