A Survey of Approaches to Social Network Analysis

The final session at AoIR 2012 this evening is a panel on online social network analysis. Jakob Linaa Jensen starts us off by reflecting on the methods for studying online social networks, and notes the importance of both tracking social media use in practice and asking users about their uses. Jakob also draws a distinction between social media and social networks, where social media are spaces where users can create profiles and share content, and social networks are a smaller subset which is focussed centrally on the user and their networking activities, less on content.

Online social networks are also different from online communities in that in online communities the entire amount of community interactions are visible, while in networks users can generally only see their interactions with their established friends (even if, as on Twitter, wider information gathering is possible through APIs).

Standard analyses for social networks include usability analyses, Website analyses, and Web sphere analyses, and new methods for studying social network sites and platforms are continuously being developed. Most broadly, we can focus on the users, content, relations, context (design, policies, algorithms), or aesthetics of social network sites – and we can take subjective (surveys, focus groups), technical (traffic analyses, network visualisations), and observational approaches (aesthetic analyses of profiles, content analyses of profiles and dialogues).

Jakob conducted a general survey of the public sphere (media use, participation, citizenship) in Denmark, and part of this focussed on the uses of social media; this involved a quantitative survey, focus groups, and examinations of user profiles and content. This provided a useful overview of respondents' general media use, of their specific knowledge of and attitudes towards social media, and of their actual behaviours and practices as opposed to the norms and ideal conceptions they articulated.

Such triangulation can generate richer data than any of these methods alone, and illustrates discrepancies between the different elements; what still needs to be added to this are the meta-analyses of content.