Adapting boosting for information retrieval measures

By Qiang Wu, Qiang Wu, C. Burges, Krysta Marie Svore, Jianfeng Gao. 2010

463 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Information Systems, Information Retrieval & Search, Optimization & Training

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Jianfeng Gao
2. Michel Galley
3. Lihong Li
4. Jason Aaron Edward Weston
5. Harry Shum
6. Di Li
7. Minlie Huang
8. Mohit Bansal

In Information Systems

1. Harry Shum
2. Jianfeng Gao
3. Michel Galley
4. Di Li
5. Li Zhou
6. Lihong Li
7. Xiaodong He
8. Li Deng

In Information Retrieval & Search

1. Michel Galley
2. Jianfeng Gao
3. Lihong Li
4. Xiaodong He
5. Li Deng
6. Wen-tau Yih
7. Jimmy J. Lin
8. Bhaskar Mitra

In Optimization & Training

1. Harry Shum
2. Jianfeng Gao
3. Michel Galley
4. Lihong Li

Citations in the corpus

• TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehension (2017, neutral): “The function score is learnt using LambdaMART (Wu et al., 2010), 6 a boosted tree based ranking algorithm.”
• A Thorough Examination of the CNN/Daily Mail Reading Comprehension Task (2016, neutral): “For training our conventional classifier, we use the implementation of LambdaMART (Wu et al., 2010) in the RankLib package.”
• Neural Approaches to Conversational AI (2018, positive): “ney buy lucasfilms?”, a set of candidate documents are retrieved from Web Index via a fast, primary ranker. Then in the Document Ranking module, a sophisticated document ranker based on boosted trees (Wu et al., 2010) is used to assign relevance scores for these documents. The top-ranked relevant doc”
• The Design and Implementation of XiaoIce, an Empathetic Social Chatbot (2018, positive): “The response candidates generated by three generators are aggregated and ranked using a boosted tree ranker (Wu et al. 2010).”
• A Hierarchical Recurrent Encoder-Decoder for Generative Context-Aware Query Suggestion (2015, neutral): “Learning to Rank Challenge in 2010 [41].”
• Neural information retrieval: at the end of the early years (2017, unknown): “To recommend possible query expansion, they use LambdaMART (Wu et al. 2010) with two sets of ranking features.”
• CatBoost: unbiased boosting with categorical features (2017, unknown): “For a number of years, it has remained the primary method for learning problems with heterogeneous features, noisy data, and complex dependencies: web search, recommendation systems, weather forecasting, and many others [5, 19, 21, 23].”
• Collaborative Intent Prediction with Real-Time Contextual Data (2017, unknown): “BoostedTree (Wu et al. 2010) is an ensemble of regression trees (decision trees).”
• Convolutional Neural Networks for Soft-Matching N-Grams in Ad-hoc Search (2018, unknown): “One can leverage the power of any learning-to-rank model such as RankSVM [16] or LambdaMART [27].”
• A Short Introduction to Learning to Rank (2011, unknown): “The pairwise approach includes Ranking SVM [7], RankBoost [8], RankNet [9], GBRank [14], IR SVM [15], Lambda Rank [16], and LambdaMART [17].”
• Collective Entity Disambiguation with Structured Gradient Tree Boosting (2018, unknown): “Structured gradient tree boosting Gradient tree boosting has been used in some of the most accurate systems for a variety of classification and regression problems (Babenko et al., 2011; Wu et al., 2010; Yamada et al., 2016).”
• Seq2Slate: Re-ranking and Slate Optimization with RNNs (2018, unknown): “Using the generated training data, we train both our seq2slate model and baseline rankers from the RankLib package: AdaRank [Xu and Li, 2007], Coordinate Ascent [Metzler and Croft, 2007], LambdaMART [Wu et al., 2010], ListNet [Cao et al., 2007], MART [Friedman, 2001], Random Forests [Breiman, 2001],”
• Learning to Rank for Information Retrieval and Natural Language Processing (2011, unknown): “IR SVM [13] GBRank [115] Frank [97] LambdaMART [102] LambdaRank [12] Listwise SVM MAP [111] AdaRank [108] ListNet [14] SoftRank [95]”
• Machine learning and AI in marketing – Connecting computing power to human insights (2020, unknown): “The search ranking component used the LambdaMART method (Wu, Burges, Svore, & Gao, 2010) which combines the lambda ranking algorithm with multiple additive regression trees.”
• Asking Clarifying Questions in Open-Domain Information-Seeking Conversations (2019, unknown): “– LambdaMART, RankNet: we consider the task of question selection as a ranking problem where a list of candidate questions should be ranked and the one with the highest rank is chosen.”
• From RankNet to LambdaRank to LambdaMART: An Overview (2010, unknown): “Clearly, since MART models derivatives, and since LambdaRank works by specifying the derivatives at any point during training, the two algorithms are well matched: LambdaMART is the marriage of the two [11].”
• CatBoost: gradient boosting with categorical features support (2018, unknown): “For a number of years, it has remained the primary method for learning problems with heterogeneous features, noisy data, and complex dependencies: web search, recommendation systems, weather forecasting, and many others [2, 15, 17, 18].”