GraphQuestions is a characteristic-rich dataset designed for factoid question answering. The dataset aims to provide a systematic way of constructing QA datasets with rich and explicitly specified question characteristics. Here are some key details about GraphQuestions:
GraphQuestions consists of a set of 5,166 factoid questions.
Each question is associated with logical forms and ground-truth answers.
The dataset is constructed based on Freebase, a large-scale knowledge base.
An array of question characteristics is formalized for each question.
Variants: GraphQuestions
This dataset is used in 2 benchmarks:
| Task | Model | Paper | Date |
|---|---|---|---|
| Question Answering | ChatGPT | Can ChatGPT Replace Traditional KBQA … | 2023-03-14 |
| Semantic Parsing | ReasonBERTR | ReasonBERT: Pre-trained to Reason with … | 2021-09-10 |
Recent papers with results on this dataset: