Cryptology ePrint Archive: Report 2009/401

Longest Common Subsequence as Private Search

Mark Gondree and Payman Mohassel

Abstract: At STOC 2006 and CRYPTO 2007, Beimel et. al. introduced a set of privacy requirements for algorithms that solve search problems. In this paper, we consider the longest common subsequence (LCS) problem as a private search problem, where the task is to find a string of (or embedding corresponding to) an LCS. We show that deterministic selection strategies do not meet the privacy guarantees considered for private search problems and, in fact, may ``leak'' an amount of information proportional to the entire input.

We then put forth and investigate several privacy structures for the LCS problem and design new and efficient output sampling and equivalence protecting algorithms that provably meet the corresponding privacy notions. Along the way, we also provide output sampling and equivalence protecting algorithms for finite regular languages, which may be of independent interest.

Category / Keywords: applications / private search, genomic computation, longest common subsequence

Publication Info: Full version of a paper to appear at WPES 2009

Date: received 15 Aug 2009

Contact author: pmohasse at cpsc ucalgary ca

Available format(s): PDF | BibTeX Citation

Version: 20090817:121844 (All versions of this report)

Short URL: ia.cr/2009/401