Here is some basic usage and privacy information for the SPARQL endpoints at https://qlever.cs.uni-freiburg.de, in particular, for the Wikidata SPARQL endpoint at https://qlever.cs.uni-freiburg.de/wikidata.

Usage information

When launching a query via the UI on https://qlever.cs.uni-freiburg.de, you can track the progress of a query by clicking on the Analysis button. When a query takes longer than you expected, or you realize that you made a mistake, it is good practice to cancel your query (the Execute button turns into a Cancel button while the query is running). You can download results via the Download button. You can also access the SPARQL endpoints via a variety of API calls; see the Share button for details. The current timeout is at least 600s.

If you are a heavy user, we encourage you to set up a SPARQL endpoint on your own machine. You don't have to be a pro for this and it can be done with only a few commands; see https://github.com/ad-freiburg/qlever-control . Downloading the complete Wikidata dataset takes 5-7 hours (with a reasonably fast internet connection). Loading the data into QLever takes 4 hours on a modern PC, and twice as long if you have an old machine. 32 GB of RAM is sufficient.

Privacy information

We currently store standard Apache logs (containing IP addresses), which are deleted after a week. We only look at these logs when necessary, for example, when there is a very large number of queries coming from the same IP address. The SPARQL endpoint also has a log, but there we only see the queries and no IP addresses. These logs are currently also not stored permanently, but we reserve the right to save interesting queries for later analysis.

Idiosyncrasies

QLever is SPARQL-compliant since 30.06.2025. Like for most engines, there are small deviations, which fall into one of the following three categories: intentional, very rare border cases, and bugs. Here is a list of the current differences to the SPARQL 1.1 standard. Here is an overview of the differences between the QLever SPARQL endpoint for Wikidata and the Wikidata Query Service (WDQS).

Issues

When you encounter an error or an unexpected query result, first check whether the error is on your side (in our experience, that is the most frequent case), or whether you ran into a missing feature or idiosyncrasy as listed in the previous section. The "Analysis" button can help you to understand, how QLever processed your query and which part took how long. If you can't figure out a problem on your own or if you think you hit a bug, you are welcome to open an issue on https://github.com/ad-freiburg/qlever/issues.