Knowledge hypergraph-based approach for data integration and querying: Application to Earth Observation

According to the world economic forum report, around 70% of generated data are not used. This limitation of usage is mainly due to the lack of interoperability and linking of data that resides in isolated silos. Indeed, the overwhelming amount of data has worsened heterogeneity problems, as have the types of sources generating data in heterogeneous formats and different semantics. Those data related problematics are frequent in the domain of Earth Observation (EO). Earth observed data use different terminologies, which are difficult to reconcile because they reflect overlapped disciplines. These issues lead to misunderstandings and inefficient exchange and management of data in terms of access, pricing, and data rights, which can hamper environmental phenomena understanding. Virtual Knowledge Graph (VKG), allows semantic integration of existing data sources into a wide Knowledge Graph. In this work we propose a knowledge hypergraph-based approach for data integration and querying, with an application to Earth Observation data. Our proposal takes place in two phases (1) a knowledge hypergraph-based virtual data integration and (2) a hypergraph-based query processing. The first phase allows to generate a virtual knowledge hypergraph consisting of RML mappings between an ontology and the data. The second phase consists of enhancing the user’s query by extracting and consolidating a global view of data from different sources based on the generated knowledge hypergraph. The proposed approach is implemented in the Onto-KIT tool (Ontology-based Knowledge hypergraph data Integration and querying Tool) and evaluated through real use case studies. The obtained results show that our proposal enhances query processing in terms of accuracy, completeness, and semantic richness of response.