Task Info

ToxHabits challenges participants to create systems for three subtasks: (a) detect instances of substance use and abuse in texts, (b) detect pieces of information in text relevant that describe substance use and abuse instances, and (c) create extractive Question Answering systems that can fully characterize the extracted instances. Participants are free to create their systems in any way they want, and the use of creative solutions is encouraged. All systems must be trained and evaluated using the ToxHabits corpus (see Data tab), a collection of 1,500 clinical case reports manually annotated by experts.

• If you want to learn about the relevance of detecting and characterizing substance use and abuse, check the Motivation page.
• For more information and examples about the different subtasks, check the Subtasks page.
• To keep up with all important dates and the task timing, check the Schedule page.
• If you want to participate, check the Registration page.
• If you want to submit your results, check the Submission page.