eng
Schloss Dagstuhl – Leibniz-Zentrum für Informatik
Leibniz International Proceedings in Informatics
1868-8969
2023-12-12
12:1
12:21
10.4230/LIPIcs.FSTTCS.2023.12
article
Approximate Maximum Rank Aggregation: Beyond the Worst-Case
Alvin, Yan Hong Yao
1
Chakraborty, Diptarka
1
National University of Singapore, Singapore
The fundamental task of rank aggregation is to combine multiple rankings on a group of candidates into a single ranking to mitigate biases inherent in individual input rankings. This task has a myriad of applications, such as in social choice theory, collaborative filtering, web search, statistics, databases, sports, and admission systems. One popular version of this task, maximum rank aggregation (or the center ranking problem), aims to find a ranking (not necessarily from the input set) that minimizes the maximum distance to the input rankings. However, even for four input rankings, this problem is NP-hard (Dwork et al., WWW'01, and Biedl et al., Discrete Math.'09), and only a (folklore) polynomial-time 2-approximation algorithm is known for finding an optimal aggregate ranking under the commonly used Kendall-tau distance metric. Achieving a better approximation factor in polynomial time, ideally, a polynomial time approximation scheme (PTAS), is one of the major challenges.
This paper presents significant progress in solving this problem by considering the Mallows model, a classical probabilistic model. Our proposed algorithm outputs an (1+ε)-approximate aggregate ranking for any ε > 0, with high probability, as long as the input rankings come from a Mallows model, even in a streaming fashion. Furthermore, the same approximation guarantee is achieved even in the presence of outliers, presumably a more challenging task.
https://drops.dagstuhl.de/storage/00lipics/lipics-vol284-fsttcs2023/LIPIcs.FSTTCS.2023.12/LIPIcs.FSTTCS.2023.12.pdf
Rank Aggregation
Center Problem
Mallows Model
Approximation Algorithms
Clustering with Outliers