It’s been fun to observe the centenary of the First World War through twitter and other social media. These forms of communication allow people to experience and share commemorative events in new ways. Attendees at ceremonies can broadcast their thoughts, perspectives, and reactions and people unable to attend the ceremonies in person can use social media to participate from the other side of the world. Tapping into these interactions offers a whole new perspective into the process of commemoration because it provides access to thousands of individual reflections and thousands of responses to those reflections.
I’ve been musing about the potential of tracking hashtags for a while. Thanks to the generous terms of my postdoc, I have a decent amount of time to play around with this data. So about a week ago I started collecting tweets that mention #Somme100 and #Verdun2016. The first thing I wanted to do was try mapping them so I created a map for each hashtag on Google My Maps.
What do the maps show?
Twitter’s API allows people to collect tweets based on a number of parameters. It’s possible to collect tweets generated by a certain account or tweets that include a particular hashtag. I collected tweets based on the latter, so this corpus does not represent all tweets about commemorations of the Somme or Verdun, only those that used the hashtags #Somme100 and #Verdun2016.
Twitter’s API has made some changes recently, one of which is that it is no longer possible to collect tweets older than nine days. I didn’t get the idea to start collecting these tweets until the first week of March, so this corpus only goes back to 24 February 2016. Unfortunately, that omits a lot of tweets that were generated during the Verdun commemorations of 21 February. I’ll have to plan further ahead next time I try this.
The Twitter API allows users to collect a lot of information about a tweet, such as the username of the person who generated that tweet and the location associated with their account. The points on these maps reflect the location listed on a Twitter user’s profile, which is not necessarily the location from which the tweet was sent. Not all Twitter users list a location on their profile, so tweets generated by such an account are not mapped.
What do the maps hide?
These maps were generated rather hastily on Google’s My Maps, a very user-friendly mapping application. My Maps also integrates nicely into hosting providers such as WordPress, which makes it very easy to embed the maps into my blog. The most serious limitation of Google My Maps is that it can only show one point for every set of coordinates. This means that all of the tweets generated by the same account, or by all the accounts listed in the same place, appear as one single dot on the map. Dozens tweets originate from London, for example, by My Maps will only display one dot and one tweet. This is why you might not be able to find your own tweet on these maps. I’m endlessly annoyed by this limitation but My Maps is the easiest way to make and share an interactive map and Google provides this service free of charge, so I shouldn’t complain.
Zoom, click, explore!
Where to go from here?
There are many fields of metadata attached to these tweets that can be analyzed beyond simply mapping tweets. But the mapping can definitely be improved upon. Other GIS software is better suited to visualizing factors like the frequency of tweets generated in a given location or the magnitude of its audience, but I’m not yet familiar with ways of making those maps kinds interactive and accessible online.
One of the pieces of data that has been collected is the number of followers a tweet was viewed by and the number of times a tweet has been re-tweeted. This makes it relatively easy to find the most visible tweets in the corpus. With a bit of cleaning and re-structuring, these tweets can be rearranged into a relational database to show how tweets are tweeted and re-tweeted within a network followers.
We can already see from these maps that tweets reach across the globe. The next step is to start connecting the dots.