ABSTRACT: Automatically generating question answer (QA) pairs from the rapidly growing coronavirus-related literature is of great valueto the medical community. Creating high quality QA pairs would allow researchers to build models to address scientific queries for answers which are not readily available in support of the ongoing fight against the pandemic. QA pair generation is, however, a very tedious and time consuming task requiring domain expertise for annotation and evaluation. In this paper we present our contribution in addressing some of the challenges of building a QA system without gold data. We first present a method to create QA pairs from a large semi-structured dataset through the use of transformer and rule-based models. Next, we propose a means of engaging subject matter experts (SMEs) for annotating the QA pairs through the usage of a web application. Finally, we demonstrate some experiments showcasing the effectiveness of leveraging active learning in designing a high performing model with a substantially lower annotation effort from the domain experts.
FIGURE 1: Generation of the Silver QA Data and the Process of Obtaining Gold Data using Active Learning
For details about our dataset and experimental setup, please read our paper titled: A Smart System to Generate and Validate Question Answer Pairs for COVID-19 Literature
Pre-processing of the CORD-19 dataset and building the rule-based QA can be done from the following notebook: rulebased_model/Preprocess-and-RuleBased.ipynb
Scripts for the seq2seq (generated QA) model can be accessed from here
Notebooks for the active learning experiments are available here
Please consider citing our work if you found it useful in your research:
@inproceedings{bhambhoria-etal-2020-smart,
title = "A Smart System to Generate and Validate Question Answer Pairs for {COVID}-19 Literature",
author = "Bhambhoria, Rohan and
Feng, Luna and
Sepehr, Dawn and
Chen, John and
Cowling, Conner and
Kocak, Sedef and
Dolatabadi, Elham",
booktitle = "Proceedings of the First Workshop on Scholarly Document Processing",
month = nov,
year = "2020",
address = "Online",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2020.sdp-1.4",
doi = "10.18653/v1/2020.sdp-1.4",
pages = "20--30",
abstract = "Automatically generating question answer (QA) pairs from the rapidly growing coronavirus-related literature is of great value to the medical community. Creating high quality QA pairs would allow researchers to build models to address scientific queries for answers which are not readily available in support of the ongoing fight against the pandemic. QA pair generation is, however, a very tedious and time consuming task requiring domain expertise for annotation and evaluation. In this paper we present our contribution in addressing some of the challenges of building a QA system without gold data. We first present a method to create QA pairs from a large semi-structured dataset through the use of transformer and rule-based models. Next, we propose a means of engaging subject matter experts (SMEs) for annotating the QA pairs through the usage of a web application. Finally, we demonstrate some experiments showcasing the effectiveness of leveraging active learning in designing a high performing model with a substantially lower annotation effort from the domain experts.",
}
For any queries about this work, please contact: Rohan Bhambhoria at [email protected]
The authors would like to thank all the organizers of the COVID-19 Open Research Dataset Kaggle Challenge. We would like to thank Vector Institute for making this collaboration possible and providing academic infrastructure and computing support during all phases of this work. We would also like to thank Richard Pito from Thomson Reuters for his invaluable feedback and support throughout this project. Last but not least, special thanks to Dr. Frank Rudzicz and Dr. Xiaodan Zhu for their academic supervision and insights.