AIMWORKS – Agentic Multimodal Workflows for Ontology-based Representation and Knowledge Systems

AIMWORKS is developing an AI-enhanced knowledge graph framework tailored for materials science within hydrogen technologies. The process involves several AI agents responsible for planning, collecting and preprocessing data, extracting features from multiple data modes, and iteratively refining and building a robust knowledge graph. AIMWORKS integrates Retrieval Augmented Generation techniques to dynamically pull relevant information from extensive data sources, enhancing the graph’s ability to support sophisticated semantic searches and precise data retrieval. By leveraging the Elementary Multiperspective Material Ontology and Neo4j for graph data management, the framework accommodates the heterogeneous nature of scientific data and enhances system functionality.

The project will extend and synergize the Helmholtz Association’s unHIDE project and Helmholtz-KG with domain capabilities, particularly by enhancing its semantic depth with the specialized ontology for materials science.

AIMWORKS is a joint project between the Helmholtz Centres FZJ and KIT funded within the HMC Project Cohort 2024.