Preference learning for category-ranking based interactive text categorization

Fabio Aiolli, Fabrizio Sebastiani, Alessandro Sperduti

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

1 Citation (Scopus)

Abstract

Category Ranking is a variant of the multi-label classification problem, in which, rather than performing a (hard) assignment to an object of categories from a predefined set, we rank all categories according to their estimated "degree of suitability" to the object. Category ranking has many applications, all pertaining to "interactive" classification contexts in which the system, rather than taking a final categorization decision, is simply required to support a human expert who is in charge of taking this decision. Despite its high applicative potential in information retrieval applications, and in text categorization in particular, category ranking has mainly been tackled by standard text categorization methods. In this paper, we take a radically different stand to category ranking, i.e. one in which supervision is provided to the learner not in the standard form of labels attached to training documents, but in the form of preferences of type "category c1 is to be preferred to category c2 for document d". We apply to this problem a recently proposed, very general model for preferential learning, and show, through experiments performed on the standard Reuters-21578 benchmark, that this largely outperforms support vector machines, the learning method which has up to now proved the best-performing one in text categorization comparative experiments.

Original languageEnglish
Title of host publicationIEEE International Conference on Neural Networks - Conference Proceedings
Pages2034-2039
Number of pages6
DOIs
Publication statusPublished - 2007
Externally publishedYes
Event2007 International Joint Conference on Neural Networks, IJCNN 2007 - Orlando, FL
Duration: 12 Aug 200717 Aug 2007

Other

Other2007 International Joint Conference on Neural Networks, IJCNN 2007
CityOrlando, FL
Period12/8/0717/8/07

Fingerprint

Labels
Information retrieval
Support vector machines
Experiments

ASJC Scopus subject areas

  • Software

Cite this

Aiolli, F., Sebastiani, F., & Sperduti, A. (2007). Preference learning for category-ranking based interactive text categorization. In IEEE International Conference on Neural Networks - Conference Proceedings (pp. 2034-2039). [4371271] https://doi.org/10.1109/IJCNN.2007.4371271

Preference learning for category-ranking based interactive text categorization. / Aiolli, Fabio; Sebastiani, Fabrizio; Sperduti, Alessandro.

IEEE International Conference on Neural Networks - Conference Proceedings. 2007. p. 2034-2039 4371271.

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

Aiolli, F, Sebastiani, F & Sperduti, A 2007, Preference learning for category-ranking based interactive text categorization. in IEEE International Conference on Neural Networks - Conference Proceedings., 4371271, pp. 2034-2039, 2007 International Joint Conference on Neural Networks, IJCNN 2007, Orlando, FL, 12/8/07. https://doi.org/10.1109/IJCNN.2007.4371271
Aiolli F, Sebastiani F, Sperduti A. Preference learning for category-ranking based interactive text categorization. In IEEE International Conference on Neural Networks - Conference Proceedings. 2007. p. 2034-2039. 4371271 https://doi.org/10.1109/IJCNN.2007.4371271
Aiolli, Fabio ; Sebastiani, Fabrizio ; Sperduti, Alessandro. / Preference learning for category-ranking based interactive text categorization. IEEE International Conference on Neural Networks - Conference Proceedings. 2007. pp. 2034-2039
@inproceedings{c38ecc47bacb41008f6f83d99834efa9,
title = "Preference learning for category-ranking based interactive text categorization",
abstract = "Category Ranking is a variant of the multi-label classification problem, in which, rather than performing a (hard) assignment to an object of categories from a predefined set, we rank all categories according to their estimated {"}degree of suitability{"} to the object. Category ranking has many applications, all pertaining to {"}interactive{"} classification contexts in which the system, rather than taking a final categorization decision, is simply required to support a human expert who is in charge of taking this decision. Despite its high applicative potential in information retrieval applications, and in text categorization in particular, category ranking has mainly been tackled by standard text categorization methods. In this paper, we take a radically different stand to category ranking, i.e. one in which supervision is provided to the learner not in the standard form of labels attached to training documents, but in the form of preferences of type {"}category c1 is to be preferred to category c2 for document d{"}. We apply to this problem a recently proposed, very general model for preferential learning, and show, through experiments performed on the standard Reuters-21578 benchmark, that this largely outperforms support vector machines, the learning method which has up to now proved the best-performing one in text categorization comparative experiments.",
author = "Fabio Aiolli and Fabrizio Sebastiani and Alessandro Sperduti",
year = "2007",
doi = "10.1109/IJCNN.2007.4371271",
language = "English",
isbn = "142441380X",
pages = "2034--2039",
booktitle = "IEEE International Conference on Neural Networks - Conference Proceedings",

}

TY - GEN

T1 - Preference learning for category-ranking based interactive text categorization

AU - Aiolli, Fabio

AU - Sebastiani, Fabrizio

AU - Sperduti, Alessandro

PY - 2007

Y1 - 2007

N2 - Category Ranking is a variant of the multi-label classification problem, in which, rather than performing a (hard) assignment to an object of categories from a predefined set, we rank all categories according to their estimated "degree of suitability" to the object. Category ranking has many applications, all pertaining to "interactive" classification contexts in which the system, rather than taking a final categorization decision, is simply required to support a human expert who is in charge of taking this decision. Despite its high applicative potential in information retrieval applications, and in text categorization in particular, category ranking has mainly been tackled by standard text categorization methods. In this paper, we take a radically different stand to category ranking, i.e. one in which supervision is provided to the learner not in the standard form of labels attached to training documents, but in the form of preferences of type "category c1 is to be preferred to category c2 for document d". We apply to this problem a recently proposed, very general model for preferential learning, and show, through experiments performed on the standard Reuters-21578 benchmark, that this largely outperforms support vector machines, the learning method which has up to now proved the best-performing one in text categorization comparative experiments.

AB - Category Ranking is a variant of the multi-label classification problem, in which, rather than performing a (hard) assignment to an object of categories from a predefined set, we rank all categories according to their estimated "degree of suitability" to the object. Category ranking has many applications, all pertaining to "interactive" classification contexts in which the system, rather than taking a final categorization decision, is simply required to support a human expert who is in charge of taking this decision. Despite its high applicative potential in information retrieval applications, and in text categorization in particular, category ranking has mainly been tackled by standard text categorization methods. In this paper, we take a radically different stand to category ranking, i.e. one in which supervision is provided to the learner not in the standard form of labels attached to training documents, but in the form of preferences of type "category c1 is to be preferred to category c2 for document d". We apply to this problem a recently proposed, very general model for preferential learning, and show, through experiments performed on the standard Reuters-21578 benchmark, that this largely outperforms support vector machines, the learning method which has up to now proved the best-performing one in text categorization comparative experiments.

UR - http://www.scopus.com/inward/record.url?scp=51749109788&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=51749109788&partnerID=8YFLogxK

U2 - 10.1109/IJCNN.2007.4371271

DO - 10.1109/IJCNN.2007.4371271

M3 - Conference contribution

SN - 142441380X

SN - 9781424413805

SP - 2034

EP - 2039

BT - IEEE International Conference on Neural Networks - Conference Proceedings

ER -