Limit number of results that are returned from knowledge bases

[JanVana]

Sometimes there are too many results returned from the knowledge base. In these situations, it could be better to limit the number returned results and add warning that the value is too vague.

[tomas-knap]

I would suggest to have in the config file an (optional) param "max number of suggestions". If available, this should cause that only such number of suggestions is returned per each case (those with highest score). By default, this should not be restricted.

[JanVana]

@tomas-knap I agree that it should be an optional parameter. But returning a limited number of results based on score would defeat the purpose of this improvement. The main motivation is to handle situations, where there are so many results returned from the knowledge base, that processing them would take impractically long.

For this reason I propose to limit the number of results in main SPARQL search queries (exact match and fulltext search) and display a warning for the user when this limit is reached.

[JanVana]

An example of query, that returns quite a lot of results:

SELECT DISTINCT ?subject ?object
WHERE
  { { SELECT ?subject ?object
      WHERE
        {   { ?subject <http://xmlns.com/foaf/0.1/name> ?object .
              ?object <bif:contains> "\"Book\""
            }
          UNION
            { ?subject  <http://www.w3.org/2000/01/rdf-schema#label> ?object .
              ?object <bif:contains> "\"Book\""
            }
        }
    }
    ?subject a ?class
    { SELECT ?class
      WHERE
        {   { ?class a <http://www.w3.org/2000/01/rdf-schema#class> }
          UNION
            { ?class a <http://www.w3.org/2002/07/owl#Class> }
        }
    }
  }

[tomas-knap]

Ok, but still I would prefer to have config.param - which may be turned on by default (which is a difference from what I proposed above)

[tomas-knap]

(so that I can easily switch that off when testing/playing with the algorithm, without the need to remove all limits etc. )