The speed and volume of scientific publishing is accelerating, both in terms of number of authors and in terms of the number of publications by each author. At the same time, the demand for knowledge synthesis and dissemination is increasing in times of upheaval in the education sector. For systematic reviewers in the field of education, this poses a challenge in the balance between not excluding too many possibly relevant studies and handling increasingly large corpora that result from document retrieval. Efforts to manually summarise and synthesise knowledge within or across domains are increasingly running into constraints on resources or scope, but questions about the coverage and quality of automated review procedures remain. This article makes the case for integrating computational text analysis into current review practices in education research. It presents a framework for incorporating computational techniques for automated content analysis at various stages in the traditional workflow of systematic reviews, in order to increase their scope or improve validity. At the same time, it warns against naively using models that can be complex to understand and to implement without devoting enough resources to implementation and validation steps.
Read full abstract