Zerber+R

摘要: Privacy-preserving document exchange among collaboration groups in an enterprise as well across enterprises requires techniques for sharing and search of access-controlled information through largely untrusted servers. In these settings systems need to provide confidentiality guarantees shared while offering IR properties comparable the ordinary engines. Top-k is a standard technique which enables fast query execution on very large indexes makes highly scalable. However, indexing top-k retrieval challenging task due sensitivity term statistics used ranking.In this paper we present Zerber+R -- ranking model allows privacy-preserving from outsourced inverted index. We propose relevance score transformation function scores different terms indistinguishable, such that even if stored server they do not reveal about indexed data. Experiments two real-world data sets show economical usage bandwidth offers with