Andrea Giovanni Nuzzolese, researcher at the National Research Council of Italy Institute of Cognitive Sciences and Technologies and leader of WP5 of FOSSR, received the Best Resource Paper Award at the European Semantic Web Conference 2026, one of the leading international events dedicated to semantic technologies and knowledge graphs.
The recognition was awarded for the paper dedicated to Bench4KE (https://link.springer.com/chapter/10.1007/978-3-032-25159-6_12), an open platform developed in the context of the collaboration with the National Interuniversity Consortium for Informatics, within the scope of the FOSSR Ontogenia service, together with Anna Sofia Lippolis of CNR-ISTC and Davide Ragagni, Paolo Ciancarini, and Valentina Presutti of the University of Bologna.
Bench4KE is a benchmarking system designed to evaluate, in a standardized way, tools based on Large Language Models (LLMs) that automatically generate Competency Questions (CQs), namely natural language questions used in knowledge engineering to define the requirements of ontologies and knowledge graphs.
Today, automatic CQ generation systems are often evaluated using different criteria and on isolated case studies, making it difficult to truly compare their performance. Bench4KE addresses this issue by providing:
- a gold standard of 843 manually curated Competency Questions;
- datasets from 17 real-world knowledge engineering projects;
- a suite of metrics to measure the quality of generated CQs;
- an extensible infrastructure also designed for future automation activities in knowledge engineering.
- The platform therefore makes it possible to measure more rigorously how well AI systems are able to support activities typically carried out by knowledge engineers.
Competency Questions are in fact a fundamental tool for the construction of ontologies and knowledge graphs, technologies that make it possible to connect data from heterogeneous sources while preserving their semantic meaning. This approach is particularly relevant in complex domains such as health, environment, cultural heritage, public policy, and social sciences.
The “Resource Paper” category of ESWC recognizes works that introduce scalable and high-quality resources – software, benchmarks, knowledge graphs, and datasets – considered fundamental for the development of high-impact research. The award received represents an important international recognition for the work carried out within FOSSR on open infrastructures for research and on the automation of knowledge engineering.
The Bench4KE code and datasets are publicly available under the Apache 2.0 license.