Speaker
Description
In biodiversity research, synthesizing data from different sources is frequently needed as a prerequisite to answering important questions. Performing these integrations remains a tedious process requiring significant human effort. Often, results of these efforts are not easily reusable for other questions. Knowledge graphs have been proposed as an approach to alleviate this problem in the literature, but have gained little traction in biodiversity research practice so far.
Our contribution showcases the approach to knowledge graph creation, management and usage implemented in the context of PlantHub and the former iKNOW project. We present a knowledge graph combining plant trait sources within the PlantHub project (planthub.idiv.de) including preprocessed data from TRY, a plant trait database, with citizen science occurrence data from naturgucker.de and add taxonomic and additional information from multiple sources (e.g., Wikidata, GBIF, OpenElevation, ...).
We present the workflow needed to create such a graph and show different options for its management using features from Ontotext Refine for data cleaning, API & URL fetching, RDF mapping and export, as well as SQL mapping and export and Ontotext GraphDB for hosting, virtualization, querying, and visualization.
For data that is already stored in databases, we also explored the option to directly integrate those into the knowledge graph via Ontop. This tool enables knowledge graph like access to datasets stored in traditional databases without the need to transform them into RDF. All that is needed is a mapping connecting the data points to show their relationship. Our experiences suggest that this might be easier than the creation of a knowledge graph from scratch and thus lower the barrier to their adoption in biodiversity research.
We present our work and its application in the biodiversity research domain and contribute to bridging the gap towards advanced computer science technologies.
| Status Group | Master Student |
|---|
