Mapping Online Publics: Methodological Observations

Bremen.
The next speaker at ‘Doing Global Media Studies’ is my CCI colleague Jean Burgess, presenting on our Mapping Online Publics research project; this presentation is the methodological part, and I’ll show some more results at the main ECREA 2010 conference later in the week. Our research is part of an ARC Discovery project exploring methods for examining Australian social media use – the aim is to develop methods for computer-assisted cultural analysis. Over the course of the three years, we’ll examine blogs, Twitter, Flickr, and YouTube.

Here’s Jean’s Powerpoint, and my transcript is below, too. I’ll add the audio later.

The main intent here is to advance beyond both mapping only the political blogosphere (which ignores a great deal of other online social interaction) as well as beyond time-based snapshots; in the process, we’ll also need to address problems of scale, of course, as we’re now faced with a great deal more data.

The underlying theoretical base is that the overall media ecology is changing, especially through the rise of user-created content; this means that the dynamics of public communication are also changing – there is no longer a unified public sphere, but rather a range of more or less ephemeral networked publics. To understand this requires some very fundamental empirical work: what are the baseline levels of user activity, what patterns do they follow, what clusters and communities of users exist, and how does all of this change over time?

This requires very large-scale data gathering, even for a project limited only to Australian online activities, and the development of computer-assisted techniques for analysing these data. How do we activate key events, clusters, and communities; how do we trace these through time? Further, what are the cultural implications of the developments we observe? This might be best studied in the context of acute media events.

Tools we use for this include the in-house content crawlers and scrapers developed for us by Sociomantic Labs; these data are processed by helper scripts written in Gawk, and examined through automated textual analysis tools like Leximancer and WordStat and visualised using the open-source network visualisation tool Gephi.

What we’re able to do with this is to track, for blogs, patterns of activity over time, networks of hyperlink interlinkage, and thematic clusters in the text – both in the longer term as well as around specific short-term events. For Twitter, we can similarly observe patterns over time, networks of @replies between individual users, and key themes of the tweets.

For Twitter, we’ve examined tweets about the Masterchef reality TV programme, for example, and the network map of @replies clearly shows that @MattsCravat, the account of one of the judges, receives a very large number of tweets, but sends very few, while other users send many, but don’t necessarily receive many replies from the network. We’ve also examined the most-tweeted hyperlinks to YouTube videos during the recent Australian election, as well as thematic patterns in tweets during the election campaign. (You can see detailed graphs on this on our project blog.)

So, on the one hand, there is a re-emergence of national frames in Internet research and practice – YouTube, for example, offers a localised homepage depending on the user’s location, and different platforms have differing importance in different nations. At the same time, issue publics can be both culturally bound and transcultural. To understand this, we need to develop further cross-national approaches.