by Luc Jonveaux
The PROBONO Horizon 2020 project is exploring new ways of processing, structuring, and using knowledge in sector-specific approaches. In particular, as part of a task aiming and strengthening resilience for health in Green Building Neighbourhoods (GBNs), the task team undertook a feasibility experiment titled "Creating Knowledge Graphs from the Literature: The Case of Health Resilience in GBNs", based on using new technology to accelerate literature review and create synthetic knowledge for enhancing community resilience.
The task specifically aims at reviewings state- of- the- art GBN interventions to mitigate contagious disease outbreaks, and the team has utilized Large Language Models (LLMs) among other technologies to develop a Knowledge Graph (KG) that consolidates a corpus of scientific literature on the subject and use this Knowledge Graph to summarize research in a usable form. This new approach demonstrates the potential for building extensive domain-specific knowledge bases and hints a new way for further exploration in the intersection between LLMs and knowledge graphs.
The technical process relies on a set of tools to parse, structure, and process information from scientific literature. Tools like GROBID, Owlready2, and vector databases can be used for initial data preparation, while natural language processing (backed by tools like NLTK, Spacy) and different LLMs API and tools, facilitated the extraction of themes, entity recognition, and text processing.
This methodical approach allows for the actual structuring of data into a usable knowledge graph, enabling for a comprehensive mapping of key risks, stakeholders, technologies, and mitigation measures at both building and neighbourhood scales. The task outcomes are promising, with the knowledge graph encompassing a vast array of articles, risks, mitigation measures, stakeholders, and technologies related to specific risk mitigation in GBNs, supporting in practice the creation of possible interventions, or 'Blueprints'.
The planned future steps include integrating more robust graph management solutions and enriching the semantic content of the graph, indicating a commitment to evolving the knowledge base to better serve community health strategies through multi-stakeholder discussions. This work, still a research-based exploration, will offer valuable systemic insights and a framework for further dialogue and action about public health and sustainability themes for Green Building Neighbourhoods.
Comments