Democracy is good for ranking

Towards multi-view rank learning and adaptation in web search

Wei Gao, Pei Yang

Research output: Chapter in Book/Report/Conference proceedingConference contribution

9 Citations (Scopus)

Abstract

Web search ranking models are learned from features originated from different views or perspectives of document relevancy, such as query dependent or independent features. This seems intuitively conformant to the principle of multi-view approach that leverages distinct complementary views to improve model learning. In this paper, we aim to obtain optimal separation of ranking features into non-overlapping subsets (i.e., views), and use such different views for rank learning and adaptation. We present a novel semi-supervised multi-view ranking model, which is then extended into an adaptive ranker for search domains where no training data exists. The core idea is to proactively strengthen view consistency (i.e., the consistency between different rankings each predicted by a distinct view-based ranker) especially when training and test data follow divergent distributions. For this purpose, we propose a unified framework based on listwise ranking scheme to mutually reinforce the view consistency of target queries and the appropriate weighting of source queries that act as prior knowledge. Based on LETOR and Yahoo Learning to Rank datasets, our method significantly outperforms some strong baselines including single-view ranking models commonly used and multi-view ranking models that do not impose view consistency on target data.

Original languageEnglish
Title of host publicationWSDM 2014 - Proceedings of the 7th ACM International Conference on Web Search and Data Mining
PublisherAssociation for Computing Machinery
Pages63-72
Number of pages10
ISBN (Print)9781450323512
DOIs
Publication statusPublished - 1 Jan 2014
Event7th ACM International Conference on Web Search and Data Mining, WSDM 2014 - New York, NY, United States
Duration: 24 Feb 201428 Feb 2014

Other

Other7th ACM International Conference on Web Search and Data Mining, WSDM 2014
CountryUnited States
CityNew York, NY
Period24/2/1428/2/14

Keywords

  • multi-view rank learning
  • rank adaptation
  • view consistency

ASJC Scopus subject areas

  • Computer Networks and Communications
  • Information Systems

Cite this

Gao, W., & Yang, P. (2014). Democracy is good for ranking: Towards multi-view rank learning and adaptation in web search. In WSDM 2014 - Proceedings of the 7th ACM International Conference on Web Search and Data Mining (pp. 63-72). Association for Computing Machinery. https://doi.org/10.1145/2556195.2556267

Democracy is good for ranking : Towards multi-view rank learning and adaptation in web search. / Gao, Wei; Yang, Pei.

WSDM 2014 - Proceedings of the 7th ACM International Conference on Web Search and Data Mining. Association for Computing Machinery, 2014. p. 63-72.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Gao, W & Yang, P 2014, Democracy is good for ranking: Towards multi-view rank learning and adaptation in web search. in WSDM 2014 - Proceedings of the 7th ACM International Conference on Web Search and Data Mining. Association for Computing Machinery, pp. 63-72, 7th ACM International Conference on Web Search and Data Mining, WSDM 2014, New York, NY, United States, 24/2/14. https://doi.org/10.1145/2556195.2556267
Gao W, Yang P. Democracy is good for ranking: Towards multi-view rank learning and adaptation in web search. In WSDM 2014 - Proceedings of the 7th ACM International Conference on Web Search and Data Mining. Association for Computing Machinery. 2014. p. 63-72 https://doi.org/10.1145/2556195.2556267
Gao, Wei ; Yang, Pei. / Democracy is good for ranking : Towards multi-view rank learning and adaptation in web search. WSDM 2014 - Proceedings of the 7th ACM International Conference on Web Search and Data Mining. Association for Computing Machinery, 2014. pp. 63-72
@inproceedings{5279b241d56740f69870a8d2dbf2df51,
title = "Democracy is good for ranking: Towards multi-view rank learning and adaptation in web search",
abstract = "Web search ranking models are learned from features originated from different views or perspectives of document relevancy, such as query dependent or independent features. This seems intuitively conformant to the principle of multi-view approach that leverages distinct complementary views to improve model learning. In this paper, we aim to obtain optimal separation of ranking features into non-overlapping subsets (i.e., views), and use such different views for rank learning and adaptation. We present a novel semi-supervised multi-view ranking model, which is then extended into an adaptive ranker for search domains where no training data exists. The core idea is to proactively strengthen view consistency (i.e., the consistency between different rankings each predicted by a distinct view-based ranker) especially when training and test data follow divergent distributions. For this purpose, we propose a unified framework based on listwise ranking scheme to mutually reinforce the view consistency of target queries and the appropriate weighting of source queries that act as prior knowledge. Based on LETOR and Yahoo Learning to Rank datasets, our method significantly outperforms some strong baselines including single-view ranking models commonly used and multi-view ranking models that do not impose view consistency on target data.",
keywords = "multi-view rank learning, rank adaptation, view consistency",
author = "Wei Gao and Pei Yang",
year = "2014",
month = "1",
day = "1",
doi = "10.1145/2556195.2556267",
language = "English",
isbn = "9781450323512",
pages = "63--72",
booktitle = "WSDM 2014 - Proceedings of the 7th ACM International Conference on Web Search and Data Mining",
publisher = "Association for Computing Machinery",

}

TY - GEN

T1 - Democracy is good for ranking

T2 - Towards multi-view rank learning and adaptation in web search

AU - Gao, Wei

AU - Yang, Pei

PY - 2014/1/1

Y1 - 2014/1/1

N2 - Web search ranking models are learned from features originated from different views or perspectives of document relevancy, such as query dependent or independent features. This seems intuitively conformant to the principle of multi-view approach that leverages distinct complementary views to improve model learning. In this paper, we aim to obtain optimal separation of ranking features into non-overlapping subsets (i.e., views), and use such different views for rank learning and adaptation. We present a novel semi-supervised multi-view ranking model, which is then extended into an adaptive ranker for search domains where no training data exists. The core idea is to proactively strengthen view consistency (i.e., the consistency between different rankings each predicted by a distinct view-based ranker) especially when training and test data follow divergent distributions. For this purpose, we propose a unified framework based on listwise ranking scheme to mutually reinforce the view consistency of target queries and the appropriate weighting of source queries that act as prior knowledge. Based on LETOR and Yahoo Learning to Rank datasets, our method significantly outperforms some strong baselines including single-view ranking models commonly used and multi-view ranking models that do not impose view consistency on target data.

AB - Web search ranking models are learned from features originated from different views or perspectives of document relevancy, such as query dependent or independent features. This seems intuitively conformant to the principle of multi-view approach that leverages distinct complementary views to improve model learning. In this paper, we aim to obtain optimal separation of ranking features into non-overlapping subsets (i.e., views), and use such different views for rank learning and adaptation. We present a novel semi-supervised multi-view ranking model, which is then extended into an adaptive ranker for search domains where no training data exists. The core idea is to proactively strengthen view consistency (i.e., the consistency between different rankings each predicted by a distinct view-based ranker) especially when training and test data follow divergent distributions. For this purpose, we propose a unified framework based on listwise ranking scheme to mutually reinforce the view consistency of target queries and the appropriate weighting of source queries that act as prior knowledge. Based on LETOR and Yahoo Learning to Rank datasets, our method significantly outperforms some strong baselines including single-view ranking models commonly used and multi-view ranking models that do not impose view consistency on target data.

KW - multi-view rank learning

KW - rank adaptation

KW - view consistency

UR - http://www.scopus.com/inward/record.url?scp=84898665654&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84898665654&partnerID=8YFLogxK

U2 - 10.1145/2556195.2556267

DO - 10.1145/2556195.2556267

M3 - Conference contribution

SN - 9781450323512

SP - 63

EP - 72

BT - WSDM 2014 - Proceedings of the 7th ACM International Conference on Web Search and Data Mining

PB - Association for Computing Machinery

ER -