kNN processing with co-space distance in SoLoMo systems

作者：

Highlights：

• We propose a new distance function, cospace distance, to measure the similarity between users.

• We design a progressive kNN search algorithm for the cospace distance.

• A new caching strategy is adopted to reduce the overhead of kNN processing.

• We use MapReduce and key-value store to support parellel processing of large social data.

摘要

•We propose a new distance function, cospace distance, to measure the similarity between users.•We design a progressive kNN search algorithm for the cospace distance.•A new caching strategy is adopted to reduce the overhead of kNN processing.•We use MapReduce and key-value store to support parellel processing of large social data.

论文关键词：kNN,SoLoMo,SVM,Crowdsourcing,Location-based search

论文评审过程：Available online 12 June 2014.

论文官网地址：https://doi.org/10.1016/j.eswa.2014.06.008