Text Mining in Systematic Literature Reviews

Systematic Literature Reviews (SLRs) are becoming increasingly important to the decision-making process across healthcare. However, the large quantity and vast scope of information are creating an array of challenges, with resources exceeding the capacity of current research methods and tools. To deal with this mass increase of data, the social sciences are increasingly turning to powerful new technologies such as text mining.

By using text mining techniques to harvest relevant articles, process information and interpret findings, the overall time scale of projects can be reduced and depth of information increased. Literature can now be mapped to identify frequent themes, using search strategies with a balance of sensitivity and precision to capture key terms and concepts. To round off the project, text mining tools can create visualisations of correlations and trends to demonstrate the information required by the client. A great advantage of using text mining software is that it is entirely desk-based, meaning in the current climate, projects can be researched from the comfort of our own homes and offices.

Text mining was used in SLRs by Steve Brewer and the Health Protection Agency over 10 years ago on the FastVac project to research the production and evaluation of emergency vaccinations. An extensive SLR of scientific records was conducted through text mining techniques with the aim of producing a set of predictive rules that would enable rapid deployment of an emergency vaccine.

Text Mining Solutions has recently used SLRs in the agricultural industry by researching sheep welfare. The following diagram shows the welfare risks related to the farming of sheep for wool, meat and milk production. The thickness of the chords between welfare factors encodes the strength of the correlation between the two topics. The chord diagram is a relatively new and powerful visualisation, especially when wanting to display and analyse connections.

You can view our interactive chord diagram by clicking here or the image below

Results achieved through text mining compare favourably to those obtained by traditional methods, however, the tools ensure improved efficiency of data collection, processing and interpretation. It is important to uphold the idea that in SLRs text mining tools must be seen as an asset, rather than an obstacle or substitution.