Pairwise versus pointwise ranking: A case study

Vitalik Melnikov; Pritha Gupta; Bernd Frick; Daniel Kaimann; Eyke Hüllermeier

doi:10.4467/20838476SI.16.006.6187

Pairwise versus pointwise ranking: A case study

Vitalik Melnikov
, Pritha Gupta
, Bernd Frick
, Daniel Kaimann
, Eyke Hüllermeier

Department of Arts and Culture Studies

Paderborn University

Research output: Contribution to journal › Article › Academic › peer-review

12 Citations (Scopus)

Abstract

Object ranking is one of the most relevant problems in the realm of preference learning and ranking. It is mostly tackled by means of two different techniques, often referred to as pairwise and pointwise ranking. In this paper, we present a case study in which we systematically compare two representatives of these techniques, a method based on the reduction of ranking to binary classification and so-called expected rank regression (ERR). Our experiments are meant to complement existing studies in this field, especially previous evaluations of ERR. And indeed, our results are not fully in agreement with previous findings and partly support different conclusions.

Original language	English
Pages (from-to)	73-83
Number of pages	11
Journal	Schedae Informaticae
Volume	25
DOIs	https://doi.org/10.4467/20838476SI.16.006.6187
Publication status	Published - 2016

Bibliographical note

Funding Information:
This work has been supported by the German Research Foundation (Deutsche Forsch-ungsgesellschaft,DFG) within the Collaborative Research Centre “On-The-Fly Computing” (CRC 901).

Research programs

ESHCC A&CS

Access to Document

10.4467/20838476SI.16.006.6187

Cite this

@article{d6df2df18c8f422a8d3107e958d2edbc,

title = "Pairwise versus pointwise ranking: A case study",

abstract = "Object ranking is one of the most relevant problems in the realm of preference learning and ranking. It is mostly tackled by means of two different techniques, often referred to as pairwise and pointwise ranking. In this paper, we present a case study in which we systematically compare two representatives of these techniques, a method based on the reduction of ranking to binary classification and so-called expected rank regression (ERR). Our experiments are meant to complement existing studies in this field, especially previous evaluations of ERR. And indeed, our results are not fully in agreement with previous findings and partly support different conclusions.",

author = "Vitalik Melnikov and Pritha Gupta and Bernd Frick and Daniel Kaimann and Eyke H{\"u}llermeier",

note = "Funding Information: This work has been supported by the German Research Foundation (Deutsche Forsch-ungsgesellschaft,DFG) within the Collaborative Research Centre {\textquotedblleft}On-The-Fly Computing{\textquotedblright} (CRC 901).",

year = "2016",

doi = "10.4467/20838476SI.16.006.6187",

language = "English",

volume = "25",

pages = "73--83",

journal = "Schedae Informaticae",

issn = "1732-3916",

publisher = "Jagiellonian University",

}

TY - JOUR

T1 - Pairwise versus pointwise ranking

T2 - A case study

AU - Melnikov, Vitalik

AU - Gupta, Pritha

AU - Frick, Bernd

AU - Kaimann, Daniel

AU - Hüllermeier, Eyke

N1 - Funding Information: This work has been supported by the German Research Foundation (Deutsche Forsch-ungsgesellschaft,DFG) within the Collaborative Research Centre “On-The-Fly Computing” (CRC 901).

PY - 2016

Y1 - 2016

N2 - Object ranking is one of the most relevant problems in the realm of preference learning and ranking. It is mostly tackled by means of two different techniques, often referred to as pairwise and pointwise ranking. In this paper, we present a case study in which we systematically compare two representatives of these techniques, a method based on the reduction of ranking to binary classification and so-called expected rank regression (ERR). Our experiments are meant to complement existing studies in this field, especially previous evaluations of ERR. And indeed, our results are not fully in agreement with previous findings and partly support different conclusions.

AB - Object ranking is one of the most relevant problems in the realm of preference learning and ranking. It is mostly tackled by means of two different techniques, often referred to as pairwise and pointwise ranking. In this paper, we present a case study in which we systematically compare two representatives of these techniques, a method based on the reduction of ranking to binary classification and so-called expected rank regression (ERR). Our experiments are meant to complement existing studies in this field, especially previous evaluations of ERR. And indeed, our results are not fully in agreement with previous findings and partly support different conclusions.

UR - https://www.scopus.com/pages/publications/85027835054

U2 - 10.4467/20838476SI.16.006.6187

DO - 10.4467/20838476SI.16.006.6187

M3 - Article

AN - SCOPUS:85027835054

SN - 1732-3916

VL - 25

SP - 73

EP - 83

JO - Schedae Informaticae

JF - Schedae Informaticae

ER -

Pairwise versus pointwise ranking: A case study

Abstract

Bibliographical note

Research programs

Access to Document

Other files and links

Fingerprint

Cite this