TY - JOUR T1 - revtools: bibliographic data visualization for evidence synthesis in R JF - bioRxiv DO - 10.1101/262881 SP - 262881 AU - Martin J. Westgate Y1 - 2018/01/01 UR - http://biorxiv.org/content/early/2018/02/12/262881.abstract N2 - Evidence synthesis (ES) is the process of summarizing scientific information, incorporating a range of methods including systematic reviews, systematic maps, and meta-analyses. ES is critical for translating scientific information into recommendations for management or policy, but is becoming increasingly challenging due to rapid growth in publication of scientific literature. From a computational perspective, the ES process can be viewed as a sequence of operations on bibliographic data, including importing, de-duplicating, and classifying large numbers of articles or reports. These are common tasks in natural language processing and machine learning, yet few software tools are available to help researchers access these techniques for manipulating bibliographic data in a systematic way. Here, I present ‘revtools’, an R package for exploratory investigation of bibliographic data during reviews and evidence syntheses. It provides tools for the import and de-duplication of bibliographic data formats, and cluster analysis and visualization of article titles, abstracts and keywords using topics models. The key function of ‘revtools’ is an interactive viewer window that allows users to view and select and export the articles, terms or topics most relevant to their study. Rather than generating lists of text commonly provided by other article sorting software, ‘revtools’ displays this content as points in an ordination cloud, allowing rapid, intuitive assessment of patterns within the corpus. These tools will assist users to identify key topics and outlying articles or terms, increasing the navigability and ease of processing of bibliographic datasets. ER -