Annotation is the process or result of attaching labels, tags, or metadata to data so that the data can be interpreted, searched, evaluated, or used to train AI systems. Annotated data may include text, images, audio, video, or structured records, and labels may identify objects, categories, sentiment, grammar, meaning, or other relevant features. High-quality annotation is essential for supervised learning, evaluation, and many model fine-tuning workflows.
In machine learning, metadata attached to a data point or data set to indicate what the data represents.