Probabilistic models for personalizing Web search

David Sontag, Kevyn Collins-Thompson, Paul N. Bennett, Ryen W. White, Susan Dumais, Bodo Billerbeck

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

Abstract

We present a new approach for personalizing Web search results to a specific user. Ranking functions for Web search engines are typically trained by machine learning algorithms using either direct human relevance judgments or indirect judgments obtained from click-through data from millions of users. The rankings are thus optimized to this generic population of users, not to any specific user. We propose a generative model of relevance which can be used to infer the relevance of a document to a specific user for a search query. The user-specific parameters of this generative model constitute a compact user profile. We show how to learn these profiles from a user's long-term search history. Our algorithm for computing the personalized ranking is simple and has little computational overhead. We evaluate our personalization approach using historical search data from thousands of users of a major Web search engine. Our findings demonstrate gains in retrieval performance for queries with high ambiguity, with particularly large improvements for acronym queries.

Original languageEnglish (US)
Title of host publicationWSDM 2012 - Proceedings of the 5th ACM International Conference on Web Search and Data Mining
Pages433-442
Number of pages10
DOIs
StatePublished - 2012
Event5th ACM International Conference on Web Search and Data Mining, WSDM 2012 - Seattle, WA, United States
Duration: Feb 8 2012Feb 12 2012

Other

Other5th ACM International Conference on Web Search and Data Mining, WSDM 2012
CountryUnited States
CitySeattle, WA
Period2/8/122/12/12

Fingerprint

Search engines
Learning algorithms
Learning systems
Statistical Models

Keywords

  • Machine learning
  • Personalization
  • Probabilistic models
  • Re-ranking

ASJC Scopus subject areas

  • Computer Networks and Communications

Cite this

Sontag, D., Collins-Thompson, K., Bennett, P. N., White, R. W., Dumais, S., & Billerbeck, B. (2012). Probabilistic models for personalizing Web search. In WSDM 2012 - Proceedings of the 5th ACM International Conference on Web Search and Data Mining (pp. 433-442) https://doi.org/10.1145/2124295.2124348

Probabilistic models for personalizing Web search. / Sontag, David; Collins-Thompson, Kevyn; Bennett, Paul N.; White, Ryen W.; Dumais, Susan; Billerbeck, Bodo.

WSDM 2012 - Proceedings of the 5th ACM International Conference on Web Search and Data Mining. 2012. p. 433-442.

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

Sontag, D, Collins-Thompson, K, Bennett, PN, White, RW, Dumais, S & Billerbeck, B 2012, Probabilistic models for personalizing Web search. in WSDM 2012 - Proceedings of the 5th ACM International Conference on Web Search and Data Mining. pp. 433-442, 5th ACM International Conference on Web Search and Data Mining, WSDM 2012, Seattle, WA, United States, 2/8/12. https://doi.org/10.1145/2124295.2124348
Sontag D, Collins-Thompson K, Bennett PN, White RW, Dumais S, Billerbeck B. Probabilistic models for personalizing Web search. In WSDM 2012 - Proceedings of the 5th ACM International Conference on Web Search and Data Mining. 2012. p. 433-442 https://doi.org/10.1145/2124295.2124348
Sontag, David ; Collins-Thompson, Kevyn ; Bennett, Paul N. ; White, Ryen W. ; Dumais, Susan ; Billerbeck, Bodo. / Probabilistic models for personalizing Web search. WSDM 2012 - Proceedings of the 5th ACM International Conference on Web Search and Data Mining. 2012. pp. 433-442
@inproceedings{65067c9ce5294b0a87f5d7277246c088,
title = "Probabilistic models for personalizing Web search",
abstract = "We present a new approach for personalizing Web search results to a specific user. Ranking functions for Web search engines are typically trained by machine learning algorithms using either direct human relevance judgments or indirect judgments obtained from click-through data from millions of users. The rankings are thus optimized to this generic population of users, not to any specific user. We propose a generative model of relevance which can be used to infer the relevance of a document to a specific user for a search query. The user-specific parameters of this generative model constitute a compact user profile. We show how to learn these profiles from a user's long-term search history. Our algorithm for computing the personalized ranking is simple and has little computational overhead. We evaluate our personalization approach using historical search data from thousands of users of a major Web search engine. Our findings demonstrate gains in retrieval performance for queries with high ambiguity, with particularly large improvements for acronym queries.",
keywords = "Machine learning, Personalization, Probabilistic models, Re-ranking",
author = "David Sontag and Kevyn Collins-Thompson and Bennett, {Paul N.} and White, {Ryen W.} and Susan Dumais and Bodo Billerbeck",
year = "2012",
doi = "10.1145/2124295.2124348",
language = "English (US)",
isbn = "9781450307475",
pages = "433--442",
booktitle = "WSDM 2012 - Proceedings of the 5th ACM International Conference on Web Search and Data Mining",

}

TY - GEN

T1 - Probabilistic models for personalizing Web search

AU - Sontag, David

AU - Collins-Thompson, Kevyn

AU - Bennett, Paul N.

AU - White, Ryen W.

AU - Dumais, Susan

AU - Billerbeck, Bodo

PY - 2012

Y1 - 2012

N2 - We present a new approach for personalizing Web search results to a specific user. Ranking functions for Web search engines are typically trained by machine learning algorithms using either direct human relevance judgments or indirect judgments obtained from click-through data from millions of users. The rankings are thus optimized to this generic population of users, not to any specific user. We propose a generative model of relevance which can be used to infer the relevance of a document to a specific user for a search query. The user-specific parameters of this generative model constitute a compact user profile. We show how to learn these profiles from a user's long-term search history. Our algorithm for computing the personalized ranking is simple and has little computational overhead. We evaluate our personalization approach using historical search data from thousands of users of a major Web search engine. Our findings demonstrate gains in retrieval performance for queries with high ambiguity, with particularly large improvements for acronym queries.

AB - We present a new approach for personalizing Web search results to a specific user. Ranking functions for Web search engines are typically trained by machine learning algorithms using either direct human relevance judgments or indirect judgments obtained from click-through data from millions of users. The rankings are thus optimized to this generic population of users, not to any specific user. We propose a generative model of relevance which can be used to infer the relevance of a document to a specific user for a search query. The user-specific parameters of this generative model constitute a compact user profile. We show how to learn these profiles from a user's long-term search history. Our algorithm for computing the personalized ranking is simple and has little computational overhead. We evaluate our personalization approach using historical search data from thousands of users of a major Web search engine. Our findings demonstrate gains in retrieval performance for queries with high ambiguity, with particularly large improvements for acronym queries.

KW - Machine learning

KW - Personalization

KW - Probabilistic models

KW - Re-ranking

UR - http://www.scopus.com/inward/record.url?scp=84858033182&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84858033182&partnerID=8YFLogxK

U2 - 10.1145/2124295.2124348

DO - 10.1145/2124295.2124348

M3 - Conference contribution

SN - 9781450307475

SP - 433

EP - 442

BT - WSDM 2012 - Proceedings of the 5th ACM International Conference on Web Search and Data Mining

ER -