Hierarchical multi-label conditional random fields for aspect-oriented opinion mining

Diego Marcheggiani, Oscar Täckström, Andrea Esuli, Fabrizio Sebastiani

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

26 Citations (Scopus)

Abstract

A common feature of many online review sites is the use of an overall rating that summarizes the opinions expressed in a review. Unfortunately, these document-level ratings do not provide any information about the opinions contained in the review that concern a specific aspect (e.g., cleanliness) of the product being reviewed (e.g., a hotel). In this paper we study the finer-grained problem of aspect-oriented opinion mining at the sentence level, which consists of predicting, for all sentences in the review, whether the sentence expresses a positive, neutral, or negative opinion (or no opinion at all) about a specific aspect of the product. For this task we propose a set of increasingly powerful models based on conditional random fields (CRFs), including a hierarchical multi-label CRFs scheme that jointly models the overall opinion expressed in the review and the set of aspect-specific opinions expressed in each of its sentences. We evaluate the proposed models against a dataset of hotel reviews (which we here make publicly available) in which the set of aspects and the opinions expressed concerning them are manually annotated at the sentence level. We find that both hierarchical and multi-label factors lead to improved predictions of aspect-oriented opinions.

Original languageEnglish
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
PublisherSpringer Verlag
Pages273-285
Number of pages13
Volume8416 LNCS
ISBN (Print)9783319060279
DOIs
Publication statusPublished - 2014
Externally publishedYes
Event36th European Conference on Information Retrieval, ECIR 2014 - Amsterdam
Duration: 13 Apr 201416 Apr 2014

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume8416 LNCS
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other36th European Conference on Information Retrieval, ECIR 2014
CityAmsterdam
Period13/4/1416/4/14

Fingerprint

Opinion Mining
Conditional Random Fields
Labels
Hotels
Review
Express
Model-based
Evaluate
Prediction
Model

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science

Cite this

Marcheggiani, D., Täckström, O., Esuli, A., & Sebastiani, F. (2014). Hierarchical multi-label conditional random fields for aspect-oriented opinion mining. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 8416 LNCS, pp. 273-285). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8416 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-319-06028-6_23

Hierarchical multi-label conditional random fields for aspect-oriented opinion mining. / Marcheggiani, Diego; Täckström, Oscar; Esuli, Andrea; Sebastiani, Fabrizio.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 8416 LNCS Springer Verlag, 2014. p. 273-285 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8416 LNCS).

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

Marcheggiani, D, Täckström, O, Esuli, A & Sebastiani, F 2014, Hierarchical multi-label conditional random fields for aspect-oriented opinion mining. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). vol. 8416 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 8416 LNCS, Springer Verlag, pp. 273-285, 36th European Conference on Information Retrieval, ECIR 2014, Amsterdam, 13/4/14. https://doi.org/10.1007/978-3-319-06028-6_23
Marcheggiani D, Täckström O, Esuli A, Sebastiani F. Hierarchical multi-label conditional random fields for aspect-oriented opinion mining. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 8416 LNCS. Springer Verlag. 2014. p. 273-285. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-319-06028-6_23
Marcheggiani, Diego ; Täckström, Oscar ; Esuli, Andrea ; Sebastiani, Fabrizio. / Hierarchical multi-label conditional random fields for aspect-oriented opinion mining. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 8416 LNCS Springer Verlag, 2014. pp. 273-285 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{36f5e1ccfeb54677add41d641de5d4a8,
title = "Hierarchical multi-label conditional random fields for aspect-oriented opinion mining",
abstract = "A common feature of many online review sites is the use of an overall rating that summarizes the opinions expressed in a review. Unfortunately, these document-level ratings do not provide any information about the opinions contained in the review that concern a specific aspect (e.g., cleanliness) of the product being reviewed (e.g., a hotel). In this paper we study the finer-grained problem of aspect-oriented opinion mining at the sentence level, which consists of predicting, for all sentences in the review, whether the sentence expresses a positive, neutral, or negative opinion (or no opinion at all) about a specific aspect of the product. For this task we propose a set of increasingly powerful models based on conditional random fields (CRFs), including a hierarchical multi-label CRFs scheme that jointly models the overall opinion expressed in the review and the set of aspect-specific opinions expressed in each of its sentences. We evaluate the proposed models against a dataset of hotel reviews (which we here make publicly available) in which the set of aspects and the opinions expressed concerning them are manually annotated at the sentence level. We find that both hierarchical and multi-label factors lead to improved predictions of aspect-oriented opinions.",
author = "Diego Marcheggiani and Oscar T{\"a}ckstr{\"o}m and Andrea Esuli and Fabrizio Sebastiani",
year = "2014",
doi = "10.1007/978-3-319-06028-6_23",
language = "English",
isbn = "9783319060279",
volume = "8416 LNCS",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "273--285",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

}

TY - GEN

T1 - Hierarchical multi-label conditional random fields for aspect-oriented opinion mining

AU - Marcheggiani, Diego

AU - Täckström, Oscar

AU - Esuli, Andrea

AU - Sebastiani, Fabrizio

PY - 2014

Y1 - 2014

N2 - A common feature of many online review sites is the use of an overall rating that summarizes the opinions expressed in a review. Unfortunately, these document-level ratings do not provide any information about the opinions contained in the review that concern a specific aspect (e.g., cleanliness) of the product being reviewed (e.g., a hotel). In this paper we study the finer-grained problem of aspect-oriented opinion mining at the sentence level, which consists of predicting, for all sentences in the review, whether the sentence expresses a positive, neutral, or negative opinion (or no opinion at all) about a specific aspect of the product. For this task we propose a set of increasingly powerful models based on conditional random fields (CRFs), including a hierarchical multi-label CRFs scheme that jointly models the overall opinion expressed in the review and the set of aspect-specific opinions expressed in each of its sentences. We evaluate the proposed models against a dataset of hotel reviews (which we here make publicly available) in which the set of aspects and the opinions expressed concerning them are manually annotated at the sentence level. We find that both hierarchical and multi-label factors lead to improved predictions of aspect-oriented opinions.

AB - A common feature of many online review sites is the use of an overall rating that summarizes the opinions expressed in a review. Unfortunately, these document-level ratings do not provide any information about the opinions contained in the review that concern a specific aspect (e.g., cleanliness) of the product being reviewed (e.g., a hotel). In this paper we study the finer-grained problem of aspect-oriented opinion mining at the sentence level, which consists of predicting, for all sentences in the review, whether the sentence expresses a positive, neutral, or negative opinion (or no opinion at all) about a specific aspect of the product. For this task we propose a set of increasingly powerful models based on conditional random fields (CRFs), including a hierarchical multi-label CRFs scheme that jointly models the overall opinion expressed in the review and the set of aspect-specific opinions expressed in each of its sentences. We evaluate the proposed models against a dataset of hotel reviews (which we here make publicly available) in which the set of aspects and the opinions expressed concerning them are manually annotated at the sentence level. We find that both hierarchical and multi-label factors lead to improved predictions of aspect-oriented opinions.

UR - http://www.scopus.com/inward/record.url?scp=84899958907&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84899958907&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-06028-6_23

DO - 10.1007/978-3-319-06028-6_23

M3 - Conference contribution

AN - SCOPUS:84899958907

SN - 9783319060279

VL - 8416 LNCS

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

SP - 273

EP - 285

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

PB - Springer Verlag

ER -