Digital Sociology by Marres Noortje;

Author:Marres, Noortje;
Language: eng
Format: epub
Publisher: Polity Press
Published: 2017-05-08T00:00:00+00:00

Pilot study: the liveliness of climate change on Twitter

Co-occurrence analysis is a basic measure of data analysis, which detects which entities (such as words) occur together in a given text. It has in recent years been implemented in online data tools, such as the aforementioned Mentionmapp, but also in the Twitter Streamgraph. The latter tool uses co-occurrence analysis to analyse what are ‘happening issues’ on Twitter. Enter ‘climate change’ and the Streamgraph will provide a curve of the activity associated with that word on this platform over a specific period of time, visualized as the rise and fall of words based on their frequency of co-occurrence (see Figure 3.4). The Streamgraph then depicts endurance and variation of word relations over time. When more words are significantly connected, the stream widens, and as such, this tool is not dissimilar to a method championed in the sociology of innovation, namely co-word analysis.

This method was developed in the 1980s by the sociologist of science and technology Michel Callon and colleagues to detect emerging or innovative topics in the scientific literature (Callon et al., 1983; see also Danowski, 2009; Marres, 2012a). Co-word analysis draws on co-occurrence, but adds a further operation, as it identifies word pairs based on proximity, assigning a value to word distances of say 3, 4 or 5 words. Co-word analysis weighs these word relations in terms of both the quantity of connections and the proximity of co-occurrence (Callon et al., 1983; Danowski, 2009). The method was advocated by Callon and colleagues as a way of detecting the emergence of topics – or so-called ‘pockets of innovation’ or, as the more sociological terminology had it, ‘areas of problematization’ – in a corpus of texts. Co-word analysis, the argument went, makes it possible to detect changes in thematic associations over time without having to rely on previously defined categories, and their implied criteria of relevance, by rendering text amendable to network analysis.

Steamgraph does something similar to co-word analysis, but there are also relevant differences: the online data tools measure only the frequency of words co-occurring as opposed to the strength of their connection based on their spatial proximity.14 It is thus possible to put popular data tools ‘in their place’, and say that they are less sophisticated than sociological method for the study of innovation dynamics. However, the resonances between sociological methods and digital analytic techniques also opens up a space of exploration for social enquiry, allowing us to interrogate the capacities of social methods anew. In doing so, we should also take into consideration the context of application of these measures: tools like Steamgraph are part of the continuously evolving infrastructure known as the real-time Web which is focused on ‘what is happening right now’. Online data tools are thus closely implicated in the valorization of live content or ‘currency’ (Marres and Weltevrede, 2013): which word appears most often in Tweets? Which news article gets the most links or retweets? Which actor is mentioned most often? By contrast, co-word analysis as

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.