This paper is concerned with learning to rank for information retrieval (IR). Ranking is the central problem for information retrieval, and employing machine learning techniques to learn the ranking function is viewed as a promising approach to IR. Unfortunately, there was no benchmark dataset that could be used in comparison of existing learning algorithms and in evaluation of newly proposed algorithms, which stood in the way of the related research. To deal with the problem, we have constructed a benchmark dataset referred to as LETOR and distributed it to the research communities. Specifically we have derived the LETOR data from the existing data sets widely used in IR, namely, OHSUMED and TREC data. The two collections contain queries, the contents of the retrieved documents, and human judgments on the relevance of the documents with respect to the queries. We have extracted features from the datasets, including both conventional features, such as term frequency, inverse document frequency, BM25, and language models for IR, and features proposed recently at SIGIR, such as HostRank, feature propagation, and topical PageRank. We have then packaged LETOR with the extracted features, queries, and relevance judgments. We have also provided the results of several state-ofthe-arts learning to rank algorithms on the data. This paper describes in details about LETOR.

This paper aims to conduct a study on the listwise approach to learning to rank. The listwise approach learns a ranking function by taking individual lists as instances and minimizing a loss function defined on the predicted list and the ground-truth list. Existing work on the approach mainly focused on the development of new algorithms; methods such as RankCosine and ListNet have been proposed and good performances by them have been observed. Unfortunately, the underlying theory was not sufficiently studied so far. To amend the problem, this paper proposes conducting theoretical analysis of learning to rank algorithms through investigations on the properties of the loss functions, including consistency, soundness, continuity, differentiability, convexity, and efficiency. A sufficient condition on consistency for ranking is given, which seems to be the first such result obtained in related research. The paper then conducts analysis on three loss functions: likelihood loss, cosine loss, and cross entropy loss. The latter two were used in RankCosine and ListNet. The use of the likelihood loss leads to the development of

A machine learning approach to learning to rank trains a model to optimize a target evaluation measure with repect to training data. Currently, existing information retrieval measures are impossible to optimize directly except for models with a very small number of parameters. The IR community thus faces a major challenge: how to optimize IR measures of interest directly. In this paper, we present a solution. Specifically, we show that LambdaRank [1], which smoothly approximates the gradient of the target measure, can be adapted to work with four popular IR target evaluation measures using the same underlying gradient construction. It is likely, therefore, that this construction is extendable to other evaluation measures. We empirically show that LambdaRank finds a locally optimal solution for mean NDCG@10, mean NDCG, MAP and MRR with a 99% confidence rate. We also show that the amount of effective training data varies with IR measure and that with a sufficiently large training set size, matching the training optimization measure to the target evaluation measure yields the best accuracy.

Learning to rank is a new statistical learning technology on creating a ranking model for sorting objects. The technology has been successfully applied to web search, and is becoming one of the key machineries for building search engines. Existing approaches to learning to rank, however, did not consider the cases in which there exists relationship between the objects to be ranked, despite of the fact that such situations are very common in practice. For example, in web search, given a query certain relationships usually exist among the the retrieved documents, e.g., URL hierarchy, similarity, etc., and sometimes it is necessary to utilize the information in ranking of the documents. This paper addresses the issue and formulates it as a novel learning problem, referred to as, ‘learning to rank relational objects’. In the new learning

One of the central issues in learning to rank for information retrieval is to develop algorithms that construct ranking models by directly optimizing evaluation measures used in information retrieval such as Mean Average Precision (MAP) and Normalized Discounted Cumulative Gain (NDCG). Several such algorithms including SVM map and AdaRank have been proposed and their effectiveness has been verified. However, the relationships between the algorithms are not clear, and furthermore no comparisons have been conducted between them. In this paper, we conduct a study on the approach of directly optimizing evaluation measures in learning to rank for Information Retrieval (IR). We focus on the methods that minimize loss functions upper bounding the basic loss function defined on the IR measures. We first provide a general framework for the study and analyze the existing algorithms of SVM map and AdaRank within the framework. The framework is based on upper bound analysis and two types of upper bounds are discussed. Moreover, we show that we can derive new algorithms on the basis of this analysis and create one example algorithm called PermuRank. We have also conducted comparisons between SVM map, AdaRank, PermuRank, and conventional methods of Ranking SVM and Rank-Boost, using benchmark datasets. Experimental results show that the methods based on direct optimization of evaluation measures can always outperform conventional methods of Ranking SVM and RankBoost. However, no significant difference exists among the performances of the direct optimization methods themselves.

We consider the problem of ranking refinement, i.e., to improve the accuracy of an existing ranking function with a small set of labeled instances. We are, particularly, interested in learning a better ranking function using two complementary sources of information, ranking information given by the existing ranking function (i.e., the base ranker) and that obtained from users ’ feedbacks. This problem is very important in information retrieval where feedbacks are gradually collected. The key challenge in combining the two sources of information arises from the fact that the ranking information presented by the base ranker tends to be imperfect and the ranking information obtained from users’ feedbacks tends to be noisy. We present a novel boosting algorithm for ranking refinement that can effectively leverage the uses of the two sources of information. Our empirical study shows that the proposed algorithm is effective for ranking refinement, and furthermore it significantly outperforms the baseline algorithms that incorporate the outputs from the base ranker as an additional feature.

LETOR is a benchmark collection for the research on learning to rank for information retrieval, released by Microsoft Research Asia. In this paper, we describe the details of the LETOR collection and show how it can be used in different kinds of researches. Specifically, we describe how the document corpora and query sets in LETOR are selected, how the documents are sampled, how the learning features and meta information are extracted, and how the datasets are partitioned for comprehensive evaluation. We then compare several state-of-the-art learning to rank algorithms on LETOR, report their ranking performances, and make discussions on the results. After that, we discuss possible new research topics that can be supported by LETOR, in addition to algorithm comparison. We hope that this paper can help people to gain deeper understanding of LETOR, and enable more interesting research projects on learning to rank and related topics.

Capturing the context of a user’s query from the previous queries and clicks in the same session may help understand the user’s information need. A context-aware approach to document re-ranking, query suggestion, and URL recommendation may improve users ’ search experience substantially. In this paper, we propose a general approach to context-aware search. To capture contexts of queries, we learn a variable length Hidden Markov Model (vlHMM) from search sessions extracted from log data. Although the mathematical model is intuitive, how to learn a large vlHMM with millions of states from hundreds of millions of search sessions poses a grand challenge. We develop a strategy for parameter initialization in vlHMM learning which can greatly reduce the number of parameters to be estimated in practice. We also devise a method for distributed vlHMM learning under the map-reduce model. We test our approach on a real data set consisting of 1.8 billion queries, 2.6 billion clicks, and 840 million search sessions, and evaluate the effectiveness of the vlHMM learned from the real data on three search applications: document re-ranking, query suggestion, and URL recommendation. The experimental results show that our approach is both effective and efficient.

One shortfall of existing machine learning (ML) methods when applied to information retrieval (IR) is the inability to directly optimize for typical IR performance measures. This is in part due to the discrete nature, and thus non-differentiability, of these measures. When cast as an optimization problem, many methods require computing the gradient. In this paper, we explore conditions where the gradient might be numerically estimated. We use Simultaneous Perturbation Stochastic Approximation as our gradient approximation method. We also examine the empirical optimality of LambdaRank, which has performed very well in practice. 1

Learning to rank is a relatively new field of study, aiming to learn a ranking function from a set of training data with relevancy labels. The ranking algorithms are often evaluated using information retrieval measures, such as Normalized Discounted Cumulative Gain (NDCG) [1] and Mean Average Precision (MAP) [2]. Until recently, most learning to rank algorithms were not using a loss function related to the above mentioned evaluation measures. The main difficulty in direct optimization of these measures is that they depend on the ranks of documents, not the numerical values output by the ranking function. We propose a probabilistic framework that addresses this challenge by optimizing the expectation of NDCG over all the possible permutations of documents. A relaxation strategy is used to approximate the average of NDCG over the space of permutation, and a bound optimization approach is proposed to make the computation efficient. Extensive experiments show that the proposed algorithm outperforms state-of-the-art ranking algorithms on several benchmark data sets. 1