Planetary Knowledge Base: Semantic transcription using graph neural networks

The Natural History Museum, London (NHM), in collaboration with Amazon Web Services (AWS), has embarked on a project to build the Planetary Knowledge Base (PKB), comprehensive graph network comprising data all specimens, collectors, and localities. In initial prototype, we have concentrated botanical using plant taxa specimens within Global Biodiversity Information Facility (GBIF), combined geographic from GeoNames biographic WikiData, Bionomia, Harvard Index of Botany, TL2, Tropicos. Development PKB is huge undertaking—our first proof concept more than 100 million nodes. primary application this knowledge (KG) powering automated transcription specimen labels. Using Graph Convolutional Neural Networks, textual information labels can be aligned entities graph, creating structured semantic raw text. Text extracted images services AWS ecosystem, including Optical Character Recognition Language Processing identify units information, high-throughput auto-digitisation workflow for extracting data. enables new ways interrogate collections. It help species that may require re-examination or re-identification due taxonomic updates inconsistencies. also flag potential discrepancies conflicts data, such as cases where same recorded under different names classifications across various sources. Moreover, detect possible errors outliers point out could represent misidentified collection. By cross-validating International Union Conservation Nature (IUCN) Red List, it assist analysing populations insufficient being developed cloud service, so researchers other institutions experiment transformative technology, support their own digitisation efforts.