Overview
PeerScout was a machine learning tool that eLife developed to suggest peer reviewers for manuscripts submitted to eLife, choosing from an existing pool of reviewers based on the papers they had written and reviewed. Daniel Ecer built the first version in 2017 as a six-week project on the ASI Fellowship. At that time eLife editors chose reviewers from a database of about 40,000 people by filtering on subject area and then on keywords. Ecer’s model combined those keywords with natural language processing of the manuscript text to give each potential reviewer a suitability score, so that editors could find reviewers who knew the topic but had described their work with different keywords. For each manuscript, the prototype listed potential reviewers with their suitability score, their average review time and the number of review invitations they had not yet answered. It also flagged early career researchers, using an external data source to match them because eLife held less data about them.
In 2018 eLife described PeerScout as using machine learning, including natural language processing, to find relevant reviewers, with a deliberate bias towards surfacing early career researchers while leaving the final choice to editors. eLife released the code as open source under the MIT licence.
Notes and quotes
- In 2018 Ecer and Paul Shannon, eLife’s head of technology, said that with PeerScout the team faced “a social problem in convincing editors that these recommendations are worthwhile”.
- eLife archived the PeerScout code repository on GitHub on 25 October 2019, which made it read-only.
Sources
- Daniel Ecer at ASI Demo Day 7, Faculty, YouTube, 29 March 2017
- AI for automation and influence in open science publishing, Daniel Ecer and Paul Shannon, O’Reilly Artificial Intelligence Conference, 10 October 2018
- elifesciences/peerscout, GitHub
This entry was written with the help of AI, from the sources listed. Please check the sources before relying on it, and let us know if anything needs correcting.
