What problem does this paper attempt to address?

The problem that this paper attempts to solve is the multi - class minmax ranking aggregation problem. Specifically, the author introduced a new class of minmax ranking aggregation problems based on Kendall τ - distance and Spearman footrule distance, and proposed a series of constant - approximation algorithms to solve these problems. ### Problem Background Ranking data has received extensive attention in the field of machine learning, especially in applications such as recommendation systems, social voting, and product ranking platforms. Rank aggregation is one of the main tasks for processing ranking data, which usually involves calculating the median of a set of permutations or partial rankings and evaluating using an appropriate distance function. ### Specific Problem Description The minmax ranking aggregation problem proposed in this paper can be described as follows: Given a set of rankings of multiple classes, find a ranking π such that it minimizes the maximum weighted distance to the rankings in different classes. Formally, the optimization problem can be expressed as: \[ \text{MinMax: } \min_{\pi} \max_{k} \lambda_k d(\pi, \Sigma_k) \] where: - \( \lambda_k>0 \) represents the cost of violating the ranking consistency of the k - th class. - \( d(\pi, \Sigma_k) \) represents the distance between the ranking π and the set of rankings Σ_k in class k, and the median distance or the minimum distance can be selected. ### Problem Challenges Since this problem has been proven to be NP - hard, the author proposed several constant - approximation algorithms to solve this problem. These algorithms include: 1. **Pick - Rnd - Perm** and **Pick - Opt - Perm** algorithms for solving the problem under the median distance. 2. **mmKT - Conv** and **mmSP - Conv** algorithms, which solve more complex cases through convex optimization and rounding methods. 3. **min - Pick - Perm** algorithm for solving the problem under the minimum distance. ### Application Examples The author also demonstrated the applications of these aggregation methods on the Mallows model and genomic data, especially for the ancestral genome reconstruction problem of mitochondrial DNA (mtDNA). In conclusion, this paper aims to solve the multi - class minmax ranking aggregation problem and provides a series of effective approximation algorithms to deal with the complexity and computational difficulty in practical applications.

Multiclass MinMax Rank Aggregation

Fair Rank Aggregation

Multivariate Spearman's rho for aggregating ranks using copulas

Approximate and Dynamic Rank Aggregation

Approximate Rank Aggregation

A New Probabilistic Model for Rank Aggregation.

The complexity of MinRank

A New Weighted Spearman's Footrule as A Measure of Distance between Rankings

Partition-Mallows Model and Its Inference for Rank Aggregation

Maximizing Agreements for Ranking, Clustering and Hierarchical Clustering via MAX-CUT

Diversity in Kemeny Rank Aggregation: A Parameterized Approach

Supervised rank aggregation

Faster Algorithms for Fair Max-Min Diversification in $\mathbb{R}^d$

Worst-Case Discriminative Feature Learning via Max-Min Ratio Analysis

Bayesian Aggregation of Order-Based Rank Data

Towards Non-Parametric Learning to Rank

Heuristic Search for Rank Aggregation with Application to Label Ranking

Monotone Retargeting for Unsupervised Rank Aggregation with Object Features

Rank Aggregation via Nuclear Norm Minimization

Adaptive Low-Nonnegative-Rank Approximation for State Aggregation of Markov Chains

Spectral MLE: Top-$K$ Rank Aggregation from Pairwise Comparisons