Every fact back to its source: an evening of data talks in Vienna

A knowledge graph built from sneaker news filled the screen, and the demonstration was about checking it, not admiring it. We were at TUtheSKY, on Vienna's Getreidemarkt, for From Demo to Production, a meetup of the volunteer-run Vienna Data Science Group and part of the European Commission's AI Innovation Month.

Two talks went straight to the work we do for clients. Moritz Wegener of ATVANTAGE showed a pipeline in which every fact in the graph traces back to the document it came from, while an ontology agreed with domain experts decides what a correct fact looks like. In our programme coordination for a family office, a claim is only as good as the evidence filed behind it.

Georg Heiler and Hernán Picatto, of ASCII and the Complexity Science Hub, showed how they run supply-chain graphs drawn from the open web on a supercomputer. One of their slides made the business case in a line: early warning of a shortage is only useful while the graph is fresh. Our supplier qualification for a West African solar venture runs on the same clock, because a supplier stays qualified for as long as its last check holds.

Neo4j and AI Factory Austria co-organised the evening, Rania Wazir of leiwand.ai took on what bias in a system really means and how to test for it, and we ended the night outside with the group's chair. Demos can be admired. A system in production, like a qualified supplier list, has to be checked fact by fact.

Gallery