Learning to Rank for Information Retrieval

Overview

contents

1 Introduction
1.1 Ranking in IR
1.2 Learning to Rank
1.3 About this Tutorial
2 The Pointwise Approach
2.1 Regression based Algorithms
2.2 Classification based Algorithms
2.3 Ordinal Regression based Algorithms
2.4 Discussions
3 The Pairwise Approach
3.1 Example Algorithms
3.2 Discussions
4 The Listwise Approach
4.1 Direct Optimization of IR Evaluation Measures
4.2 Minimization of Listwise Ranking Losses
4.3 Discussions
5 Analysis of the Approaches
5.1 The Pointwise Approach
5.2 The Pairwise Approach
5.3 The Listwise Approach
5.4 Discussions
6 Benchmarking Learning-to-Rank Algorithms
6.1 The LETOR Collection
6.2 Experimental Results on LETOR
7 Statistical Ranking Theory
7.1 Conventional Generalization Analyses on Ranking
7.2 A Query-level Ranking Framework
7.3 Query-level Generalization Analysis
7.4 Discussions
8 Summary and Outlook
Acknowledgments References