Automating Quality Checks in the Publishing Process

Current Landscape

The scientific community contains many stakeholders: researchers, funders, institutions, publishers, and the public. Each stakeholder plays a key role in this ecosystem, yet no one has full control. Publishers are not responsible for the science but hold the crucial role of reporting the research. Funders are responsible for paying for but typically not conducting the science. Researchers conduct the science and are experts who may still lack expertise in all areas of fully reporting science. Improvement in research quality will come when multiple actors in this network make changes.

To radically change the scientific publishing review, the scientific community must move from manual processes based on peer-review and editor knowledge to automated pipelines programmatically assessing the reported research. Technically this entails items such as i) Obtaining machine-readable information from publications across disciplines and publishers; ii) Transforming the data into a normalized and meaningful structure for future use; then, iii) Using machine learning and natural language processing to accurately identify and extract text. As with any technological solution, there also needs to be the ability to have this solution adopted in practice. Let us first delve into the stakeholders and current practices taking place to ensure scientific rigor in preventing the dissemination and consumption of non-rigorous, non-reproducible science.