Page 427 -

418 CHAPTER 14 Online and ubiquitous HCI research

14.2.3.2 Following trends: Social media and online interaction data
What can we learn from behavior on the online sites that seem to occupy so much of
our collective attention? Moving beyond the closed confines of online communities,
broader studies of both content and patterns of online activity can tell us a great deal
about how people interact, how ideas spread, and what meaning might be attributed
to those patterns.
Studies of online activity can be classified into three broad categories, dis-
tinguished by data source. Social media studies explore participation in familiar
sites such as Twitter and Facebook to understand how these tools can be used to
find and share information. In this context, we use the term “social media” to
refer to general-purpose sites supporting individually selected lists of “friends”
or “contacts,” as opposed to interest-specific communities described in Section
14.2.3.1. Examples include studies of how people use social media to meet in-
formation needs (Menefee et al., 2016), and examinations of the impact of so-
cial media on dissemination of information from research conferences (Winandy
et al., 2016). Web search studies examine queries submitted to general Internet
search engines, looking for behaviors common to many web users, such as search-
ing for information about flu outbreaks (Ginsberg et al., 2009) and other health
conditions (White et al., 2013; Paparrizos et al., 2016). Examinations of blogs,
wikis, and other user-generated content explore how users interact in creating
and sharing information on the web, including video blogs (Huh et al., 2014),
Wikipedia editing (Viégas et al., 2004, 2007a,b; Kittur and Kraut, 2008), and on-
line reviews (Hedegaard and Simonsen, 2013, 2014), to name a few. Boundaries
between these categories are fuzzy, and many of these goals can be met by mul-
tiple sources of interaction data.
Identification of appropriate data sources, and of the means of accessing that
data, is often the first step in conducting studies of online interactions. Designing a
study to investigate the use of “social media” in examining a topic of interest is a rea-
sonable start, but details are important—which social media sites will you consider?
Which content types? Various sources will differ significantly in their willingness
to share data and in the tools available to access any data that is openly available.
Open-source sites like Wikipedia might allow access to data that might be considered
proprietary by for-profit search engines. Some social media sites such as Facebook
(https://developers.facebook.com/docs/graph-api) and Twitter (https://dev.twitter.
com/overview/documentation) sites provide API access suitable for querying data
sets, while others may require the use of more manual tools to “screen-scrape” data
off of web pages. However, the mere presence of an API might not be sufficient—
APIs that limit the quantity or range of content that can be retrieved might not be
sufficient for some tasks.
An examination of selected papers provides a sampling of some of the approaches
researchers have used to access social media interaction data. Small-scale studies—
such as examining the impact and diffusion of social media content for a specific
issue among a small community—can be conducted relatively easily. Organizers of
a 2011 health research conference established social media presences on Facebook,