Twitter Sentiment Analysis with deep convolutional neural networks

Aliaksei Severyn, Alessandro Moschitti

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

176 Citations (Scopus)

Abstract

This paper describes our deep learning system for sentiment analysis of tweets. The main contribution of this work is a new model for initializing the parameter weights of the convolutional neural network, which is crucial to train an accurate model while avoiding the need to inject any additional features. Briefly, we use an unsupervised neural language model to train initial word embeddings that are further tuned by our deep learning model on a distant supervised corpus. At a final stage, the pre-trained parameters of the network are used to initialize the model. We train the latter on the supervised training data recently made available by the official system evaluation campaign on Twitter Sentiment Analysis organized by Semeval-2015. A comparison between the results of our approach and the systems participating in the challenge on the official test sets, suggests that our model could be ranked in the first two positions in both the phrase-level subtask A (among 11 teams) and on the message-level subtask B (among 40 teams). This is an important evidence on the practical value of our solution.

Original languageEnglish
Title of host publicationSIGIR 2015 - Proceedings of the 38th International ACM SIGIR Conference on Research and Development in Information Retrieval
PublisherAssociation for Computing Machinery, Inc
Pages959-962
Number of pages4
ISBN (Print)9781450336215
DOIs
Publication statusPublished - 9 Aug 2015
Event38th International ACM SIGIR Conference on Research and Development in Information Retrieval, SIGIR 2015 - Santiago, Chile
Duration: 9 Aug 201513 Aug 2015

Other

Other38th International ACM SIGIR Conference on Research and Development in Information Retrieval, SIGIR 2015
CountryChile
CitySantiago
Period9/8/1513/8/15

Fingerprint

Neural networks
Learning systems
Deep learning

Keywords

  • Convolutional neural networks
  • Twitter Sentiment Analysis

ASJC Scopus subject areas

  • Information Systems
  • Software

Cite this

Severyn, A., & Moschitti, A. (2015). Twitter Sentiment Analysis with deep convolutional neural networks. In SIGIR 2015 - Proceedings of the 38th International ACM SIGIR Conference on Research and Development in Information Retrieval (pp. 959-962). Association for Computing Machinery, Inc. https://doi.org/10.1145/2766462.2767830

Twitter Sentiment Analysis with deep convolutional neural networks. / Severyn, Aliaksei; Moschitti, Alessandro.

SIGIR 2015 - Proceedings of the 38th International ACM SIGIR Conference on Research and Development in Information Retrieval. Association for Computing Machinery, Inc, 2015. p. 959-962.

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

Severyn, A & Moschitti, A 2015, Twitter Sentiment Analysis with deep convolutional neural networks. in SIGIR 2015 - Proceedings of the 38th International ACM SIGIR Conference on Research and Development in Information Retrieval. Association for Computing Machinery, Inc, pp. 959-962, 38th International ACM SIGIR Conference on Research and Development in Information Retrieval, SIGIR 2015, Santiago, Chile, 9/8/15. https://doi.org/10.1145/2766462.2767830
Severyn A, Moschitti A. Twitter Sentiment Analysis with deep convolutional neural networks. In SIGIR 2015 - Proceedings of the 38th International ACM SIGIR Conference on Research and Development in Information Retrieval. Association for Computing Machinery, Inc. 2015. p. 959-962 https://doi.org/10.1145/2766462.2767830
Severyn, Aliaksei ; Moschitti, Alessandro. / Twitter Sentiment Analysis with deep convolutional neural networks. SIGIR 2015 - Proceedings of the 38th International ACM SIGIR Conference on Research and Development in Information Retrieval. Association for Computing Machinery, Inc, 2015. pp. 959-962
@inproceedings{0f6dfdfdab3b4684810f0b5ebcbe985d,
title = "Twitter Sentiment Analysis with deep convolutional neural networks",
abstract = "This paper describes our deep learning system for sentiment analysis of tweets. The main contribution of this work is a new model for initializing the parameter weights of the convolutional neural network, which is crucial to train an accurate model while avoiding the need to inject any additional features. Briefly, we use an unsupervised neural language model to train initial word embeddings that are further tuned by our deep learning model on a distant supervised corpus. At a final stage, the pre-trained parameters of the network are used to initialize the model. We train the latter on the supervised training data recently made available by the official system evaluation campaign on Twitter Sentiment Analysis organized by Semeval-2015. A comparison between the results of our approach and the systems participating in the challenge on the official test sets, suggests that our model could be ranked in the first two positions in both the phrase-level subtask A (among 11 teams) and on the message-level subtask B (among 40 teams). This is an important evidence on the practical value of our solution.",
keywords = "Convolutional neural networks, Twitter Sentiment Analysis",
author = "Aliaksei Severyn and Alessandro Moschitti",
year = "2015",
month = "8",
day = "9",
doi = "10.1145/2766462.2767830",
language = "English",
isbn = "9781450336215",
pages = "959--962",
booktitle = "SIGIR 2015 - Proceedings of the 38th International ACM SIGIR Conference on Research and Development in Information Retrieval",
publisher = "Association for Computing Machinery, Inc",

}

TY - GEN

T1 - Twitter Sentiment Analysis with deep convolutional neural networks

AU - Severyn, Aliaksei

AU - Moschitti, Alessandro

PY - 2015/8/9

Y1 - 2015/8/9

N2 - This paper describes our deep learning system for sentiment analysis of tweets. The main contribution of this work is a new model for initializing the parameter weights of the convolutional neural network, which is crucial to train an accurate model while avoiding the need to inject any additional features. Briefly, we use an unsupervised neural language model to train initial word embeddings that are further tuned by our deep learning model on a distant supervised corpus. At a final stage, the pre-trained parameters of the network are used to initialize the model. We train the latter on the supervised training data recently made available by the official system evaluation campaign on Twitter Sentiment Analysis organized by Semeval-2015. A comparison between the results of our approach and the systems participating in the challenge on the official test sets, suggests that our model could be ranked in the first two positions in both the phrase-level subtask A (among 11 teams) and on the message-level subtask B (among 40 teams). This is an important evidence on the practical value of our solution.

AB - This paper describes our deep learning system for sentiment analysis of tweets. The main contribution of this work is a new model for initializing the parameter weights of the convolutional neural network, which is crucial to train an accurate model while avoiding the need to inject any additional features. Briefly, we use an unsupervised neural language model to train initial word embeddings that are further tuned by our deep learning model on a distant supervised corpus. At a final stage, the pre-trained parameters of the network are used to initialize the model. We train the latter on the supervised training data recently made available by the official system evaluation campaign on Twitter Sentiment Analysis organized by Semeval-2015. A comparison between the results of our approach and the systems participating in the challenge on the official test sets, suggests that our model could be ranked in the first two positions in both the phrase-level subtask A (among 11 teams) and on the message-level subtask B (among 40 teams). This is an important evidence on the practical value of our solution.

KW - Convolutional neural networks

KW - Twitter Sentiment Analysis

UR - http://www.scopus.com/inward/record.url?scp=84953807567&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84953807567&partnerID=8YFLogxK

U2 - 10.1145/2766462.2767830

DO - 10.1145/2766462.2767830

M3 - Conference contribution

SN - 9781450336215

SP - 959

EP - 962

BT - SIGIR 2015 - Proceedings of the 38th International ACM SIGIR Conference on Research and Development in Information Retrieval

PB - Association for Computing Machinery, Inc

ER -