大规模数据的集合相似度估计研究进展

Research Progress of Set Similarity Estimation in Large Scale Data

相似度估计是实现近似集合检索的一个重要组成部分。针对于不用的相似度度量方式,主要有使用Jaccard相似度的Min Hash算法和基于海明距离的Sim Hash算法,它们是两种常用的大规模数据相似度度量技术。此外,还有基于集合包含度下的集合相似度估计算法。通过总结多篇已发表论文,详细地回顾了不同的相似度度量方式下的集合相似度估计方法。

Similarity estimation is an important component in approximate sets retrieval. For thesimilarity measure, there are mainly MinHash algorithm using Jaccard similarity and SimHash algo-rithm based on Hamming distance. They are two commonly used large -scale data similarity meas-urement techniques. In addition, there is a set similarity estimation algorithm based on set contain-ment. By summarizing a number of published papers, we review the similarity estimation method indifferent similarity measure.