ITU

Hardness of Bichromatic Closest Pair with Jaccard Similarity

Research output: Conference Article in Proceeding or Book/Report chapterArticle in proceedingsResearchpeer-review

View graph of relations

Consider collections $\mathcal{A}$ and $\mathcal{B}$ of red and blue sets,respectively. Bichromatic Closest Pair is the problem of finding apair from $\mathcal{A}\times \mathcal{B}$ that has similarity higher than a given thresholdaccording to some similarity measure. Our focus here is the classic Jaccard similarity$|\textbf{a}\cap \textbf{b}|/|\textbf{a}\cup \textbf{b}|$ for $(\textbf{a},\textbf{b})\in \mathcal{A}\times \mathcal{B}$.
We consider the approximate version of the problem where we are given thresholds $j_1>j_2$ and wish to returna pair from $\mathcal{A}\times \mathcal{B}$ that has Jaccard similarity higher than$j_2$ if there exists a pair in $\mathcal{A}\times \mathcal{B}$ with Jaccard similarity atleast $j_1$. The classic locality sensitive hashing (LSH) algorithm of Indyk and Motwani (STOC '98), instantiated with the MinHash LSH function of Broder et al., solves this problem in $\tilde O(n^{2-\delta})$ time if $j_1\ge j_2^{1-\delta}$. In particular, for $\delta=\Omega(1)$, the approximation ratio $j_1/j_2=1/j_2^{\delta}$ increases polynomially in $1/j_2$.
In this paper we give a corresponding hardness result. Assuming theOrthogonal Vectors Conjecture (OVC), we show that there cannot be ageneral solution that solves the Bichromatic Closest Pair problem in $O(n^{2-\Omega(1)})$ timefor $j_1/j_2=1/j_2^{o(1)}$. Specifically, assuming OVC, we prove that forany $\delta>0$ there exists an $\varepsilon>0$ such that BichromaticClosest Pair with Jaccard similarity requires time$\Omega(n^{2-\delta})$ for any choice of thresholds$j_2<j_1<1-\delta$, that satisfy $j_1\le j_2^{1-\varepsilon}$.
Original languageEnglish
Title of host publicationLIPIcs - Leibniz International Proceedings in Informatics - 27th Annual European Symposium on Algorithms (ESA 2019)
Number of pages13
Volume144
PublisherSchloss Dagstuhl - Leibniz-Zentrum fuer Informatik GmbH
Publication date2019
Pages80:1–80:13
Article number80
ISBN (Electronic)978-3-95977-124-5
DOIs
Publication statusPublished - 2019
SeriesLeibniz International Proceedings in Informatics
ISSN1868-8969

Downloads

No data available

ID: 84765366