### Abstract

This paper examines in detail an alternative ranking problem for search engines, movie recommendation, and other similar ranking systems motivated by the requirement to not just accurately predict pairwise ordering but also preserve the magnitude of the preferences or the difference between ratings. We describe and analyze several cost functions for this learning problem and give stability bounds for their generalization error, extending previously known stability results to non-bipartite ranking and magnitude of preference- preserving algorithms. We present algorithms optimizing these cost functions, and, in one instance, detail both a batch and an on-line version. For this algorithm, we also show how the leave-one-out error can be computed and approximated efficiently, which can be used to determine the optimal values of the trade-off parameter in the cost function. We report the results of experiments comparing these algorithms on several datasets and contrast them with those obtained using an AUC-maximization algorithm. We also compare training times and performance results for the on-line and batch versions, demonstrating that our on-line algorithm scales to relatively large datasets with no significant loss in accuracy.

Original language | English (US) |
---|---|

Title of host publication | Experimental Algorithms - 6th International Workshop, WEA 2007, Proceedings |

Pages | 1-22 |

Number of pages | 22 |

Volume | 4525 LNCS |

State | Published - 2007 |

Event | 6th International Workshop on Experimental Algorithms, WEA 2007 - Rome, Italy Duration: Jun 6 2007 → Jun 8 2007 |

### Publication series

Name | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
---|---|

Volume | 4525 LNCS |

ISSN (Print) | 03029743 |

ISSN (Electronic) | 16113349 |

### Other

Other | 6th International Workshop on Experimental Algorithms, WEA 2007 |
---|---|

Country | Italy |

City | Rome |

Period | 6/6/07 → 6/8/07 |

### Fingerprint

### ASJC Scopus subject areas

- Computer Science(all)
- Biochemistry, Genetics and Molecular Biology(all)
- Theoretical Computer Science

### Cite this

*Experimental Algorithms - 6th International Workshop, WEA 2007, Proceedings*(Vol. 4525 LNCS, pp. 1-22). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 4525 LNCS).

**An alternative ranking problem for search engines.** / Cortes, Corinna; Mohri, Mehryar; Rastogi, Ashish.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

*Experimental Algorithms - 6th International Workshop, WEA 2007, Proceedings.*vol. 4525 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 4525 LNCS, pp. 1-22, 6th International Workshop on Experimental Algorithms, WEA 2007, Rome, Italy, 6/6/07.

}

TY - GEN

T1 - An alternative ranking problem for search engines

AU - Cortes, Corinna

AU - Mohri, Mehryar

AU - Rastogi, Ashish

PY - 2007

Y1 - 2007

N2 - This paper examines in detail an alternative ranking problem for search engines, movie recommendation, and other similar ranking systems motivated by the requirement to not just accurately predict pairwise ordering but also preserve the magnitude of the preferences or the difference between ratings. We describe and analyze several cost functions for this learning problem and give stability bounds for their generalization error, extending previously known stability results to non-bipartite ranking and magnitude of preference- preserving algorithms. We present algorithms optimizing these cost functions, and, in one instance, detail both a batch and an on-line version. For this algorithm, we also show how the leave-one-out error can be computed and approximated efficiently, which can be used to determine the optimal values of the trade-off parameter in the cost function. We report the results of experiments comparing these algorithms on several datasets and contrast them with those obtained using an AUC-maximization algorithm. We also compare training times and performance results for the on-line and batch versions, demonstrating that our on-line algorithm scales to relatively large datasets with no significant loss in accuracy.

AB - This paper examines in detail an alternative ranking problem for search engines, movie recommendation, and other similar ranking systems motivated by the requirement to not just accurately predict pairwise ordering but also preserve the magnitude of the preferences or the difference between ratings. We describe and analyze several cost functions for this learning problem and give stability bounds for their generalization error, extending previously known stability results to non-bipartite ranking and magnitude of preference- preserving algorithms. We present algorithms optimizing these cost functions, and, in one instance, detail both a batch and an on-line version. For this algorithm, we also show how the leave-one-out error can be computed and approximated efficiently, which can be used to determine the optimal values of the trade-off parameter in the cost function. We report the results of experiments comparing these algorithms on several datasets and contrast them with those obtained using an AUC-maximization algorithm. We also compare training times and performance results for the on-line and batch versions, demonstrating that our on-line algorithm scales to relatively large datasets with no significant loss in accuracy.

UR - http://www.scopus.com/inward/record.url?scp=37149011781&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=37149011781&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:37149011781

SN - 3540728449

SN - 9783540728443

VL - 4525 LNCS

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 1

EP - 22

BT - Experimental Algorithms - 6th International Workshop, WEA 2007, Proceedings

ER -