DynBench: instantiated Benchmarks

Examples of new benchmark datasets produced by DynBench. DynQALD contains records in five different languages (en, de, fr, ru, uk). DynRuBQ contains records in wo languages (en, ru). The generated DynQALD dataset has 1791 unique questions in total, i.e., 356 unique new entities. The average length of the questions is 7.4 words in English, 5.9 words in Russian, 7.0 words in German, 7.6 words in French, and 5.8 words in Ukrainian. The complexity level is the same as in QALD-9-plus. The generated DynRuBQ dataset has 4011 unique questions in total, i.e., 1773 unique new entities. The average length of the questions is 8.8 words in English and 6.5 words in Russian. The complexity level is the same as in RuBQ 2.0. The main difference to the original dataset is that DynRuBQ does not contain, at the moment, the unanswerable questions.