Rapid Review on Publicly Available Datasets for Health Misinformation Detection

Area of Science:

• Public Health
• Information Science
• Computational Linguistics

Background:

• The rapid spread of health misinformation necessitates robust detection methods.
• Numerous datasets have been developed since 2020 to address this challenge.
• A significant portion of these datasets focus on COVID-19 related misinformation.

Approach:

• This review systematically examines publicly available datasets for health misinformation detection.
• It analyzes implementation strategies and dataset characteristics.
• The review categorizes datasets based on their data sources and annotation methods.

Key Points:

• Half of the identified datasets are specifically related to COVID-19 misinformation.
• Most datasets rely on fact-checking websites, with fewer expert-annotated resources.
• Some datasets include valuable metadata like social engagement and explanations.

Conclusions:

• The reviewed datasets represent a crucial resource for researchers and practitioners in the field of health misinformation.
• These datasets facilitate the study of misinformation propagation and the development of effective countermeasures.
• Further development of expert-annotated datasets is recommended for enhanced accuracy.