关键词: top-k算法, 分布式网络, histogram blooms技术

Abstract: A new algorithm FT to answer top-k queries（find the k objects with the highest aggregate scores） in a distributed network is proposed.Prior research such as Threshlod Algorithm（TA），Three-Phase Uniform-Threshold（TPUT） algorithm and Hybrid-Threshold（HT） algorithm consume an excessive amount of bandwidth.KLEE algorithm can reduce network bandwidth consumption but it can’t give exact top-k answers.The algorithm FT can both reduce network bandwidth and give the exact top-k answer based on the pretreatment in the first phase and the histogram bloom.This paper implements FT and related algorithms and conducts a comprehensive performance evaluation.Evaluation employs real-world and synthetic data sets.The experiment shows that FT can achieve major performance in terms of network bandwidth.

Key words: top-k algorithm, distributed networks, histogram blooms