AI-Powered Knowledge Graphs for Science-Based Journalism

Short description of the innovation study

This innovation study aims to develop a specialised, lightweight large language model (LLM) to convert academic papers into dynamic knowledge graphs and enable system querying, with an initial focus on climate and energy. The Newsroom will tackle the challenge of bridging the gap between academic knowledge and public discourse by efficiently extracting structured information that journalists can utilise for science-based reporting.

Key innovations include conflict resolution between contradictory sources, temporal tracking of evolving scientific understanding, and rapid indexing of arguments and conclusions. The technology will allow journalists to quickly access comprehensive insights from academic literature—insights that would be difficult to achieve through manual review or traditional database queries. Ultimately, this will enhance the quality of science reporting and contribute to informed public discourse on complex topics.

Organisations involved: 

End User: The Newsroom