Replace Scoring with Arrangement: A Contextual Set-to-Arrangement Framework for Learning-to-Rank

By J. Jin et al.

Published on Oct. 21, 2023

Read the original document by opening this link in a new tab.

Table of Contents

1. Introduction
2. Preliminaries
3. The STARank Framework

Summary

This document presents a new framework called STARank that aims to generate the permutations of candidate items in a learning-to-rank context. The framework consists of a Reader Module for encoding candidate items and a Plackett-Luce Module for arranging the items into a permutation. The document discusses the limitations of existing ranking methods and proposes a list-wise loss function to optimize the arrangement of items based on ground-truth permutations. STARank has been shown to outperform state-of-the-art methods in various ranking metrics.