Standard

Ad lingua: Text classification improves symbolism prediction in image advertisements. / Savchenko, Andrey; Alekseev, Anton; Kwon, Sejeong; Tutubalina, Elena; Myasnikov, Evgeny; Nikolenko, Sergey.

Proceedings of the 28th International Conference on Computational Linguistics. 2020. p. 1886-1892.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Harvard

Savchenko, A, Alekseev, A, Kwon, S, Tutubalina, E, Myasnikov, E & Nikolenko, S 2020, Ad lingua: Text classification improves symbolism prediction in image advertisements. in Proceedings of the 28th International Conference on Computational Linguistics. pp. 1886-1892, 28th International Conference on Computational Linguistics, Barcelona, Spain, 8/12/20. https://doi.org/10.18653/v1/2020.coling-main.171

APA

Savchenko, A., Alekseev, A., Kwon, S., Tutubalina, E., Myasnikov, E., & Nikolenko, S. (2020). Ad lingua: Text classification improves symbolism prediction in image advertisements. In Proceedings of the 28th International Conference on Computational Linguistics (pp. 1886-1892) https://doi.org/10.18653/v1/2020.coling-main.171

Vancouver

Savchenko A, Alekseev A, Kwon S, Tutubalina E, Myasnikov E, Nikolenko S. Ad lingua: Text classification improves symbolism prediction in image advertisements. In Proceedings of the 28th International Conference on Computational Linguistics. 2020. p. 1886-1892 https://doi.org/10.18653/v1/2020.coling-main.171

Author

Savchenko, Andrey ; Alekseev, Anton ; Kwon, Sejeong ; Tutubalina, Elena ; Myasnikov, Evgeny ; Nikolenko, Sergey. / Ad lingua: Text classification improves symbolism prediction in image advertisements. Proceedings of the 28th International Conference on Computational Linguistics. 2020. pp. 1886-1892

BibTeX

@inproceedings{0a33ca3d351a4ad8a19258901864a8ae,
title = "Ad lingua: Text classification improves symbolism prediction in image advertisements",
abstract = "Understanding image advertisements is a challenging task, often requiring non-literal interpretation. We argue that standard image-based predictions are insufficient for symbolism prediction. Following the intuition that texts and images are complementary in advertising, we introduce a multimodal ensemble of a state of the art image-based classifier, a classifier based on an object detection architecture, and a fine-tuned language model applied to texts extracted from ads by OCR. The resulting system establishes a new state of the art in symbolism prediction.",
author = "Andrey Savchenko and Anton Alekseev and Sejeong Kwon and Elena Tutubalina and Evgeny Myasnikov and Sergey Nikolenko",
note = "Andrey Savchenko, Anton Alekseev, Sejeong Kwon, Elena Tutubalina, Evgeny Myasnikov, and Sergey Nikolenko. 2020. Ad Lingua: Text Classification Improves Symbolism Prediction in Image Advertisements. In Proceedings of the 28th International Conference on Computational Linguistics, pages 1886–1892, Barcelona, Spain (Online). International Committee on Computational Linguistics.; 28th International Conference on Computational Linguistics, COLING 2020 ; Conference date: 08-12-2020 Through 13-12-2020",
year = "2020",
doi = "10.18653/v1/2020.coling-main.171",
language = "English",
pages = "1886--1892",
booktitle = "Proceedings of the 28th International Conference on Computational Linguistics",

}

RIS

TY - GEN

T1 - Ad lingua: Text classification improves symbolism prediction in image advertisements

AU - Savchenko, Andrey

AU - Alekseev, Anton

AU - Kwon, Sejeong

AU - Tutubalina, Elena

AU - Myasnikov, Evgeny

AU - Nikolenko, Sergey

N1 - Andrey Savchenko, Anton Alekseev, Sejeong Kwon, Elena Tutubalina, Evgeny Myasnikov, and Sergey Nikolenko. 2020. Ad Lingua: Text Classification Improves Symbolism Prediction in Image Advertisements. In Proceedings of the 28th International Conference on Computational Linguistics, pages 1886–1892, Barcelona, Spain (Online). International Committee on Computational Linguistics.

PY - 2020

Y1 - 2020

N2 - Understanding image advertisements is a challenging task, often requiring non-literal interpretation. We argue that standard image-based predictions are insufficient for symbolism prediction. Following the intuition that texts and images are complementary in advertising, we introduce a multimodal ensemble of a state of the art image-based classifier, a classifier based on an object detection architecture, and a fine-tuned language model applied to texts extracted from ads by OCR. The resulting system establishes a new state of the art in symbolism prediction.

AB - Understanding image advertisements is a challenging task, often requiring non-literal interpretation. We argue that standard image-based predictions are insufficient for symbolism prediction. Following the intuition that texts and images are complementary in advertising, we introduce a multimodal ensemble of a state of the art image-based classifier, a classifier based on an object detection architecture, and a fine-tuned language model applied to texts extracted from ads by OCR. The resulting system establishes a new state of the art in symbolism prediction.

U2 - 10.18653/v1/2020.coling-main.171

DO - 10.18653/v1/2020.coling-main.171

M3 - Conference contribution

SP - 1886

EP - 1892

BT - Proceedings of the 28th International Conference on Computational Linguistics

T2 - 28th International Conference on Computational Linguistics

Y2 - 8 December 2020 through 13 December 2020

ER -

ID: 95169147