Text mining to support abstract screening for knowledge syntheses: A workflow approach

Human and computing resources required for implementing the workflow

This project was part of an on-going collaboration between a knowledge synthesis team, and a laboratory for systems, software and semantics.^1,2 The project team is inter-disciplinary, including reviewers, review coordinators, review methodologists, data analysts and research scientists in semantic computing and knowledge engineering. All analyses were conducted in R.³ One author (JJ) initially developed the R codes for a workshop for doctoral students (all material available online).⁴ Another author (BP) adapted the R codes to the case studies, with problem solving support from other team members through 1-hour weekly meetings. Coding was initially done on a laptop (Intel core i3-4000M CPU@2.4GHz, 4GB RAM, 32-bit operating system) in R Studio,⁵ and run on a server (Linux Ubuntu, 64-bit operating system) through remote-communication freeware between the laptop and the server (Putty and Xming, see the references for the related URL’s),^6,7 and parallel computing.⁸ The initial investment from the SR team was time and effort to foster collaboration with researchers with TM/ML expertise, to arrange for access to computing power, and to provide a supporting environment for SR automation.

Resources are required to integrate the implemented workflow into a review team, including access to TM/ML expertise and high performance computing (through a collaboration as described above), and the acquisition of TM skills by a team member. With TM tools that are increasingly accessible to non-specialists, the dedicated member could consider taking an introductory short-course in TM, or learning the relevant topics online. Our experience suggested that any reviewer with some interest in data analysis could learn to peruse the sample R codes to conduct the TM/ML analysis over a short time period (e.g., 3 months).

Download files

References

KT. Knowledge Translation Program. https://knowledgetranslation.net/.
LS3. Laboratory for Systems, Software and Semantics. http://ls3.rnet.ryerson.ca/.
Team RC. R: A language and environment for statistical computing. 2013; http://www.R-project.org/.
Jovanović J. Society for Learning Analytics Research. 2018; https://solaresearch.org/events/lasi/lasi-2018/lasi18-workshops/.
RStudio. R Studio.
Putty. https://www.putty.org/.
Xming. https://sourceforge.net/projects/xming/.
Zhou M. High performance computing in R using doSNOW package 2014.