Analytics of Learning Analytics
Learning analytics as a nasent field of scholarship is evolving rapidly and garnering broad interest in both educational research and practice (Johnson 2011). Since the 2011 Learning Analytics and Knowledge (LAK) conference in Banff, Canada, exciting changes have happened during the past four years, including the launches of the Society of Learning Analytics Research (SoLAR) in 2012, Learning Analytics Summer Institute (LASI) in 2013, and the Journal of Learning Analytics in 2014. These events together indicate the establishment of learning analytics as an independent field of research and practice.
Since learning analytics as a field is still in its early stage, efforts have been made to understand the evolution of the field and its linkages with others. For instance, researchers have attempted to understand the similarities and distinctions between Learning Analytics and Educational Data Mining, which as two research communities share similar interests but grow seperately in their early years (Siemens 2012). Efforts have also been made to study the roots of learning analytics and its relations with fields including learning sciences, machine learning, and data-driven analytics (Ferguson 2012, Balacheff 2013). More recently, colleagues have made use of an open dataset—which contains structured metadata from research publications in the field of learning analytics and educational data mining (Taibi 2013, Proceedings of the LA...)1—and developed topic models (Sharkey 2014), ontology (Zouaq 2013), visualizations (Scheffel 2014), and knowledge systems (Lopes 2014, Hu 2014) to make sense of the field. These efforts uncover key themes of the field and identify major challenges faced by the community (Siemens 2012a).
The present study attempts to contribute to the ongoing reflection upon learning analytics by analyzing Twitter archives of the past four LAK conferences from 2011 to 2014. Using the set of tweets posted by the conference participants, who attended LAK either in person or remotely, we are hoping to uncover new insights about the evolution of the field. The significance of this work is two-fold. First, because learning analytics is a relatively new field attracting participation of both academics and practitioners, many community members have not published in conference proceedings or academic journals or are not intended to publish at all; as a result, analysis of data from academic publications, as those included in the LAK Open Dataset, falls short in revealing the reach and development of the community. In contrast, Twitter as an information sharing and social networking platform broadly used at LAK conferences affords us with authentic, multimodal data from many more “participants” other than those who have published in the learning analytics literature. Second, Twitter supports rich social interactions among conference participants, in the form of retweeting, mentioning, and replying, which are not supported by traditional academic publishing venues. By analyzing social interactions on Twitter, we could obtain a richer picture of community dynamics of learning analytics, by identifying the leaders, characterizing information diffusion patterns, and detecting sub-communities. Therefore, analysis of Twitter archives of LAK conferences could potentially afford us new insights about the learning analytics community.
In this article, we start with a brief introduction of Twitter and Twitter analytics in the context of academic conferences. Then we introduce the LAK Twitter dataset and analytic approaches applied in the study. After that, we present results from our analysis and discuss their implications for the field of learning analytics.
Twitter is, as we write, an online social networking service that enables users to share short messages known as “tweets.” While Twitter is normally conceptualized as a social network, or a microblogging service, it has really grown into an information or news network (Kwak 2010). Because of the agileness offered by its 140-character limit, it has emerged to become “a personal news-wire,” in Twitter’s own words (Stone 2008), on which all types of world events are posted and further spread through “retweeting.” These user behaviours together give rise to trending topics and facilitate broad-scale social phenomena, such as Haiti eartiquake relief efforts (Smith 2010), “Arab Spring” (Lotan 2011) and Iran Election (Gaffney 2010). It has contributed to transform journalism, by either changing how people become aware of news (Hermida 2010) or how journalists engage with their profession (Lasorsa 2012).
Twitter is also widely used at conferences, from its impressive “debut” at the 2007 South by Southwest Interactive (SXSWi) conference to almost every academic conference the authors have attended. At conferences, Twitter has been used to establish a backchannel to enable richer communication among attendees and extend conversations beyond the conference venue (Atkinson 2009). In a conventional academic conference setting, the space is divided into a “front” stage for the speaker and a large “back” area for the audience (Ross 2010). In this context, attention is solely focused to the front and interactions are usually limited. Because of the constrains posed by time and space, opportunities provided for the audience to interact with each other and to collectively construct understanding of a given speech are usually rare. As a result, the traditional conference model has a number of problems, including feedback lag, stress for asking questions, and participation decrease caused by the “single speaker paradigm” (Anderson 2008, Ebner 2009). Twitter could help close the gap of social interactions at conferences, thanks to its simplcity and the ubiquity of Internect connection. Conversationality and collaboration afforded by Twitter, especially through
#LAK11 (Honey 2009), could help mitigate the disconnect among conference participants. It thus becomes not surprising that Twitter has become widely used at academic conferences.
Because the extensive use of Twitter in various social sectors, the analysis of Twitter data carries the potential to offer actionable knowledge for stakeholders or to help us understand information diffusion on social media. For instance, sentiment analysis of tweets has also been broadly applied to understand customer perception of certain products (Chamlertwat 2012, Davidov 2010), or to characterise presidential debates by combining tweets with live television programs (Diakopoulos 2010). New information diffusion mechanisms could also be discovered from various angles such as social linkages (Kwak 2010) and retweeting behaviors (Suh 2010). In education, Twitter and other social media platforms are increasingly used in classrooms to facilitate communication between teachers and students, as well as among students (Grosseck 2008, Junco 2013, Junco 2011). Researchers also combine data-driven approaches and ethnographical approaches to study the unique online culture among the digital natives (Ito 2009, Boyd 2014), students identity performance on social media (Junco 2014), and students’ learning issues and learning experiences (Chen 2014).
Twitter’s useage at academic conferences has also attracted some reseach attention, given its special appeal to such social events. Previous studies have mainly focused on three aspects. The first aspect focuses on users and usages of Twitter at academic conferences. For example, some studies seek to understand who use Twitter at conferences, why they use it, and how (Ebner 2009, DeVoe 2010, Ross 2010). Based on a