The paper “MammoTab 25: A Large-Scale Dataset for Semantic Table Interpretation – Training, Testing, and Detecting Weaknesses” by Marco Cremaschi, Federico Belotti, Jennifer D’Souza (TIB Leibniz Information Centre for Science and Technology) and Matteo Palmonari has received the Best Resource Paper award at the 24th International Semantic Web Conference (ISWC 2025), held in Nara, Japan, from 2 to 6 November 2025.
The contribution was selected from among 19 articles published in the conference proceedings, themselves chosen out of a total of 82 submissions to the ISWC 2025 Resource Track, highlighting the high level of scientific competitiveness of this year’s edition.
MammoTab 25 is a large-scale dataset designed for Semantic Table Interpretation and for evaluating the tabular data reasoning capabilities of modern Large Language Models. The corpus contains 838,930 tables extracted from over 63 million English Wikipedia pages and annotated via Wikidata.
Thanks to these features, MammoTab 25 positions itself as a reference resource for the Semantic Web and AI community.