Towards more systematic Twitter analysis: metrics for tweeting activities

Abstract

Twitter is an important and influential social media platform, but much research into its uses remains centred around isolated cases – e.g. of events in political communication, crisis communication, or popular culture, often coordinated by shared hashtags (brief keywords, prefixed with the symbol ‘#’). In particular, a lack of standard metrics for comparing communicative patterns across cases prevents researchers from developing a more comprehensive perspective on the diverse, sometimes crucial roles which hashtags play in Twitter-based communication. We address this problem by outlining a catalogue of widely applicable, standardised metrics for analysing Twitter-based communication, with particular focus on hashtagged exchanges. We also point to potential uses for such metrics, presenting an indication of what broader comparisons of diverse cases can achieve.

Keywords:

• Twitter
• social media
• hashtags
• metrics
• communicative exchanges

View correction statement:

Erratum

Notes

This article was originally published with errors. This version has been corrected. Please see Erratum (http://dx.doi.org/10.1080/13645579.2013.770300).

1. Some very early Twitter research projects continue to enjoy free large-scale data access under a grandfather clause (see e.g. Jürgens, Jungherr, & Schoen, 2011) – but Twitter has stopped granting new fee waivers.

2. Our thanks to yourTwapperkeeper developer John O’Brien III for pointing out this problem; see http://code.google.com/p/twitter-api/issues/detail?id=214 for details.

3. News coverage of the 2011 UK riots suggested that Blackberry messaging was used to incite the riots, for example (see Halliday, 2011); here, it may be useful to examine whether Blackberry users showed divergent trends in their participation in the #londonriots or #ukriots hashtags.

4. This is documented in detail at http://mappingonlinepublics.net/; also see Bruns & Liang (2012); Bruns (2011).

5. A Gawk script for generating these metrics from yourTwapperkeeper data-sets is available at http://mappingonlinepublics.net/2012/01/31/more-twitter-metrics-metrify-revisited/

6. For want of a better term, we use “community” loosely here: overall, hashtag participants may act more or less strongly as a genuine community. Indeed, our metrics provide useful measures for the extent to which they do so.

7. Various equivalent conventions for marking messages as retweets now exist, and should be tested for in this context: in addition to RT @user [message], formats such as MT @user [message] (for ‘manual tweet’), [message] (via @user), and ‘‘@user [message]” (username and original message enclosed in quotation marks) are also common, with or without added comments from the retweeting user. At present, a pattern matching expression which searches for RT @user, MT @user, via @user, and “@user will identify the vast majority of retweets; as Twitter formatting conventions evolve, additional variants may need to be added, however.

8. Depending on context, it may be necessary to remove outliers from this calculation. This may include users who tweeted once but received substantial retweets or @replies, or users who did not participate in the hashtag, but were @mentioned frequently. Conversely, such unusual cases may themselves be worthy of closer investigation.