The Hover dataset is a collection of labeled examples for many-hop fact extraction and claim verification tasks. It contains claims, with each claim labeled as either "Supports" or "Refutes". The dataset was created by Yichen Jiang, Shikha Bordia, Zheng Zhong, Charles Dognin, Maneesh Singh, and Mohit Bansal, and was presented in their paper "HoVer: A Dataset for Many-Hop Fact Extraction and Claim Verification" at the 2020 Conference on Empirical Methods in Natural… See the full description on the dataset page:
https://huggingface.co/datasets/Dzeniks/hover.
