Quotation marks in names have caused errors before when encoding the SPARQL queries. This was fixed by stripping quotation marks from the names, when replacing the placeholder in the question and query.
However, when building the graph ttl dataset, the quotation marks are weirdly placed in the identifyingName:
The name in the translated query contains no quotation marks, so this causes the query to return empty result.
TO DO:
[x] use strip_item function from generator_utils to strip quotation marks from the legalNames object when copied to the identifyingName object
As follow up, the evaluation on models A to H has to be rerun. Some accuracy gains should be seen. Also the gap in performance between name match and query result should decrease.
Quotation marks in names have caused errors before when encoding the SPARQL queries. This was fixed by stripping quotation marks from the names, when replacing the placeholder in the question and query.
However, when building the graph ttl dataset, the quotation marks are weirdly placed in the identifyingName:
The name in the translated query contains no quotation marks, so this causes the query to return empty result.
TO DO:
strip_item
function fromgenerator_utils
to strip quotation marks from the legalNames object when copied to the identifyingName objectAs follow up, the evaluation on models A to H has to be rerun. Some accuracy gains should be seen. Also the gap in performance between name match and query result should decrease.