Workshop: How close are we to reliably (semi-)automating study selection in systematic reviews?

Objectives: To discuss: 1. The different ways that text mining technologies might be able to help with study identification 2. The current evidence base for these technologies – i.e., do they work? 3. The broader implications of (semi-)automating study selection, including the potential risks, the skills needed to execute, and the acceptability to different types of review users. Description: The large and growing number of publications makes identifying relevant studies both complex and time consuming. Text-mining has been offered as a potential solution: through automating some of the screening process, reviewer time can be saved. We will outline the different ways in which text mining technologies are being applied to study selection in systematic reviews. This will include advanced techniques that combine text mining with machine learning, including term-based clustering and ‘topic modelling’, where the traditional lines which distinguish searching, screening, and mapping become blurred. We will also provide an update on current functionality of the Evidence Pipeline, which is part of the Cochrane ‘Project Transform’. We will also summarise the current evidence base in an attempt to answer the question: how close are we to being able to reliably replace some of the manual labour in study identification with text mining technologies? Small groups (with whole-group feedback) will try out a selection of tools, including the Evidence Pipeline, and discuss the methodological issues that the use of these technologies raise. This may feed in to the content of emerging guidance on their use