Richard Jonker presents knowledge-anchored generation of synthetic data at ECIR 2026

Richard Jonker, PhD student at IEETA, was at ECIR 2026 presenting BioGraphletQA, a framework for systematically generating complex, factually-grounded Question Answering (QA) data. The core of this framework is a novel graphlet-anchored generation process, where small subgraphs from a Knowledge Graph are used as a factual base to control the complexity and ensure the grounding of questions generated by Large Language Models (LLMs). The code and dataset are openly available on https://github.com/ieeta-pt/BioGraphletQA.