[QA] Make knowledge base and version selectable in config

[Suganya31]

As a user, I want to see the knowledge base and its version (e.g. DBpedia 2016-10 or Wkidata live) used for each dataset and there should be an option to compare the system answers against the answers from the dataset file or against the live endpoint (via a radio button).

[MichaelRoeder]

Comments

• The KB is bound to the dataset. The version could be bound to it as well (please correct me if I am wrong, here). It might be possible to change the version later on to a newer version but I am not sure whether it makes sense for all KBs. So this version change feature might be skipped for the beginning.
• For reproducability it is important to be able to track whether a live endpoint or loaded answers have been used.
• I would suggest to leave the front end as it is and do not let the user choose this for uploaded datasets since it would make things much more complicated (e.g., tracking whether an endpoint has been used and the address of it becomes challenging without the new database structure from the main gerbil branch).

Solution

• [ ] add the possibility to define the name and version of the KB when defining a dataset (in the dataset.properties file)
• [ ] add the possibility to define a live endpoint for a KB
  • This might be done together with the dataset OR in the gerbil.properties file (or even an additional kb.properties if necessary)
• [ ] update the evaluator factory to ask the dataset object for the way it needs to be evaluated. Based on that, it can choose a answer set based solution (the current implementation) or the live endpoint. Note that a new datatype might be necessary to store the SPARQL query and evaluate it.
• [ ] when a SPARQL query is used, make sure to store the date on when this happened on the experiment metadata, i.e. RDF, also, to have a repeatable experiment