Foundations and Trends
R
in
sample
Vol. xx, No xx (2008) 1–112
c
2008 xxxxxxxxx
DOI: xxxxxx
Learning to Rank for Information Retrieval
Tie-Yan Liu
1
1
Microsoft Research Asia, Sigma Center, No. 49, Zhichun Road, Haidian
District, Beijing, 100190, P. R. China, Tie-Yan.Liu@microsoft.com
Abstract
Learning to rank for information retrieval (IR) is a task to automat-
ically construct a ranking model using training data, such that the
model can sort new objects according to their degrees of relevance,
preference, or importance. Many IR problems are by nature ranking
problems, and many IR technologies can be potentially enhanced by
using learning-to-rank techniques. The objective of this tutorial is to
give an introduction to this research direction. Specifically, the exist-
ing learning-to-rank algorithms are reviewed and categorized into three
approaches: the pointwise, pairwise, and listwise approaches. The ad-
vantages and problems with each approach are analyzed, and the rela-
tionships between the loss functions used in these approaches and IR
evaluation measures are discussed. Then the empirical evaluations on
typical learning-to-rank methods are shown, with the LETOR collec-
tion as a benchmark dataset, which seem to suggest that the listwise ap-
proach be the most effective one among all the approaches. After that,
a statistical ranking theory is introduced, which can describe different
learning-to-rank algorithms, and be used to analyze their query-level
generalization abilities. At the end of the tutorial, we make a summary
and discuss potential future work on learning to rank.
剩余114页未读，继续阅读
评论4