Apologies for cross-posting. --------------------------------------------------------------------------- There is still time to participate in this shared task---evaluation phase from April 17 to 24.
**Social Media Mining For Health 2024** https://healthlanguageprocessing.org/smm4h-2024/
The Social Media Mining for Health (SMM4H) workshop and shared tasks have been running successfully since 2016. They now go into the 9th round, with the workshop being co-located at ACL 2024 in Bangkok. https://2024.aclweb.org/ Bangkok, Thailand , August 12–17, 2024
**Important Dates for all SMM4H Shared Tasks**
Training data available: January 10, 2024 CodaLab Available: January 17, 2024 Evaluation Phase: April 17 - 24, 2024 System description paper due: May 17, 2024 Paper acceptance notification: June 17, 2024 Camera-ready papers due: July 1, 2024 Workshop in Bangkok, Thailand , August 15, 2024
**Task 2: Task Description**
Adverse Drug Events (ADEs) are negative medical side effects related to a drug. Mining ADEs from user-generated text has become a popular topic and is an important use case for research, as it could help detecting crowd signals from users online. Being able to make use of information across languages by applying multi-lingual methods further supports this endeavor. Our task targets the languages *German, French and Japanese* and is split into two subtasks. Subtask 2a focuses on Named Entity Recognition (NER) of of medication, disorder, and function mentions from user-generated texts. Subtask 2b performs joint NER and Relation Extraction (RE) to determine if these disorders are ADEs by finding the correct relations between medications, disorders and functions. We distinguish two types of relations between medication mentions and disorder/function mentions:
- "caused": the disorder/function was caused by a medication, i.e., the disorder/function is an ADE - "treatment_for": the disorder/function is the reason for the medication, i.e., the medication is supposed to treat the disorder/function
*Tasks*: Participants can choose between participating in subtask 2a, or subtask 2b, or both. ~~ We explicitly encourage the submission of new and creative approaches! ~~
- Subtask 2a) Named entity recognition of the entities "drug", "disorder" and "function" from user-generated texts. - Subtask 2b) Joint named entity and relation extraction of the entities "drug", "disorder" and "function" and the relations "caused" and "treatment_for".
*Data*: The data originates from social media platforms, e.g., patient fora and X (Twitter). We provide data in German and Japanese, and a few examples in French. The submitted systems will be evaluated on German, French and Japanese data. Please find more information here: https://healthlanguageprocessing.org/smm4h-2024/
Please use this form to register: https://forms.gle/7w4si27uJrCMiTyL8
Organizers of Subtask 2:
Pierre Zweigenbaum, Université Paris-Saclay, CNRS, LISN, France Sebastian Möller, Technische Universität Berlin, DFKI GmbH, Germany Roland Roller, DFKI GmbH, Germany Philippe Thomas, DFKI GmbH, Germany Eiji Aramaki, NAIST, Japan Shoko Wakamiya, NAIST, Japan Shuntaro Yada, NAIST, Japan Katherine Yeh, Université Paris-Saclay, CNRS, LISN, France Lisa Raithel, Technische Universität Berlin, Germany & Université Paris-Saclay, CNRS, LISN, France
-- Pierre Zweigenbaum Senior Researcher Université Paris-Saclay, CNRS, LISN