Towards Multimodal Knowledge Graphs for Data Spaces
Atiya Usmani, Muhammad Jaleed Khan, John G. Breslin, Edward Curry · 2023
Multimodal knowledge graphs have the potential to enhance data spaces by providing a unified and semantically grounded structured representation of multimodal data produced by multiple sources. With the ability to integrate and analyze data in real-time, multimodal knowledge graphs offer a wealth of insights for smart city applications, such as monitoring traffic flow, air quality, public safety, and identifying potential hazards. Knowledge enrichment can enable a more comprehensive representation of multimodal data and intuitive decision-making with improved expressiveness and generalizability. However, challenges remain in effectively modelling the complex relationships between and within different types of modalities in data spaces and infusing common sense knowledge from external sources. This paper reviews the related literature and identifies major challenges and key requirements for effectively developing multimodal knowledge graphs for data spaces, and proposes an ontology for their construction.