Extreme Systematic Reviews: A Large Literature Screening Dataset to Support Environmental Policymaking

Short conference paper for CIKM2022, 31st ACM International Conference on Information and Knowledge Management.

Impact/Purpose

ISAs can be viewed as unusually large systematic reviews that are cyclical, comprise an unusual diversity of disciplines and study types, and require extremely high recall. These characteristics result in high time demands on EPA scientific experts who screen published articles for inclusion. Ranking models and machine-assisted screening methods can decrease the time of screening by prioritizing the order in which references found through literature searches are screened, such that the non-relevant references are gathered far down the screening queue, and exhausting the queue becomes unnecessary. This sub-product describes first steps in improving ranking methods and screening methods by leveraging the the unique characteristics of ISAs. Results will extend to other EPA assessments, and to systematic reviews in general.

Citation

Hou, J., X. Wang, J. Dubois, R. Rice, A. Haddock, AND Y. Wang. Extreme Systematic Reviews: A Large Literature Screening Dataset to Support Environmental Policymaking. In Proceedings, CIKM2022, Atlanta, GA, October 17 - 22, 2022 Association for Computing Machinery (ACM), New York, NY, 4029�4033, (2022). [DOI: 10.1145/3511808.3557600]

DOI: Extreme Systematic Reviews: A Large Literature Screening Dataset to Support Environmental Policymaking