Big data, little people: The case for connecting data back to the people it is about

Donald Matheson, University of Canterbury, New Zealand

This presentation argues that a key contribution of social scientists – in my case a communication researcher – to the emerging scholarly world of big data studies is to connect data back to the people it is about. There are good ethical and epistemological reasons for attempting to do this and I begin the presentation by discussing them in relation to the key underlying problem of context. Computational research on large datasets, particularly those made up of the digital traces people leave behind, is making significant insights possible into the relationship between structure and agency, the way society is connected and related questions, because of the way it decontextualises and aggregates material. Yet in doing so, particularly in its frequent reliance on probabilistic analytics, it remains poorly equipped to tag, sort or otherwise mark up datasets according to local meaning, function or practice being performed. Ethically, the data researcher risks further marginalising those already marginalised by dominant forms and shapers of knowledge in working at this level of abstraction. Epistemologically, the researcher may over-read what the data provides, including the importance of keywords or topics and networked dimensions of social practice.

The presentation then describes some attempts, in my own research on tweets after a major New Zealand earthquake, to reconnect data with practice. I argue for the value of tracking language use in particular as a form of data that can be read within interpretivist frameworks for traces of human intention and social interaction. I discuss, firstly, the use of a concordancer to cycle between large-scale patterns and the detail of text in co-text and context and, secondly, the value of working within a team where data scientists are able to tailor tools and queries. Thirdly, at a more abstract level, I discuss ways to take social media data back to some of the people whom it is about, so as to draw them into the process of both question-setting and interpretation.


The Shifting Temporalities of Online News: The Guardian’s Website from 1996-2015

Niels Brügger & Henrik Bødker, School of Communication and Culture, Aarhus University

As much as news websites news can be characterised by speed and immediacy they are also recognisable online periodicals, which accumulate preceding news coverage. This is, as with the constitution of time in more general, linked to relations between change and continuity. This article aims to understand how online conventions for the temporalities of online news have developed since the first news sites in the mid-1990s. The analytical starting point for this is that such temporalities must be understood as a complex interplay between textual elements on different and overlapping levels of the webpage.
This article consequently employs a framework for webpage analysis that focuses on the syntactical level where temporalities emerge as relations between textual elements that change at very different intervals. This framework is applied to examples from the different stages of The Guardian’s webpage from 1996 to the present retrieved from The Internet Archive (www.archive.org). The shifting constitutions of time that emerge through these analyses highlight interesting relations between emerging online conventions and journalistic practices. The analysis reveals a historical process through which the temporal affordances of the online are reflexively negotiated in relation to each other and as such highlight interesting relations between emerging online conventions and journalistic practices.
Due to an inbuilt instability between textual elements on stored websites (as well as other characteristics of online archives) the construction of the empirical base stands in a complex relation to the analytical framework applied. As much as the article is a historical analysis of the temporality of online news it thus also offers a range of methodological considerations as well as thoughts on avenues for further study of how journalism constitute time within different institutional settings and on different media platforms.

Time and place:
The seminar takes place 23 May 2016 from 13:00 to 15:30 at room 295 in the Nygaard-building (5335).