Give it a shot: Few-shot learning to normalize ADR mentions in Social Media posts

E. Manousogiannis; Sepideh Mesbah; Selene Baez Santamaria; Alessandro Bozzon; Robert-Jan Sips

Give it a shot: Few-shot learning to normalize ADR mentions in Social Media posts

E. Manousogiannis, Sepideh Mesbah, Selene Baez Santamaria, Alessandro Bozzon, Robert-Jan Sips

Research output: Chapter in Book/Conference proceedings/Edited volume › Conference contribution › Scientific › peer-review

52 Downloads (Pure)

Abstract

This paper describes the system that team MYTOMORROWS-TU DELFT developed for the 2019 Social Media Mining for Health Applications (SMM4H) Shared Task 3, for the end-to-end normalization of ADR tweet mentions to their corresponding MEDDRA codes. For the first two steps, we reuse a state-of-theart approach, focusing our contribution on the final entity-linking step. For that we propose a simple Few-Shot learning approach, based on pre-trained word embeddings and data from the UMLS, combined with the provided training data. Our system (relaxed F1: 0.337- 0.345) outperforms the average (relaxed F1 0.2972) of the participants in this task, demonstrating the potential feasibility of few-shot learning in the context of medical text normalization.

Original language	English
Title of host publication	Proceedings of the 4th Social Media Mining for Health Applications (#SMM4H) Workshop & Shared Task
Editors	Davy Weissenbacher, Graciela Gonzalez-Hernandez
Publisher	Association for Computational Linguistics
Pages	114–116
Number of pages	4
ISBN (Electronic)	978-1-950737-46-8
Publication status	Published - 2019
Event	4th Social Media Mining for Health Applications (#SMM4H) Workshop & Shared Task - Florence, Italy Duration: 2 Aug 2019 → 2 Aug 2019 Conference number: 4 https://www.aclweb.org/anthology/W19-32

Conference

Conference	4th Social Media Mining for Health Applications (#SMM4H) Workshop & Shared Task
Abbreviated title	#SMM4H
Country/Territory	Italy
City	Florence
Period	2/08/19 → 2/08/19
Internet address	https://www.aclweb.org/anthology/W19-32

Access to Document

W19_3219Accepted author manuscript, 149 KB

Cite this

Manousogiannis, E., Mesbah, S., Baez Santamaria, S., Bozzon, A., & Sips, R.-J. (2019). Give it a shot: Few-shot learning to normalize ADR mentions in Social Media posts. In D. Weissenbacher, & G. Gonzalez-Hernandez (Eds.), Proceedings of the 4th Social Media Mining for Health Applications (#SMM4H) Workshop & Shared Task (pp. 114–116). Association for Computational Linguistics.

Manousogiannis, E. ; Mesbah, Sepideh ; Baez Santamaria, Selene et al. / Give it a shot: Few-shot learning to normalize ADR mentions in Social Media posts. Proceedings of the 4th Social Media Mining for Health Applications (#SMM4H) Workshop & Shared Task. editor / Davy Weissenbacher ; Graciela Gonzalez-Hernandez. Association for Computational Linguistics, 2019. pp. 114–116

@inproceedings{641abf0d3e2841f1ae73d7c6a01b7810,

title = "Give it a shot: Few-shot learning to normalize ADR mentions in Social Media posts",

abstract = "This paper describes the system that team MYTOMORROWS-TU DELFT developed for the 2019 Social Media Mining for Health Applications (SMM4H) Shared Task 3, for the end-to-end normalization of ADR tweet mentions to their corresponding MEDDRA codes. For the first two steps, we reuse a state-of-theart approach, focusing our contribution on the final entity-linking step. For that we propose a simple Few-Shot learning approach, based on pre-trained word embeddings and data from the UMLS, combined with the provided training data. Our system (relaxed F1: 0.337- 0.345) outperforms the average (relaxed F1 0.2972) of the participants in this task, demonstrating the potential feasibility of few-shot learning in the context of medical text normalization. ",

author = "E. Manousogiannis and Sepideh Mesbah and {Baez Santamaria}, Selene and Alessandro Bozzon and Robert-Jan Sips",

year = "2019",

language = "English",

pages = "114–116",

editor = "Davy Weissenbacher and Graciela Gonzalez-Hernandez",

booktitle = "Proceedings of the 4th Social Media Mining for Health Applications (#SMM4H) Workshop & Shared Task",

publisher = "Association for Computational Linguistics",

note = "4th Social Media Mining for Health Applications (#SMM4H) Workshop & Shared Task, #SMM4H ; Conference date: 02-08-2019 Through 02-08-2019",

url = "https://www.aclweb.org/anthology/W19-32",

}

Manousogiannis, E, Mesbah, S, Baez Santamaria, S, Bozzon, A & Sips, R-J 2019, Give it a shot: Few-shot learning to normalize ADR mentions in Social Media posts. in D Weissenbacher & G Gonzalez-Hernandez (eds), Proceedings of the 4th Social Media Mining for Health Applications (#SMM4H) Workshop & Shared Task. Association for Computational Linguistics, pp. 114–116, 4th Social Media Mining for Health Applications (#SMM4H) Workshop & Shared Task, Florence, Italy, 2/08/19.

Give it a shot: Few-shot learning to normalize ADR mentions in Social Media posts. / Manousogiannis, E.; Mesbah, Sepideh; Baez Santamaria, Selene et al.
Proceedings of the 4th Social Media Mining for Health Applications (#SMM4H) Workshop & Shared Task. ed. / Davy Weissenbacher; Graciela Gonzalez-Hernandez. Association for Computational Linguistics, 2019. p. 114–116.

Research output: Chapter in Book/Conference proceedings/Edited volume › Conference contribution › Scientific › peer-review

TY - GEN

T1 - Give it a shot: Few-shot learning to normalize ADR mentions in Social Media posts

AU - Manousogiannis, E.

AU - Mesbah, Sepideh

AU - Baez Santamaria, Selene

AU - Bozzon, Alessandro

AU - Sips, Robert-Jan

N1 - Conference code: 4

PY - 2019

Y1 - 2019

N2 - This paper describes the system that team MYTOMORROWS-TU DELFT developed for the 2019 Social Media Mining for Health Applications (SMM4H) Shared Task 3, for the end-to-end normalization of ADR tweet mentions to their corresponding MEDDRA codes. For the first two steps, we reuse a state-of-theart approach, focusing our contribution on the final entity-linking step. For that we propose a simple Few-Shot learning approach, based on pre-trained word embeddings and data from the UMLS, combined with the provided training data. Our system (relaxed F1: 0.337- 0.345) outperforms the average (relaxed F1 0.2972) of the participants in this task, demonstrating the potential feasibility of few-shot learning in the context of medical text normalization.

AB - This paper describes the system that team MYTOMORROWS-TU DELFT developed for the 2019 Social Media Mining for Health Applications (SMM4H) Shared Task 3, for the end-to-end normalization of ADR tweet mentions to their corresponding MEDDRA codes. For the first two steps, we reuse a state-of-theart approach, focusing our contribution on the final entity-linking step. For that we propose a simple Few-Shot learning approach, based on pre-trained word embeddings and data from the UMLS, combined with the provided training data. Our system (relaxed F1: 0.337- 0.345) outperforms the average (relaxed F1 0.2972) of the participants in this task, demonstrating the potential feasibility of few-shot learning in the context of medical text normalization.

M3 - Conference contribution

SP - 114

EP - 116

BT - Proceedings of the 4th Social Media Mining for Health Applications (#SMM4H) Workshop & Shared Task

A2 - Weissenbacher, Davy

A2 - Gonzalez-Hernandez, Graciela

PB - Association for Computational Linguistics

T2 - 4th Social Media Mining for Health Applications (#SMM4H) Workshop & Shared Task

Y2 - 2 August 2019 through 2 August 2019

ER -

Give it a shot: Few-shot learning to normalize ADR mentions in Social Media posts

Abstract

Conference

Access to Document

Fingerprint

Cite this