Data
A dataset is an organized collection of data, such as images, text, or records, gathered for a purpose like training or testing an AI model.
A dataset is data with structure. It might be a spreadsheet of house sales, a folder of 70,000 handwritten digits with their correct values, or a giant archive of web pages. What makes it a dataset rather than a pile is that it is collected on purpose, described, and stored in a consistent format.
In machine learning, a dataset is normally divided. Most of it becomes the training set the model learns from. A separate portion, the test set, is hidden until the end and used to measure honest performance on unseen examples. Mixing the two is a classic mistake that makes results look better than they are.
Famous datasets shaped the field. MNIST digits and ImageNet photos drove computer vision progress. Large web-scale text collections drove language models. Today, building and licensing good datasets is a major industry in itself.
MNIST, a dataset of 70,000 handwritten digits each tagged with the correct number, has been used since the 1990s to teach and test image recognition.