SAGE Research Methods explorer

What did we experiment with?

We wanted to explore the relationship between the key terms from SAGE Research Methods (SRM) using visualization and classification tools to see if we could extract meaningful relationships between SRM in a given discipline. This would allow users to understand their field better, find new methods, or find methods that are typically applied together in their discipline, while also leading them to SAGE content about these methods or a combination of methods.

HOW DID WE DO IT?

We annotated the SAGE Journals and SRM corpora with the research methods keywords using Luxid (a content enrichment platform) which also assigned an 'aboutness' score to every keyword linked to a document. For example, a chapter in a Blue Book could be scored at 6.7 about 'structural equation modelling', 2.5 about 'latent variables' and at 1.1 about 'covariance' matrix, and this doesn't just count the number of occurrences.

Based on these scores, and filtered by type of corpus (currently psychology vs education vs all SRM), we mapped the keywords in a network of co-occurrences (how often they occur within the corpus of a given field, and how often they occur together in a document). This final output was a JavaScript-based interactive visualization. When hovering over one of the popular research methods, you can also see how significant its relationships with other methods  are (or how often they are used together in this particular field). To give you a flavor of the outcome, here is a complex though static snapshot:

Screen Shot 2018-07-27 at 11.29.52.png

What did we learn and OTHER IDEAS?

The visualization shows a lot of promise in supporting an understanding of all the research methods used in particular disciplines, but it still needs a lot of refinement. At the moment, it is still quite challenging to navigate the map in its totality and identify more meaningful relationships that may be less well known. These types of data visualizations are very engaging and give a great first impression, but as we learned, they required careful crafting and a good design in order to provide genuinely useful insights. Going forward, this is exactly what we are working on right now, as we are exploring the potential for new offerings based on these results. 

TEAM

James Siddle, Alan Maloney, Adam Day