Los Angeles, USA, 9-13 April, 2013
By Kristy Revell SSI Agent and PhD student at University College London
Attending sessions beyond my academic domain, notably in relation to GIS
Observing how social media is influencing research in the digital humanities and social sciences
The Association of American Geographers Annual Meeting is a world-renowned conference that boasts over 6,000 presentations, posters, workshops and field trips. It endeavours to cover a broad range of academic disciplines with geography, sustainability and GIScience being the most represented fields.
This was my first time attending this meeting but luckily I was accompanied by some seasoned AAGers that made the whole experience much smoother. The conference was incredibly enjoyable with one of its most attractive features being the breadth of the high-quality presentations. I was pleasantly surprised to learn that there were a huge amount of sessions within my academic field of environmental sustainability (at least three
interesting and relevant sessions in each time slot). I was also pleased to see that there were many sessions on the intersection between GIScience and sustainability and of course on wider GIS issues, for example big data. As a result, during AAG I took the opportunity to attend presentations beyond my immediate area of expertise. Recently I have been working with colleagues to utilise GPS location data within my own research (to obtain information on travel patterns), as a result I decided to attend a number of sessions on GIS.
Of these GIS sessions, the most interesting presentation that I heard was in a session that focused on researchers’ use of ‘big data’, notably from social media sources such as Foursquare, Facebook and Twitter. This presentation was in response to the recent proliferation of geographical research being undertaken using such big data and was incredibly interesting because it focused on the criticisms of big data, from a social sciences and digital humanities perspective – a topic that is little addressed. The presentation drew on the popular article by Chris Anderson in Wired Magazine titled ‘The End of Theory’, where Anderson states that ‘Petabytes allow us to say: "Correlation is enough." We can stop looking for models. We can analyse the data without hypotheses about what it might show. We can throw the numbers into the biggest computing clusters the world has ever seen and let statistical algorithms find patterns where science cannot’. This
presentation challenged this with criticism.
In this presentation, the main criticism of big data social media research was that drawing conclusions from big data without being an expert in that particular field leads to the drawing conclusions without deep
contextualisation and analysis. Therefore, although the conclusions are rooted in the data, such research can fail to answer the important questions. For example, conclusions may be drawn by an expert in GIS about a city’s inhabitants, their demographics, migration patterns or language but this will fail to tell us why such patterns are being observed because such researchers are not demographers, linguists or migration experts.
Overall I felt this raised an interesting issue for the SSI about wider software literacy. It seems that many of the researchers processing these big data sets are more often than not from a computer science background. They have the expertise to process these data sets but often lack the background to interpret them fully, after all no one can be an expert in everything. This could be mitigated by working in partnership with social scientists but I believe that a lack of software literacy amongst social scientists would act as a barrier to a true contextualisation of the research problems.
Therefore, given the relentless growth in big data is appears that it may well be time for social scientists to learn the underpinning concepts of programming and modelling, if the opportunities that big data offers are
truly to be taken advantage of.
I left this presentation not really knowing what the solution is to these big data issues but I did think that it raised some very interesting questions about big data research. Big data research invests meaning in data that was never created for that purpose and although this data is powerful it isn’t like small data that is collected from a stratified sample to specifically to test a hypothesis. Therefore I believe that as researchers we should broach such research with care and potentially more criticism. One thing is for certain though, big data calls for greater software literacy amongst all researchers and it will certainly require far more inter-disciplinary.