Secure Co-Creation of Industrial Knowledge Graph: Graph Complement Method with Federated Learning and ChatGPT

Industrial areas have increasingly developed their own Knowledge Graph (KG) for organizing and leveraging vast amounts of data. One major challenge in constructing KG is the heavy reliance on available resources, restricting the scalability and accuracy of the resulting graphs. To address this issue, an end-to-end method is proposed to create a multi-benefit ecosystem by integrating Federated Learning with ChatGPT (a popular language model). Different stakeholders may leverage ChatGPT to search for novel knowledge that complements their existing KGs, however, this approach could potentially introduce ambiguous and wrong triples into the KG. To overcome this, Federated Learning is applied to align and disambiguate the triples using other industrial KGs as super-vision. The proposed method applies a multi-field hyperbolic embedding method to vectorize entities and edges, which are then associatively aggregated to achieve edge replenishment and entity fusion for each KG encrypted. Finally, an incentive win-win mechanism is proposed to motivate diverse stakeholders to contribute to this co-creation actively. A case study is conducted on different industrial KG to evaluate the proposed method. Results demonstrate that this method provides a practical solution for KG co-creation and no compromise to data security.