"Introduction. The concept of Learning Analytics is attracting significant attention within several communities with interests at the intersection of learning and information technology, including educational administrators, enterprise computing services, educators and learners. The core proposition is that, as unprecedented amounts of digital data about learners’ activities and interests become available, there is significant potential to make better use of this data to improve learning outcomes. After introducing some of the conceptual roots of Learning Analytics (§2), we propose that the implementation of effective Social Learning Analytics is a distinctive part of this broader design space, and offers a grand challenge for technology-enhanced learning research and enterprise, in three important respects (§3). 1. The first is that the educational landscape is extraordinarily turbulent at present, in no small part due to technological drivers. The move to a participatory online culture sets a new context for thinking about analytics. Online social learning is emerging as a significant phenomenon for a variety of reasons, which we review (§4) in order to clarify the concept of online social learning (§5) and ways of conceiving social learning environments as distinct from other social platforms. 2. The second challenge is to understand the possibilities offered by different types of Social Learning Analytic, both those that are either inherently social (§6) and those that can be socialised, i.e., usefully applied in social settings (§7). 3. Thirdly, we face the challenge of implementing analytics that satisfy concerns about the limitations and abuses of analytics (§8). We conclude (§9) by considering potential futures for Social Learning Analytics, if the drivers and trends reviewed continue. Learning analytics. Learning analytics has its roots in two computing endeavours not specifically concerned with learning, but rather with strong business imperatives to understand internal organisational data, and external consumer behaviour. - Business Intelligence focuses on computational tools to improve organisational decision-making through effective fusion of data collected via diverse systems. The earliest mention of the term ‘learning analytics’ that we have found relates to business intelligence about e-learning products and services (Mitchell & Costello, 2000). - Data Mining, also called Knowledge Discovery in Databases (KDD), is the field concerned with employing large amounts of data to support the discovery of novel and potentially useful information (Piatetsky-Shapiro, 1995). This field brings together many strands of research in computing, including artificial neural networks, Bayesian learning, decision tree construction, instance-based learning, logic programming, rule induction and statistical algorithms (Romero & Ventura, 2007). From data mining developed the field of: - Educational Data Mining (EDM) “an emerging discipline, concerned with developing methods for exploring the unique types of data that come from educational settings, and using those methods to better understand students, and the settings which they learn in” (Baker & Yacef, 2009). Originally, relatively fine-grained, quantitative data came from private educational software applications—Romero and Ventura (2007) trace the first EDM publications to 1995—but their overview of the field shows that research projects multiplied after widespread adoption of virtual learning environments (VLEs) in the early 21st century. Blackboard and Moodle are well-known examples of VLEs, which are also known as learning management systems (LMSs) and content management systems (CMSs). These tools automatically amass large amounts of log data relating to student activities. They not only record student activities and browse time, but also personal information such as user profiles, academic results, and interaction data. Many of them include student tracking capabilities as generic software features. Dawson (2009) reported that the depth of extraction and aggregation, reporting and visualisation functionality of these built-in analytics was often basic or non-existent, but in the last year, all of the major VLE products now include at least rudimentary analytics “dashboards.” Educational institutions have become increasingly interested in analysing the available datasets in order to support retention of students and to improve student results. This use of academic analytics stretches back for at least 50 years, but has become more significant in the last five years as datasets have grown larger and more easily available for analysis. - Academic Analytics are described by Campbell & Oblinger (2007) as ‘an engine to make decisions or guide actions. That engine consists of five steps: capture, report, predict, act, and refine.’ They note that ‘administrative units, such as admissions and fund raising, remain the most common users of analytics in higher education today.’ - Action Analytics is a related term, proposed by Norris, Baer and Offerman (2009) to emphasise the need for benchmarking both within and across institutions, with particular emphasis on the development of practices that make them effective. The Signals project at Purdue University is currently the field’s flagship example of the successful application of academic analytics, reporting significantly higher grades and retention rates than were observed in control groups (Arnold, 2010; Pistilli & Arnold, 2012). The project mines data from a VLE, and combines this with predictive modelling to provide a real-time red/amber/green traffic-light to students and educators, helping staff intervene in a timely manner where it will be most beneficial, and giving students a sense of their progress. Encouraged by such examples, educational institutions are seeking both to embed academic/action analytics and to develop a culture that values the insights that analytics provide for organisational strategic planning and improved learner outcomes. A growing number of universities are implementing data warehouse infrastructures in readiness for a future in which they see analytics as a key strategic asset (Stiles, Jones, & Paradkar, 2011). These data warehouses store and integrate data from one or more systems, allowing complex queries and analysis to take place without disrupting or slowing production systems. This brings us to the present situation; the first significant academic gathering of the learning analytics community was in 2011 at the 1st International Conference on Learning Analytics & Knowledge, doubling in size to 200 in 2012. The 2011 conference defined the term as follows: Learning analytics is the measurement, collection, analysis and reporting of data about learners and their contexts, for purposes of understanding and optimising learning and the environments in which it occurs. Clearly, this encapsulates strands from all the above fields, reflecting the topic’s interdisciplinary convergence but, in contrast to more theoretical research or artificial experimentation which might be published in some of the above fields, there is an emphasis on impacting authentic learning from real-world contexts, through the use of practical tools. There is also a shift away from an institutional perspective towards a focus on the concerns of learners and teachers. The main beneficiaries are no longer considered to be administrators, funders, marketing departments and education authorities, but instead are learners, teachers and faculty members (Long & Siemens, 2011). The challenge of social learning analytics. In a literature analysis of the field, we found that in the discourse of academic analytics there is little mention of pedagogy, theory, learning or teaching (Ferguson, 2012). This reflects the roots of these analytics in management information systems and business intelligence, whose mission has been to guide strategic action by senior leaders in organisations, and whose tools deliver abstracted summaries of key performance indicators. In such contexts, senior executives do not have the time to delve into the process details of a particular individual’s or group’s interactions, and similarly, the arguments for academic analytics seem to focus on finding variables that predict positive or negative outcomes for cohorts of learners. Performance indicators in educational settings typically involve outcomes-centric analytics based on learners’ performance on predefined tasks. Within formal education, success is typically defined as the display of expertise through summative assessment tasks (for example, assignments, exams or quizzes) intended to gauge mastery of discipline knowledge. The focus is on individual performance and on what has been achieved. This model is familiar within settings such as schools and universities, but it is less relevant in the context of online social learning, which involves lifelong learners drawing together resources and connections from across the Internet to solve real-life problems, often without access to the support of a skilled teacher or accredited learning institution. Social Learning Analytics (SLA) are strongly grounded in learning theory and focus attention on elements of learning that are relevant when learning in a participatory online culture. They shift attention away from summative assessment of individuals’ past performance in order to render visible, and in some cases potentially actionable, behaviours and patterns in the learning environment that signify effective process. In particular, the focus of social learning analytics is on processes in which learners are not solitary, and are not necessarily doing work to be marked, but are engaged in social activity, either interacting directly with others (for example, messaging, friending or following), or using platforms in which their activity traces will be experienced by others (for example, publishing, searching, tagging or rating). Social Learning Analytics is, we propose, a distinctive subset of learning analytics that draws on the substantial body of work demonstrating that new skills and ideas are not solely individual achievements, but are developed, carried forward, and passed on through interaction and collaboration. A socio-cultural strand of educational research demonstrates that language is one of the primary tools through which learners construct meaning. Its use is influenced by their aims, feelings and relationships, all of which shift according to context (Wells & Claxton, 2002). Another socio-cultural strand of research emphasises that learning cannot be understood by focusing solely on the cognition, development or behaviour of individual learners; neither can it be understood without reference to its situated nature (Gee, 1997; Wertsch, 1991). As groups engage in joint activities, their success is related to a combination of individual knowledge and skills, environment, use of tools, and ability to work together. Understanding learning in these settings requires us to pay attention to group processes of knowledge construction – how sets of people learn together using tools in different settings. The focus must be not only on learners, but also on their tools and contexts. Viewing learning analytics from a social perspective highlights types of analytic that can be employed to make sense of learner activity in a social setting. This gives us a new way to conceive of both current and emerging approaches—as tools to identify social behaviours and as patterns that signify effective process in learning environments. Social Learning Analytics should render learning processes visible and actionable at different scales: from national and international networks to small groups and individual learners. We turn now to review some of the features of the participatory online culture that drives this work. The emergence of open, social learning. In this section, we identify some of the signals that many futures analysts and horizon-scanning reports on learning technology have highlighted as significant. Taken together, these create synergies that establish a radically new context for learning. In such a context, we argue, analytics focused on summative assessment of performance remain important but do not go far enough: we need to develop new sets of analytics that can be used to support learning and teaching in these new conditions. We summarise these phenomena as: - technological drivers - the shift to ‘free’ and ‘open’ - demand for knowledge-age skills - innovation requires social learning - challenges to educational institutions. Technological drivers. A key force shaping the emerging landscape is clearly the digital revolution. Only very recently do we have almost ubiquitous Internet access in wealthy countries and mobile access in many more. In addition, we now have user interfaces that have evolved through intensive use, digital familiarity from an early age, standards enabling interoperability and commerce across diverse platforms, and scalable computing architectures capable of servicing billions of real-time users and of mining the resulting data. With the rise of social websites serving millions of users, such as Facebook, YouTube and Twitter, plus the thousands of smaller versions and niche applications for specific tasks and communities, we have witnessed a revolution in the ways in which people think about online interaction and publishing. Such social media platforms facilitate the publishing, indexing and tracking of user-generated media, provide simple-to-learn collaboration spaces, and enable social networking functions that are becoming ubiquitous: friending, following, messaging and status updates. Standards such as really simple syndication (RSS) allow information to be shared easily using structured data feeds, web services enable more sophisticated machine-machine interaction, and mobile devices expand the availability and localization of these services. Internet services may also begin to apply pressure to one of the slowest evolving elements in educational provision: accreditation. Christensen et al. (2008) argue that the agencies controlling accreditation often stifle innovation and protect the status quo, because new approaches to learning/accreditation struggle to gain credibility unless they are associated with institutions that have the power to award established qualifications. However, as the infrastructure for secure identity management matures, and as the participatory, social culture fostered by Web 2.0 becomes more deeply ingrained in younger generations, initiatives such as OpenBadges may provide new ways to accredit learning outside established institutions. Moreover, as ubiquitous tools for capturing digital material make it easier to evidence learning and practical knowledge in authentic communities of practice, an e-portfolio of evidence might come to have equivalent or greater credibility than formal certificates. However, changes in technology do not necessarily imply changes in pedagogy. Those who view education as information transfer will use interactive media for storage, drilling, testing and accessing information; those who seek conceptual change will seek to make use of their interactive qualities (Salomon, 2000). Technological shifts support analytics that draw on sets of big data—but they do not necessitate a shift towards analytics focused on such issues as conceptual change, distributed expertise, collaboration or innovation. So, if we do not accept simplistically that technology alone determines the future, we need to look elsewhere to understand the move towards online social learning and its associated analytics. The shift to free and open. There has been a huge shift in expectations of access to digital content. The Internet makes possible completely new revenue-generation models due to the radically lower transaction costs incurred (compared to bricks and mortar businesses with physical products) as one scales to hundreds of thousands of users. Andersen (2009) documents many ways in which online companies are able to offer quality services free of charge, producing an increasing expectation on the part of end-users of huge choice between free tools and sources of content hosted ‘in the cloud’. Within education, the Open Education Resource (OER) movement has been a powerful vehicle for making institutions aware of the value of making high quality learning materials available, not only free of charge, but also in formats that promote remixing, in an effort to reap the benefits seen in the open-source software movement. This has not proven to be a simple matter, but OER has made huge progress, and is gaining visibility at the highest levels of educational policy. This is amplified by efforts to make data open to machine processing as well as human interpretation. This requires not only a shift in mindset by data owners but also the construction of technological infrastructure to make it possible to publish data in useful formats. These efforts can be tracked within communities developing Linked Data and the Semantic Web, and their myriad applications communities, for example, Open Government, Open Mapping, Science 2.0 and Health 2.0. Together, these very rapid shifts contribute to a new cultural context for the provision of learning services, in which the industrial-era value chain, previously delivered by a single institution, is disaggregated into smaller and smaller elements. The provision of content, community, tools and basic analytics may increasingly be expected to come free of charge, while learners may still consider paying for other services such as personalised learning journeys, personal tuition, career guidance, accreditation against formal standards and tailored analytics that support them on a variety of sites, not just within one institution. Demand for knowledge-age skills. Technology is always appropriated to serve what people believe to be their needs and values. Since 1991, we have lived in the “knowledge age”—a period in which knowledge, rather than labour, land or capital, has been the key wealth-generating resource (Savage, 1996). This shift has occurred within a period when constant change in society has been the norm, and it is therefore increasingly difficult to tell which specific knowledge and skills will be required in the future (Lyotard, 1979). These changes have prompted an interest in “knowledge-age skills” that will allow learners to become both confident and competent designers of their own learning goals (Claxton, 2002). Accounts of knowledge-age skills vary, but they can be broadly categorized as relating to learning, management, people, information, research/enquiry, citizenship, values/attributes and preparation for the world of work (Futurelab, 2007). From one viewpoint they are important because employers are looking for “problem-solvers, people who take responsibility and make decisions and are flexible, adaptable and willing to learn new skills” (Educational Subject Center, 2007, p. 5). More broadly, knowledge-age skills are related not just to an economic imperative but to a desire and a right to know, an extension of educational opportunities, and a “responsibility to realise a cosmopolitan understanding of universal rights and acting on that understanding to effect a greater sense of community” (Willinsky, 2005, p111). In both cases, there is a perceived need to move away from a curriculum based on a central canon of information towards learning that develops skills and competencies. This implies a need for ongoing analytics that can support the development of dispositions such as creativity and curiosity, collaboration skills and resilience. Innovation requires social learning. The conditions for online social learning are also related to the pressing need for effective innovation strategy. In an accessible introduction to the literature and business trends, Hagel et al. (2010) argue that social learning is the only way in which organizations can cope in today’s fast-changing world. They invoke the concept of ‘pull’ as an umbrella term to signal some fundamental shifts in the ways in which we catalyse learning and innovation. They highlight quality of interpersonal relationships, tacit knowing, discourse and personal passion as key elements. This is a move away from having information pushed to us during spells of formal education towards a more flexible situation in which we pull resources and information to us as we need them. The move from “push” to “pull” motivates analytics that can be accessed by learners at any point, employed in both informal and formal settings, are sensitive to social relationships, and build transferable learning dispositions and skills. Challenges to educational institutions. Together, these forces create pressures on models of educational provision at all stages of education from childhood into workplace learning. Heppell (2007), amongst many, points to the need for an education system that helps people to help each other, rather than one that delivers learning. The barriers between formal and informal learning, and between online and face-to-face learning are currently being broken down, allowing the development of new models that take into account the range of learners’ experience outside formal study, and the affective elements of learning. An example of this is Gee’s “affinity spaces,” which provide a model for online social learning and were first identified in video gaming environments. Affinity spaces are organized around a passion; within them, knowledge is both distributed and dispersed, they are not age graded, experts work alongside newcomers, learning is proactive but aided as people mentor and are themselves mentored, participants are encouraged to produce as well as to consume, smart tools are available to support learning and everyone, no matter what their level of experience or expertise, remains a learner (Gee, 2004, 2009). Other new models for learning are emerging from a variety of digital sources. Some examples amongst many are the learning affordances of the World of Warcraft online game, with its guilds and carefully planned, collectively executed strategies (Thomas & Brown, 2011), learners beginning to access and create knowledge through persistent avatar identities that can move between different environments (Ferguson, Sheehy, & Clough, 2010), and the development of distributed cognition within virtual worlds (Gillen, Ferguson, Peachey, & Twining, 2012). These models suggest new ways of approaching learning analytics. Gee (2003) showed that well-designed video games incorporate analysis of the development of participants’ relevant knowledge and skills, so that their experience is constantly customized to their current level, effort and growing mastery, they are aware of ongoing achievements, and they are provided with information at the point when it can best be understood and used in practice. Having noted some of the features of the emerging landscape for open, social learning, and the implications of these features for analytics, we now consider some of the key features of social learning, and the nature of online social learning environments. Characterising online social learning. Why has someone sawn down half of the beautiful cedar tree outside my office window? I can’t find this out from a book, and I don’t know anyone with the precise knowledge that I am looking for. It is as I engage in conversations with different people that my understanding of what I see outside my window increases, and I learn more about the tree’s history, health, ecosystem and future possibilities. It is not just the social construction of understanding that is important here, since this is a part of most human interactions. My intention to learn is part of what makes this social learning, as are interactions with others. This is not a one-sided engagement with books or online content—it involves social relationships. As such, it has lots of ‘affective’ aspects: people must be motivated to engage with me and I must have the confidence to ask questions in the first place, as well as some way of assessing the expertise of the people I’m talking to. (Ferguson, 2010) Social learning has been conceptualised as societal learning in general, as processes of interaction that lead to concerted action for change, as group learning, and as the learning of individuals within a social context (Blackmore, 2010). Our conception of online social learning takes into account the changing affordances of a world in which social activity increasingly takes place at a distance and in mediated forms. It is succinctly expressed by Seely Brown and Adler (2008) as being “based on the premise that our understanding of content is socially constructed through conversations about that content and through grounded interactions, especially with others, around problems or actions.” Many others have, of course, argued for similar conceptions, unpacking this broad concept in great detail within the constructivist educational literature, and computer-supported collaborative learning (CSCL) research. Social learning adds an important dimension to CSCL, introducing a particular interest in the non-academic contexts in which it may take place (including the home, social network, and workplace) and the use of free, ready-to-hand online tools, with no neatly packaged curriculum or signed-up peer cohort, no formally prescribed way to test one’s understanding and no pre-scheduled activities (Blackmore’s (2010) edited readings remind us how far back everyday, non-digital social learning goes in learning theory, and provide us with foundations for extension into the digital realm). While OERs greatly increase the amount of good quality material available online to learners, another consequence can be that individual learners find themselves adrift in an ocean of information, struggling to solve ill-structured problems, with little clear idea of how to solve them, or how to recognise when they have solved them. At the same time, distributed networks of learners are grappling with ‘wicked problems’ such as climate change, which offer the same challenges on a grander scale. Social learning infrastructure could have a key role to play in these situations, helping learners connect with others who can provide emotional and conceptual support for locating and engaging with resources, just as in our tree story at the start of this section. This forces us to ask whether our current educational and training regimes are fit for purpose in equipping our children, students and workforce with the dispositions and skills needed under conditions of growing uncertainty—a challenge explored in detail by many others, for example in the collection edited by Deakin Crick (2009). The Open University, where we are based, has been seeking to address these issues with its SocialLearn project, aimed at supporting large-scale social learning. In the early days of the project, Weller (2008) identified six broad principles of SocialLearn: Openness, Flexibility, Disruptive, Perpetual beta, Democracy and Pedagogy. Following a series of workshops, Conole (2008) proposed a set of learning principles for the project—thinking & reflection, conversation & interaction, experience & interactivity and evidence & demonstration—and articulated how these could be linked to characteristics of social learning. Distilling this array of perspectives, we have derived a simple working definition focused on three dynamics, which serves to guide us in designing for meaningful interpersonal and conceptual connection: Online social learning can take place when people are able to: - clarify their intention—learning rather than browsing - ground their learning—by defining their question/problem, and experimenting - engage in learning conversations—increasing their understanding. A significant feature of the Web 2.0 paradigm is the degree of personalisation that end-users now expect. However, a me-centred universe has self-evident limitations as a paradigm for holistic development: learning often disorients and reorients one’s personal universe. User-centred is not the same as Learner-centred: what I want is not necessarily what I need, because my grasp of the material, and of myself as a learner, is incomplete. The centrality of good relationships becomes clear when we remind ourselves that a university’s job is to teach people to think, and that deeper learning requires leaving a place of cognitive and emotional safety where assumptions are merely reinforced—see the extensive research on learning dispositions that characterize this readiness (for example, Claxton, 2001; Perkins, Jay, & Tishman, 1993). This implies challenge to stretch learners out of their comfort zones, underlining the importance of affirmation and encouragement that give a learner the security to step out. As Figure 1 shows, the design of a social media space tuned for learning involves many alterations and additions to a generic space for social media. Within an online space tuned for learning, friends can become learning peers and mentors, informal endorsements are developed into verifiable accreditation, information exchanges become learning conversations and, likewise, generic web analytics need to be developed into learning analytics that can be used in such an environment. To summarise: we have outlined what we mean by online social learning, some of the major drivers that help to explain why it is emerging as a phenomenon, and some of the elements that may differentiate a social learning environment from other social media spaces. We have also indicated why these factors require new approaches to learning analytics. Constructivist pedagogies suggest the need for a shift away from a positivist approach to analytics and towards analytics that are concerned with conceptual change, distributed expertise, collaboration and innovation. This ties in with an increasing emphasis on knowledge-age skills and their associations with learning dispositions such as creativity and resilience. Within an open environment, there is a need for a range of analytics that can extend beyond an institutional platform in order to provide support for lifelong learners at all points in their learning journey. These learners may be organised in classes and cohorts, but they may also need analytics that help them to learn together in looser groupings such as communities and networks. These analytics, and their associated recommendations, will be informed by those developed for social media tools and platforms, but they will be tuned for learning, examples being prompting the development of conversations into educational dialogue, recommending resources that challenge learners to leave their comfort zones, or making learners aware that social presence and role are increasingly important to attend to in a complex world. Figure 1. Dimensions of the social learning design space. Together, these motivate a conception of Social Learning Analytics as a distinctive class of analytic. Inherently social learning analytics. Social learning analytics make use of data generated by learners’ online activity in order to identify behaviours and patterns within the learning environment that signify effective process. The intention is to make these visible to learners, to learning groups and to teachers, together with recommendations that spark and support learning. In order to do this, these analytics make use of data generated when learners are socially engaged. This engagement includes both direct interaction—particularly dialogue—and indirect interaction, when learners leave behind ratings, recommendations or other activity traces that can influence the actions of others. Another important source of data consists of users’ responses to these analytics and their associated visualizations and recommendations. We identify two inherently social analytics, and three socialised analytics: Inherently social analytics—only make sense in a collective context: - Social Network Analytics—interpersonal relationships define social platforms and link learners to contacts, resources and ideas. - Discourse Analytics—language is a primary tool for knowledge negotiation and construction. Socialised analytics—although these are relevant as personal analytics, they have important new attributes in a collective co"