Covid-19QA [1] is a new Covid-19 related QA dataset, where the questions were generated from six kinds of "wh" question templates. All of questions require one-hop reasoning. This dataset is based on COVID-19KG, a knowledge graph about Covid-19 via Named Entity Recognition and Relation Extraction algorithms from PubMed biomedical texts. Covid-19KG contains 11,264 entities, 8,038 relations, and 123,190 triples.
| Model / System | Year | Hits@1 | Language | Reported by |
|---|---|---|---|---|
| Qiao et al.'s proposed method | 2022 | 29.00 | EN | Qiao et al. |
| EmbedKGQA | 2022 | 27.7 | EN | Qiao et al. |
| EmbedWBioBERT | 2022 | 7.50 | EN | Qiao et al. |
| EmbedWBERT | 2022 | 6.40 | EN | Qiao et al. |
[1] Qiao, Yinbo and Yang, Zhihao and Lin, Hongfei and Wang, Jian. An End-to-End Knowledge Graph Based Question Answering Approach for COVID-19. In China Health Information Processing Conference, pp. 156–169. 2022.