AI-Ready Data in Biodiversity Informatics: Why Machine-Readability Matters

Biodiversity science relies on diverse and complex data sources to trace changes in species distribution and their causes and uses this information to help address global environmental challenges. AI tools can support in tackling these challenges, but are biodiversity data AI-ready? Only digital data that can be managed via automatic pipelines and consumed directly by a machine learning model or a deep neural network can be considered AI-ready. Therefore, making data and metadata available in a machine-readable format is the first step to make them AI-ready. For this reason, a Github repository has been created to provide an overview of suitable machine-readable data formats in biodiversity informatics, illustrate their context of application, and give advice against possible pitfalls. Both machine-readable data formats of general use (e.g., CSV) and specific to biodiversity science (e.g., DwC-A) are examined.