gms | German Medical Science

Introduction: Information retrieval plays an essential role during the conduct of any systematic review and involves the development of search strategies to allow for a comprehensive search of the literature. Here, an objective approach is increasingly used to identify candidate search terms from a test set of relevant citations [1]. However, this often requires the application of complex text analysis software, which may constitute a barrier for some information specialists. Thus, we aim to develop a fit-for-purpose and easy-to-use tool to support the routine development of high-quality search strategies for systematic reviews.

Methods: Text-analytic procedures were performed using the quanteda package and included calculating simple word frequencies and analysing keywords-in-context [2]. Further, an interactive graphical user interface was developed using the shiny package [3].

Results: We present an overview of the basic functionalities of the current development version of our tool and give a brief live demo of the corresponding “Shiny App” to demonstrate that information specialists do not require programming skills or familiarity with R to execute the underlying R code. Further, we provide a tentative roadmap towards the first stable release and discuss potential upcoming features.

Conclusion: An R-based tool has the potential to facilitate the identification of candidate search terms for objectively derived search strategies.

Keywords: literature search, text and data mining, R