Rating prediction by correcting user rating bias

Masanao Ochi, Yutaka Matsuo, Makoto Okabe, Rikio Onai

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

1 Citation (Scopus)

Abstract

We propose a novel method to improve the prediction accuracy on the rating prediction task by correcting the bias of user ratings. We demonstrate that the manner of user rating and review is biased and that it is necessary to correct this difference for more accurate prediction. Our proposed method comprises approaches based on the detection of each user value to ratings: The bias of the rating is detected using entropy of user rating and by updating word weights only when the words appear in the review, the problem of bias is reduced. We implement this idea by extending the Prank algorithm. We apply a review - item matrix as a feature matrix instead of a user - item matrix because of its volume of information. Our quantitative evaluation shows that our method improves the prediction accuracy (the Rank Loss measurement) significantly by 8.70 % compared with the normal Prank algorithm. Our proposed method helps users find out what they care about when buying something, and is applicable to newer variants of the Prank algorithm. Moreover, it is useful to most review sites because we use only rating and review data.

Original languageEnglish
Title of host publicationProceedings - 2012 IEEE/WIC/ACM International Conference on Web Intelligence, WI 2012
Pages452-456
Number of pages5
DOIs
Publication statusPublished - 2012
Externally publishedYes
Event2012 IEEE/WIC/ACM International Conference on Web Intelligence, WI 2012 - Macau, China
Duration: 2012 Dec 42012 Dec 7

Other

Other2012 IEEE/WIC/ACM International Conference on Web Intelligence, WI 2012
CountryChina
CityMacau
Period12/12/412/12/7

Fingerprint

Entropy

Keywords

  • Collaborative Filtering
  • Prank
  • Rating Prediction
  • Recommendation

ASJC Scopus subject areas

  • Artificial Intelligence
  • Software

Cite this

Ochi, M., Matsuo, Y., Okabe, M., & Onai, R. (2012). Rating prediction by correcting user rating bias. In Proceedings - 2012 IEEE/WIC/ACM International Conference on Web Intelligence, WI 2012 (pp. 452-456). [6511923] https://doi.org/10.1109/WI-IAT.2012.186

Rating prediction by correcting user rating bias. / Ochi, Masanao; Matsuo, Yutaka; Okabe, Makoto; Onai, Rikio.

Proceedings - 2012 IEEE/WIC/ACM International Conference on Web Intelligence, WI 2012. 2012. p. 452-456 6511923.

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

Ochi, M, Matsuo, Y, Okabe, M & Onai, R 2012, Rating prediction by correcting user rating bias. in Proceedings - 2012 IEEE/WIC/ACM International Conference on Web Intelligence, WI 2012., 6511923, pp. 452-456, 2012 IEEE/WIC/ACM International Conference on Web Intelligence, WI 2012, Macau, China, 12/12/4. https://doi.org/10.1109/WI-IAT.2012.186
Ochi M, Matsuo Y, Okabe M, Onai R. Rating prediction by correcting user rating bias. In Proceedings - 2012 IEEE/WIC/ACM International Conference on Web Intelligence, WI 2012. 2012. p. 452-456. 6511923 https://doi.org/10.1109/WI-IAT.2012.186
Ochi, Masanao ; Matsuo, Yutaka ; Okabe, Makoto ; Onai, Rikio. / Rating prediction by correcting user rating bias. Proceedings - 2012 IEEE/WIC/ACM International Conference on Web Intelligence, WI 2012. 2012. pp. 452-456
@inproceedings{77e7e024fdd447a5a53690749c612946,
title = "Rating prediction by correcting user rating bias",
abstract = "We propose a novel method to improve the prediction accuracy on the rating prediction task by correcting the bias of user ratings. We demonstrate that the manner of user rating and review is biased and that it is necessary to correct this difference for more accurate prediction. Our proposed method comprises approaches based on the detection of each user value to ratings: The bias of the rating is detected using entropy of user rating and by updating word weights only when the words appear in the review, the problem of bias is reduced. We implement this idea by extending the Prank algorithm. We apply a review - item matrix as a feature matrix instead of a user - item matrix because of its volume of information. Our quantitative evaluation shows that our method improves the prediction accuracy (the Rank Loss measurement) significantly by 8.70 {\%} compared with the normal Prank algorithm. Our proposed method helps users find out what they care about when buying something, and is applicable to newer variants of the Prank algorithm. Moreover, it is useful to most review sites because we use only rating and review data.",
keywords = "Collaborative Filtering, Prank, Rating Prediction, Recommendation",
author = "Masanao Ochi and Yutaka Matsuo and Makoto Okabe and Rikio Onai",
year = "2012",
doi = "10.1109/WI-IAT.2012.186",
language = "English",
isbn = "9780769548807",
pages = "452--456",
booktitle = "Proceedings - 2012 IEEE/WIC/ACM International Conference on Web Intelligence, WI 2012",

}

TY - GEN

T1 - Rating prediction by correcting user rating bias

AU - Ochi, Masanao

AU - Matsuo, Yutaka

AU - Okabe, Makoto

AU - Onai, Rikio

PY - 2012

Y1 - 2012

N2 - We propose a novel method to improve the prediction accuracy on the rating prediction task by correcting the bias of user ratings. We demonstrate that the manner of user rating and review is biased and that it is necessary to correct this difference for more accurate prediction. Our proposed method comprises approaches based on the detection of each user value to ratings: The bias of the rating is detected using entropy of user rating and by updating word weights only when the words appear in the review, the problem of bias is reduced. We implement this idea by extending the Prank algorithm. We apply a review - item matrix as a feature matrix instead of a user - item matrix because of its volume of information. Our quantitative evaluation shows that our method improves the prediction accuracy (the Rank Loss measurement) significantly by 8.70 % compared with the normal Prank algorithm. Our proposed method helps users find out what they care about when buying something, and is applicable to newer variants of the Prank algorithm. Moreover, it is useful to most review sites because we use only rating and review data.

AB - We propose a novel method to improve the prediction accuracy on the rating prediction task by correcting the bias of user ratings. We demonstrate that the manner of user rating and review is biased and that it is necessary to correct this difference for more accurate prediction. Our proposed method comprises approaches based on the detection of each user value to ratings: The bias of the rating is detected using entropy of user rating and by updating word weights only when the words appear in the review, the problem of bias is reduced. We implement this idea by extending the Prank algorithm. We apply a review - item matrix as a feature matrix instead of a user - item matrix because of its volume of information. Our quantitative evaluation shows that our method improves the prediction accuracy (the Rank Loss measurement) significantly by 8.70 % compared with the normal Prank algorithm. Our proposed method helps users find out what they care about when buying something, and is applicable to newer variants of the Prank algorithm. Moreover, it is useful to most review sites because we use only rating and review data.

KW - Collaborative Filtering

KW - Prank

KW - Rating Prediction

KW - Recommendation

UR - http://www.scopus.com/inward/record.url?scp=84878455429&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84878455429&partnerID=8YFLogxK

U2 - 10.1109/WI-IAT.2012.186

DO - 10.1109/WI-IAT.2012.186

M3 - Conference contribution

AN - SCOPUS:84878455429

SN - 9780769548807

SP - 452

EP - 456

BT - Proceedings - 2012 IEEE/WIC/ACM International Conference on Web Intelligence, WI 2012

ER -