The EU Knowledge Graph
Give a brief description of your project, specifically mentioning why you chose to use Wikibase.
The EU Knowledge Graph is a knowledge graph containing information about the European Union such as countries, capitals, funds, EU-financed projects and the beneficiaries of that funding, regional information and much more. We decided to use Wikibase because it allows us to organize heterogeneous data in one place, with many advantages: collaborative editing, non-expert user interface for reading the data, established data model, query service, OpenRefine integration, history tracking and more.
There are a number of ways to import data into Wikibase. Describe the criteria that influenced your decision to use your chosen method.
We ingest data both manually and programmatically. For manual editing Wikibase is perfect. The type-as-you-search functionality allows us to find existing concepts, and quality constraints indicate whether there is a problem. We use this type of editing mainly for ontological concepts. The bulk of the data is imported programmatically. For this we use Pywikibot. The configuration is a bit challenging, but once done, it is easy to use. One problem that we encountered was the slow data ingestion speed, which created a bottleneck in our deployment. We therefore developed a Batch Extension which allowed us to accelerate ingestion speed by a factor of ten.
Did Wikibase meet your needs? Describe the challenges and successes you faced when implementing Wikibase.
Wikibase is great; it comes with a lot of features out of the box. At the beginning there is not much to worry about, and many services and tools are available for free. This allowed us to move quickly into production and to scale to a graph with nearly 1 billion edges. However, when you need to customize things, it gets more challenging. But there is a great community out there ready to help.