Knowledge Graph Infrastructure

BERD develops knowledge graph infrastructure for German company data distributed over many providers, registers and time spans. Many valuable datasets are still confined in analogue books. We use a variety of algorithms to OCR, structure and semantify the unstructured data and to create knowledge graphs.

The knowledge graphs:

• The Aktienführer (AKF) Knowledge Graph contains structured (meta)data for the German listed stock companies from the Hoppenstedt-Aktienführer from 1956 to 2018. We enriched the metadata with numerous external identifiers.
  • AKF SPARQL
  • AKF Reconciliation API
  • AKF API
  • Advanced search
  • CSV-Export

The tools for entity linking, entity typing and data enriching:

• bbw: tool for automatic semantic annotations of tabular data (entity linking, entity typing and relation extraction on Wikidata)
• spaCyOpenTapioca: spaCy wrapper of OpenTapioca for named entity linking on Wikidata
• OpenRefine: a powerful free, open source tool for working with messy data: cleaning it; transforming it from one format into another; and extending it with web services and external data

Ongoing projects:

• Maschinen-Industrie im Deutschen Reich
• Reichsanzeiger