This paper investigates various Transformer architectures on the WikiReading Information Extraction and Machine Reading Comprehension dataset. The proposed dual-source model outperforms the current state-of-the-art by a large margin. Next, we introduce WikiReading Recycled-a newly developed public dataset and the task of multiple property extraction. It uses the same data as WikiReading but does not inherit its predecessor's identified disadvantages. In addition, we provide a human-annotated test set with diagnostic subsets for a detailed analysis of model performance.
@article{arxiv.2011.03228,
title = {From Dataset Recycling to Multi-Property Extraction and Beyond},
author = {Tomasz Dwojak and Michał Pietruszka and Łukasz Borchmann and Jakub Chłędowski and Filip Graliński},
journal= {arXiv preprint arXiv:2011.03228},
year = {2020}
}
Comments
Accepted at CoNLL 2020; this article supersedes arXiv: 2006.08281