摘要： k最近邻搜索算法无法满足数据挖掘的分布性、实时性和可扩展性要求，针对该问题提出基于P2P的自适应分布式k最近邻搜索算法(P2PAKNNs)。阐述GHT*结构，定义高维数据相似度函数HDSF(X,Y)，论述GHT*中的插入算法、范围查找算法和搜索算法。给出P2PAKNNs的实现过程，通过实验证明其正确性。

关键词: k最近邻搜索算法, 度量空间, 相似性查询

Abstract: k-nearest Neighbor search algorithm(KNNs) can not satisfy the needs of distributing, real time performance and expansibility for data mining. Aiming at this problem, a P2P-based self-adaptive distributed KNNs(P2PAKNNs) is proposed. This paper expounds GHT* structure, and gives similarity measure function HDSF(X, Y). Insert algorithm, range find algorithm and search algorithm in GHT* are discussed. Implementation process of P2PAKNNs is given, and its correctness is validated by experiment.

Key words: k-nearest Neighbor search algorithm(KNNs), metric space, similarity query

中图分类号: 

• TP391