Projects per year
Abstract
Diversity maximization is a fundamental problem with wide applications in data summarization, web search, and recommender systems. Given a set X of n elements, it asks to select a subset S of k << n elements with maximum diversity, as quantified by the dissimilarities among the elements in S. In this paper, we focus on the diversity maximization problem with fairness constraints in the streaming setting. Specifically, we consider the maxmin dispersion diversity objective, which aims to select the subset S that maximizes the minimum distance (dissimilarity) between any pair of distinct elements in S. Assuming that the set X is partitioned into m disjoint groups by some sensitive attribute, e.g., sex or race, ensuring fairness requires that the selected subset S contains k_i elements from each group i in [1,m]. A streaming algorithm should process X sequentially in one pass and return a subset with maximum diversity while guaranteeing the fairness constraint. Although diversity maximization has been extensively studied, the only known algorithms that can work with the maxmin diversity objective and fairness constraints are very inefficient for data streams. Since diversity maximization is NPhard in general, we propose two novel approximation algorithms for fair diversity maximization in data streams, the first of which is (1ε)/4approximate and specific for m=2, where ε in (0,1), and the second of which achieves a (1ε)/(3m+2)approximation for an arbitrary m. Experimental results show that both algorithms provide solutions of comparable quality to the stateoftheart algorithms while running several orders of magnitude faster in the streaming setting.
Original language  English 

Title of host publication  2022 IEEE 38TH INTERNATIONAL CONFERENCE ON DATA ENGINEERING (ICDE 2022) : ICDE 2022 
Number of pages  13 
Publisher  IEEE 
Publication date  2 Aug 2022 
Pages  4153 
ISBN (Print)  9781665408844 
ISBN (Electronic)  9781665408837 
DOIs  
Publication status  Published  2 Aug 2022 
MoE publication type  A4 Article in conference proceedings 
Event  IEEE International Conference on Data Engineering  Kuala Lumpur, Malaysia Duration: 9 May 2022 → 12 May 2022 Conference number: 38 
Publication series
Name  IEEE International Conference on Data Engineering 

Publisher  IEEE COMPUTER SOC 
ISSN (Print)  10844627 
Fields of Science
 113 Computer and information sciences
 algorithmic fairness
 diversity maximization
 maxmin dispersion
 streaming algorithm
 APPROXIMATION ALGORITHMS
Projects
 1 Active

MLDB: Model Management Systems: Machine learning meets Database Systems
Gionis, A., Mathioudakis, M., Pai, S. G., Svana, M. & Wang, Y.
Suomen Akatemia Projektilaskutus
01/09/2019 → 31/12/2023
Project: Suomen Akatemia: : Akatemiahanke