[QA] Add MQALD dataset

[lsiciliani]

Hi, I am opening this is issue to propose the inclusion of MQALD in Gerbil. MQALD is a novel dataset composed of 100 novel handcrafted questions, each of them requiring the use of one or more SPARQL modifiers to retrieve the right answer.

The dataset is now publicly available on Zenodo at the following link: https://zenodo.org/record/4479876#.YEeUFVNKius

The dataset structure is compliant with that of the QALD datasets except for an additional field (a JSON array named "modifiers" which lists the modifiers needed for each question). The underlying KB is DBpedia.

Besides adding new questions (contained in the MQALD_new_query.json file on Zenodo), we also extracted questions with modifiers from the last three editions of the QALD challenge (9-8-7). We performed a merge of the training and test files (QALD-train-MOD-multilingual.json and QALD-test-MOD-multilingual.json on Zenodo).

Thank you very much!

[RicardoUsbeck]

Dear Lucia,

we will integrate it as soon as we can but I cannot promise a timeline yet.

Best regards Ricardo