A General Boosting Method and its Application to Learning Ranking Functions for Web Search

By Zhaohui Zheng, Hongyuan Zha, Tong Zhang, Olivier Chapelle, Keke Chen, Gordon Sun. 2007

228 citations recorded by the source corpus.

How researchers see this paper

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

Citations in the corpus

Counterfactual Estimation and Optimization of Click Metrics for Search Engines (2014, background): “Such an approach has been highly successful at facilitating easy comparison and improvement of ranking functions ( e.g. , [6, 32, 34]).”
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].”
The scoring of poses in protein-protein docking: current capabilities and future directions (2013, background): “Similarly a pairwise ranking approach can be formulated as pairwise classification or pairwise regression [129-136], for instance where a model is trained on restraints derived from the fact that acceptable docking poses are superior to incorrect poses, medium are superior to acceptable and incorrec”
Towards recency ranking in web search (2010, methodology): “Machine-learned ranking refers to the automatic construction of a ranking function which optimizes retrieval performance metrics [4, 5, 9, 10, 12, 21, 23 , 16].”
Web-Search Ranking with Initialized Gradient Boosted Regression Trees (2010, methodology): “GBRT depends on three parameters: The learning rate α > 0, the tree-depth d, and the number of iterations MB. Based on experiments from previous research (Friedman, 2001; Zheng et al., 2007b), we set d = 4 and pick MB and α with a validation data set.”
The Greedy Miser: Learning under Test-time Budgets (2012, background): “The applications range from web-search engines (Zheng et al., 2008), over product recommendation (Fleck et al., 1996), to email and web spam ﬁltering (Weinberger et al., 2009).”
Learning to Rank for Information Retrieval and Natural Language Processing (2011, unknown): “…[Freund et al 2003]: Pairwise Classification Using Boosting – RankNet [Burges et al 2005]: Pairwise Classification Using Neural Net – Frank [Tsai et al 2007]: Pairwise Classification Using Fidelity Loss and Neural Net – GBRank [Zheng et al 2007]: Pairwise Regression Using Boosting Tree – IR…”
Early exit optimizations for additive machine learned ranking systems (2010, methodology): “Documents are evaluated using a learning system based on gradient boosted decision trees [33], composed of N =1200 scorers (decision trees).”
Learning to model relatedness for news recommendation (2011, methodology): “4.2 Learning a Relatedness Function Given the features described in Section 3 and editorial data obtained as in Section 4.1, we use a state-of-the-art Relatedness Seed news Candidate news Very related Polish leader, 96 others dead in Russia jet crash List of some of those who died in the crash o”
The P-Norm Push: A Simple Convex Ranking Algorithm that Concentrates at the Top of the List (2009, methodology): “…(and large deviation bounds) for supervised ranking, namely, the bounds of Freund et al. (2003), Clemençon et al. (2008), Agarwal et al. (2005), Usunier et al. (2005), Hill et al. (2002), Rudin et al. (2005) and Rudin and Schapire (2009), though we were only able to adapt techniques from the…”
Multi-task learning for boosting with application to web search ranking (2010, methodology): “This is motivated by our interest in web search ranking: gradient boosted decision trees are indeed among the state-of-the-art algorithms for largescale web-search ranking [11, 19].”
Non-Linear Domain Adaptation with Boosting (2013, methodology): “We propose to use gradient boosting [8, 9] to solve for ft(·).”
Toward Predicting the Outcome of an A/B Experiment for Search Relevance (2015, background): “In practice, π is often a deterministic function [4, 29].”
Refining Recency Search Results with User Click Feedback (2011, background): “Machine-learned ranking algorithms have been shown highly eﬀective for generalizing to unseen data from labeled training data and have been very successful especially in commercial Web search engines; see [18, 3, 4, 12, 6, 30, 23] and many references therein for more reference.”
Yahoo! Learning to Rank Challenge Overview (2010, methodology): “GBRank (Zheng et al., 2008) is similar to RankSVM, but uses a quadratic penalization, (t) = max(0, 1 − t) 2 and is combined with functional gradient boosting.”
The BellKor Solution to the Netflix Grand Prize (2009, background): “This has made them a good choice for several large scale practical problems such as ranking results of a search engine [9, 17], or query-biased summarization of search results [10].”