- Software & Tools Marketplace: FAIRenrich – FAIR Enrichment and Curation for ESS and Biodiversity Data FAIRenrich is a tool developed within the BITS project that bridges the gap between manual annotation bottlenecks and the need for FAIR-compliant scientific data management in Earth Sciences and Biodiversity research. It leverages advanced NLP models for text structure recognition and integrates with the ESS Collection of the TIB Terminology Service to generate FAIR-compliant, semantically enriched (meta)data. In general, the tool was received with great interest, the open Source development of API and Code were particularly well received (https://github.com/Senckenberg-DCBiodivIT/BITS). It became clear in many discussions that there is need for improvements concerning quality assurance: even an error rate of only 1% may lead to an enourmous amount of data(sets) that would need to be corrected by humans. Also, it should be possible to adjust the mappings retrospectively, which is currently not the case.
- Barcamp: Semantics in Science – Help or Pain?
Scientific progress depends on shared knowledge, yet semantics can act as both a bridge and a barrier. With this in mind, we offered a Barcamp Session which was very well attended and lead to a lively discussion about the usage of vocabularies, ontologies and other semantic artifacts in different domains. Many obstacles line the path, beginning with the decision which terminology to use: what are the quality criteria for "good" terminologies? Recommendation systems? Mandatory requirements for specific domains? Who is responsible for registering and maintaining domain specific terminologies? Furthermore, who should be the person responsible for a) choosing a terminology and b) using it for annotation of metadata: the scientist herself? The person that publishes the data? Even though there was no one clear answer it became obviuos that the usage of semantics is research data management is crucial to improve the FAIRness of data. The only statement all participants could agree on was that we need reliable and trustworthy tools that help to automate metadata annotation - and that AI can support but will not solve the problems in the short run. - Poster: FAIRenrich: Empowering FAIR and Scalable Annotation Workflows in Earth System and Biodiversity Sciences
Our Poster (created by Alexander Wolodkin, Jonas Grieb and Claus Weiland) highlighted the modular and configurable architecture of FAIRenrich and its impact on sustainable, FAIR-compliant workflows. Discussions at the Poster session likewise lead to principle questions concerning AI: basic quality assurance should be human centred, trustworthiness is key.
AI and its capabilities and limitations is an important topic, so the German Climate Computing Center (DKRZ) as a BITS Project Partner offered a workshop on "Quality time: Rethinking Data Trust in the Age of AI and (Cross-Domain) Collaboration" which was really well attended. After three lighning talks for AI implementations in different domains, a world café offered the opportunity to discuss and exchange both ideas and concerns. The workshop results were presented at the Plenum.
Finally, BITS members used the opportunity to gather feedback for the project deliverables and activities by doing interviews and surveys. As soon as the output is analysed, we will publish it on the RESULTS tab of this website. Watch out!
