Day one of the Business School conference is covered here. This second post focuses on the day’s keynote from Dr Farida Vis (@flygirltwo), who took a very early train up from Manchester to be with us. I’d never met Farida before, but have followed her work for a couple of years, probably through a contact or two on Twitter… such is the way of things these days. Her Twitter page currently directs visitors to this site on Researching Social Media for more information, which is a forerunner to a book on the subject with Mike Thelwall due for early 2013. Farida also made reference to her MA module of the same name at the University of Leicester, which goes by the hashtag #ms7042.
So on to the talk, which was structured around three case studies. I don’t want to repeat Farida’s presentation because I simply wouldn’t do it justice, so here instead are some of the notes I dashed off (in no particular order):
- There’s a need to identify ‘objects’ within social media in order to begin the process of analysing it: links, users, tags, relationships between users, etc.
- It can be very hard to get at stable data: by its nature social media is constantly changing, being added to, deleted, archived away, taken down, etc. This makes it hard to code as well – where do you start, how do you cope with such diversity of material?
- For Farida the key came through speaking to a colleague who had computering coding skills and could provide access to relevant APIs: suddenly lots of data was available. Depending on the site in question, researchers could now get behind the front page and gather user demographics, etc. As a result specific software has been published…
Webometric Analyst, as published by the Statistical Cybermetrics Research Group at University of Wolverhampton.
- During work on the ‘Fitna’ video, focusing on YouTube research, some network analysis was also used: who commented on whose videos; who’s subscribed to whom; who is friends with whom?
- From an ethical perspective, the relevant ethics panel may well approve such work because it’s anonymised and the data is already in the public domain (comments, tweets, etc.); yet researchers may prefer to class this as ‘semi-public’, because some of what’s being accessed is somewhat behind the scenes.
- Data visualisations: to be taken with a pinch of salt. What does the big data graphic hide that only comes through in the little data?
- Sometimes there are limitations to what the web will provide: only 1,000 videos are available for each YouTube query, so some creative work with filters is sometimes needed to get to expand what’s available.
- Performing a relatively straightforward search on a social media site will often bring up unexpected results: alternative perspectives, non-mainstream stories, people using tags simply to attract traffic to unrelated work and so on.
Farida was also heavily involved in ‘Reading the Riots’, which involved a variety of academics and featured heavily in The Guardian. (I also made reference to it on my blog a few months back.) In short, Twitter made 2.6 million tweets available to The Guardian, captured during the riots that swept through some English cities in the summer of 2011; The Guardian announced this but did little with it, then engaged with researchers such as Farida when they started requesting access.
- The research that followed demonstrated that Twitter is a key platform for breaking news, making information available quickly and often well ahead of mainstream media.
- There was of course the accusation that Twitter and social media in general was to blame for the riots, with ensuing demands that it be shut down.
- Other actors, such as the police in certain areas, protested that they were using social media as a means of communicating with the public.
- The Guardian and Twitter essentially curated the data, so its capture wasn’t controlled, planned or managed by the researchers themselves.
- There is therefore a need to recognise that the paper has an editorial line, and that Twitter also has an agenda – they are not in the habit of making this amount of data available.
- Key research questions: which tweets fit together; which are the parent tweets and the children that follow; what was the role of ‘rumour’ in the online conversation?
- Evidence coding: each tweet was coded by human hand three times – this took time.
- The results include being able to visualise how rumours start, are debunked, go quiet and may then re-emerge. What makes a particular account get noticed, who is trusted?
- Other results flag up a small set of ‘elite’ users, followed by a very, very long tail.
I urge you to have a look at more of Farida’s research and to follow her as she develops more work in the area of social media research. It's hardly a surprise that she was very well received by the audience and everyone was happy to see some of their lunch time get eaten into during the question and answer bit. She’s also very happy to talk about her allotment.
Following her talk I grabbed a two minute interview with Farida which you can hear here...
(If there's no Audioboo player displayed please refresh the page or just follow this link.)
Image: ‘Leominster Allotments’ / flickr.com/photos/knucklas09/3622383791/
(This is not, as far as I aware, Farida’s allotment.)