In order to conduct my analysis, I used the Westlaw database to source approximately 1500 articles from three separate time periods (500 articles for each date).  The articles were grouped as follows:

• 2009 - articles published in the year Bergdahl was captured in Afghanistan.

• 2014 - articles published in the year Bergdahl was released by the Taliban.

• December 2015 to present - articles published since the popular podcast, Serial, began their season on Bergdahl and Trump's Presidential campaign referenced the soldier.

I chose to take samples from these years to compare the type of coverage given to Bergdahl, across three major points in his timeline.

Once collected, I began cleaning the data as much as possible (the Westlaw format involved utilising the 'find and replace' function in Microsoft Word), and I then reformatted the three text documents and began to input the data into Leximancer.

Initially, I ran the data through Leximancer on its default settings to get an idea of what concepts could confuse the output and visualization.  This step allowed me to identify areas of the data that required further cleaning, and provided me with a basic outline of the information.  I then decided which parameters I would use to generate a visualisation for analysis.

Once satisfied with my choices, I input my data into Leximancer for a second time.  Using Project Control, I removed themes that I deemed irrelevant (e.g. 'spoke', 'today', and 'copyright') and merged a few like-terms (e.g. multiple versions of Bergdahl's name) in order to make my list of concepts more succinct.

I enabled the sentiment lense under 'user defined concepts', added file tags, and configured my insights dashboard.

After processing the data, I generated an insights PDF, a visualisation, a concept map, and began to analyse the information.