Incremental reranking for hierarchical text classification

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

Abstract

The top-down method is efficient and commonly used in hierarchical text classification. Its main drawback is the error propagation from the higher to the lower nodes. To address this issue we propose an efficient incremental reranking model of the top-down classifier decisions. We build a multiclassifier for each hierarchy node, constituted by the latter and its children. Then we generate several classification hypotheses with such classifiers and rerank them to select the best one. Our rerankers exploit category dependencies, which allow them to recover from the multiclassifier errors whereas their application in top-down fashion results in high efficiency. The experimentation on Reuters Corpus Volume 1 (RCV1) shows that our incremental reranking is as accurate as global rerankers but at least one magnitude order faster.

Original languageEnglish
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Pages726-729
Number of pages4
Volume7814 LNCS
DOIs
Publication statusPublished - 2 Apr 2013
Externally publishedYes
Event35th European Conference on Information Retrieval, ECIR 2013 - Moscow, Russian Federation
Duration: 24 Mar 201327 Mar 2013

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume7814 LNCS
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other35th European Conference on Information Retrieval, ECIR 2013
CountryRussian Federation
CityMoscow
Period24/3/1327/3/13

Fingerprint

Hierarchical Classification
Text Classification
Classifiers
Classifier
Error Propagation
Vertex of a graph
Experimentation
High Efficiency
Model
Hierarchy
Children
Corpus

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science

Cite this

Ju, Q., & Moschitti, A. (2013). Incremental reranking for hierarchical text classification. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 7814 LNCS, pp. 726-729). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7814 LNCS). https://doi.org/10.1007/978-3-642-36973-5_70

Incremental reranking for hierarchical text classification. / Ju, Qi; Moschitti, Alessandro.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 7814 LNCS 2013. p. 726-729 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 7814 LNCS).

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

Ju, Q & Moschitti, A 2013, Incremental reranking for hierarchical text classification. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). vol. 7814 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 7814 LNCS, pp. 726-729, 35th European Conference on Information Retrieval, ECIR 2013, Moscow, Russian Federation, 24/3/13. https://doi.org/10.1007/978-3-642-36973-5_70
Ju Q, Moschitti A. Incremental reranking for hierarchical text classification. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 7814 LNCS. 2013. p. 726-729. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-642-36973-5_70
Ju, Qi ; Moschitti, Alessandro. / Incremental reranking for hierarchical text classification. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 7814 LNCS 2013. pp. 726-729 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{727cded402ef45868d381f38501baadb,
title = "Incremental reranking for hierarchical text classification",
abstract = "The top-down method is efficient and commonly used in hierarchical text classification. Its main drawback is the error propagation from the higher to the lower nodes. To address this issue we propose an efficient incremental reranking model of the top-down classifier decisions. We build a multiclassifier for each hierarchy node, constituted by the latter and its children. Then we generate several classification hypotheses with such classifiers and rerank them to select the best one. Our rerankers exploit category dependencies, which allow them to recover from the multiclassifier errors whereas their application in top-down fashion results in high efficiency. The experimentation on Reuters Corpus Volume 1 (RCV1) shows that our incremental reranking is as accurate as global rerankers but at least one magnitude order faster.",
author = "Qi Ju and Alessandro Moschitti",
year = "2013",
month = "4",
day = "2",
doi = "10.1007/978-3-642-36973-5_70",
language = "English",
isbn = "9783642369728",
volume = "7814 LNCS",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
pages = "726--729",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

}

TY - GEN

T1 - Incremental reranking for hierarchical text classification

AU - Ju, Qi

AU - Moschitti, Alessandro

PY - 2013/4/2

Y1 - 2013/4/2

N2 - The top-down method is efficient and commonly used in hierarchical text classification. Its main drawback is the error propagation from the higher to the lower nodes. To address this issue we propose an efficient incremental reranking model of the top-down classifier decisions. We build a multiclassifier for each hierarchy node, constituted by the latter and its children. Then we generate several classification hypotheses with such classifiers and rerank them to select the best one. Our rerankers exploit category dependencies, which allow them to recover from the multiclassifier errors whereas their application in top-down fashion results in high efficiency. The experimentation on Reuters Corpus Volume 1 (RCV1) shows that our incremental reranking is as accurate as global rerankers but at least one magnitude order faster.

AB - The top-down method is efficient and commonly used in hierarchical text classification. Its main drawback is the error propagation from the higher to the lower nodes. To address this issue we propose an efficient incremental reranking model of the top-down classifier decisions. We build a multiclassifier for each hierarchy node, constituted by the latter and its children. Then we generate several classification hypotheses with such classifiers and rerank them to select the best one. Our rerankers exploit category dependencies, which allow them to recover from the multiclassifier errors whereas their application in top-down fashion results in high efficiency. The experimentation on Reuters Corpus Volume 1 (RCV1) shows that our incremental reranking is as accurate as global rerankers but at least one magnitude order faster.

UR - http://www.scopus.com/inward/record.url?scp=84875465007&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84875465007&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-36973-5_70

DO - 10.1007/978-3-642-36973-5_70

M3 - Conference contribution

AN - SCOPUS:84875465007

SN - 9783642369728

VL - 7814 LNCS

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

SP - 726

EP - 729

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

ER -