hide
Free keywords:
Cross-entropy Monte Carlo,
Kendall's τ,
Moderate deviation,
Partial list,
Random degeneration,
Rank aggregation,
Spearman's footrule,
Top-k ranked list
Abstract:
In this article, we describe a new approach that combines the estimation of the lengths of highly conforming sublists with their stochastic aggregation, to deal with two or more rankings of the same set of objects. The goal is to obtain a much smaller set of informative common objects in a new rank order. The input lists can be of large or huge size, their rankings irregular and incomplete due to random and missing assignments. A moderate deviation-based inference procedure and a cross-entropy Monte Carlo technique are used to handle the combinatorial complexity of the task. Two alternative distance measures are considered that can accommodate truncated list information. Finally, the outlined approach is applied to simulated data that was motivated by microarray meta-analysis, an important field of application.
