RAGE-KG 2025 at ISWC 2025: A new award for Yousouf Taghzouti, Franck Michel, Tao Jiang, Louis Felix Nothias, and Fabien Gandon
- Award
on the November 5, 2025
Yousouf Taghzouti, Franck Michel, Tao Jiang, Louis Felix Nothias, and Fabien Gandon won the Best Paper Award: Agentic AI.
The paper “User Interface and Agent Interface for Online Generation of Knowledge Graph’s Competency Questions and Question-Query Training Sets” co-authored by Yousouf Taghzouti (Postdoctoral Researcher), Franck Michel (Research Engineer and Researcher), Tao Jiang (Researcher and Associate Professor), Louis-Félix Nothias (Junior Professor associated with the 3IA), Fabien Gandon (3IA Chairholder)* has been awarded the Best Paper Award: Agentic AI at the 2nd International Workshop on Retrieval-Augmented Generation Enabled by Knowledge Graphs (RAGE-KG 2025), held during the 24th International Semantic Web Conference (ISWC 2025). This work was supported by 3IA Côte d'Azur, MetaboLinkAI, ANR, Université Côte d’Azur, Inria, CNRS, and i3S/CNRS.
Abstract: Few question-query datasets exist for fine-tuning large language models on tasks such as translating natural language questions into SPARQL queries. While it is often recommended that competency questions and their corresponding SPARQL queries accompany a knowledge graph (KG), this is rarely the case in practice. In this paper, we introduce Q2Forge, a web application designed to support the creation of question-query pairs for any KG. The tool enables users to generate, test, and refine competency questions and their SPARQL counterparts directly within the interface. It employs a retrieval-augmented generation architecture to support a wide range of KGs efficiently. The result is an open-source solution for building reusable question-query datasets applicable to any KG. We also present recent developments around the Model Context Protocol, moving toward the agentification of Q2Forge —enabling natural language interactions in addition to traditional UI-based workflows.
* The same team co-authored the paper "Q2Forge: Minting Competency Questions and SPARQL Queries for Question-Answering Over Knowledge Graphs" accepted at K-CAP 2025.
Read full paper
Learn more about ISWC 2025
Learn more about RAGE-KG 2025
