Academic, Scientific & Research Datasets

These portals provide datasets used in peer-reviewed research, academic collaborations, and institutional projects. They are often high-quality, well-documented, and domain-specific—ideal for AI model evaluation, reproducibility studies, or cross-discipline insights.

Tools:

• Papers with Code – Datasets – Links benchmark datasets with research papers and model performance. Great for comparing SOTA across tasks.

• AllenAI Datasets – Curated by the Allen Institute for AI, includes NLP, vision, and reasoning datasets like SciQ and Aristo. Ideal for language understanding and commonsense reasoning tasks.

• IEEE DataPort – A platform for scientific datasets, competitions, and academic benchmarks. Supports fields like IoT, robotics, and telecommunications.

• Mendeley Data – A repository for scientific data linked to publications. Encourages reproducibility and collaboration across research fields.

• B2SHARE (EUDAT) – European data infrastructure hosting cross-discipline research datasets. Includes geoscience, engineering, and climate data.

• DL ACM Dataset (MovieLens) – MovieLens-based benchmark for collaborative filtering and recommender systems.