Linked Data for Open and Distance Learning – Part 2
Continued from Linked Data for Open and Distance Learning – Part 1
Linked Data Applications in Open and Distance Learning
The previous section presents a general overview of the principles of linked data, with the idea that it can help sharing and connecting information across the global network of the web. However, to truly understand the benefits of adopting linked data, the most straightforward way is to look at the innovative applications and services that have already been developed using it. We therefore discuss here several illustrative examples, showing the variety of tasks and issues in which linked data can help for open and distance learning, and for education in general. These examples should not however be seen as a definitive answer to the question “what can be done with linked data for open and distance learning?” Indeed, this question is still open, as new ways to exploit the growing information base available as linked data on the web are emerging constantly. Actually, several of the applications described below were submitted to the LinkedUp Challenge12 (see d’Aquin et al., 2014b), which goal is to push forward the use of web data for education.
To illustrate the use of linked data for open and distance learning, we start with one of the simplest and most straightforward example: Supporting the publication and exploration of information about available courses and resources. An example of such a use is the “Study At the OU” mobile application from the Open University. “Study at the OU” is the website of the Open University that contains the description of the courses and qualifications that can be obtained from the University. The corresponding mobile application exposes the course catalogue and additional information about the topics covered by the available courses to smart-phones and tablets (see Figure 4). As part of this application, it is possible to select a topic and obtain information both about the courses available on this topic, and about the related resources such as podcasts, Youtube videos and OpenLearn units. This last feature is implemented using the Linked Data platform of the Open University, querying resources that are directly related to the topic being considered, or for resources attached to courses that are related to this topic. In other terms, it transparently delivers relevant links to heterogeneous resources by integrating these resources under a common Linked Data representation (d’Aquin, 2012b).
The advantage of Linked Data here is that the rich structure of the linked data-based, graph description of the resource metadata can support a range of navigation paths, using information that can spread across multiple systems and sources. Another advantage is that, by abstracting from the specifics of systems and formats in which the resources might be handled, it provides a way to homogeneously integrate heterogeneous resources. While the previous example shows the benefit of linked data in connecting information within an organisation, in this case the Open University, the real power of the approach is when such information, especially about available resources, can be exploited that spread across several organisations and systems. In this spirit, the Solvonauts search engine is an example of the way connecting metadata about resources available in several online systems can help resource discovery, integrating in an homogeneous way results from different origins, and in different formats. However, linked data can also make possible much more sophisticated scenarios. One of them, not unrelated to the applications described above, is resource recommendation. Typically, resource recommendation is based either on the user profile in the system holding the resources (and the profile) or on some form of similarity (of content or usage) between resources of interest to the user. However, this is not an easy task to achieve when trying to recommend resources of a certain type, held in one particular system, from resources of another type, held in another system. Through linked data however, it is possible to process information about these different resources homogeneously, independently of the systems, and to relate them conceptually using other, reference data.
This is the scenario addressed by the DiscOU application (see d’Aquin et al., 2012; and Figure 5). DiscOU is designed to provide recommendation of open educational resources from the Open University, based on a (potential) learner’s interest in a TV or radio program on the BBC. Open educational resources at the Open University include pieces of course material and small articles from the OpenLearn system, as well as multimedia material in the form of audio and video podcasts. These, resources are described through linked data on the Open University platform (data.open.ac.uk), including links to the full content (and transcripts in the case of podcasts). Programs are also described as linked data on the BBC website, which are accessible from the program page, or the iPlayer web system (the video player of the BBC). DiscOU takes the form of a bookmarklet, a link in the bookmark bar of the browser that can overlay functions and interfaces over the currently displayed website. When activated while displaying a BBC program or iPlayer page, the tool extract from the corresponding linked data information about the program, connect this information to concept in DBpedia, and find in the open educational resources of the Open University resources that also relate to these concepts, through pre-established connections between DBpedia and their linked data-based descriptions.
What is interesting here is not only that linked data can help establishing a connection between two completely different information systems, held by two different organisations and with completely different purpose, and use these connections to facilitate access to (open, free) educational resources. It is also that the whole process of recommending resources, is generic and based on the semantic relationships that exist between the resources. It is therefore possible to make explicit these relationships to the user – in DiscOU, in the form of the explanation under each recommendation: “Also about X”. What’s even more interesting is that the user (learner) can even interact with these relationships, tuning the weight of the concepts resources are connected to based on their interest, and obtaining personalised recommendations (see Figure 6).
Another common scenario where the web has introduced massive changes, and where linked data has a strong potential to help for open and distance learning, is collaboration. Indeed, linked data represents a natural model through which different people and organisations can contribute information that connect with each other, therefore creating a global base of information out of these smaller contributions. This is the scenario addressed by the We-Share application (Ruiz-Calleja et al., 2013), for ICT tools in education (see Figure 7). Indeed ,the aim of this application is basically to build a tool repository to help educators select the right tool in the right situation. This might include any type of ICT tool, and the description of each tool includes information not only about the tool and its features, but also about the contributor and the educational scenario in which it has been used. The information is then available for others to reuse, through a web interface and direct access to the created linked data.
The benefit of using linked data in data collection here is not only that the pieces of information that are being produced can be easily put together and re-distributed to others (or linked). It is also that it makes it easier to reuse information that already exist. Indeed, each tool can be described once, and whenever a new educational scenario applies, be simply pointed to (and possibly the description extended), contributing only the part of the information that is specific to the considered used. This is valid internally as well as externally: basic information about the tools are available in other datasets such as DBpedia. Each tool description in the system can not only connect to to existing, internal tool descriptions, but also to these external tool descriptions, enriching the information provided by contributors to the system with general information built for other purposes. Other systems have used a similar approach to collaboration and reuse on the basis of linked data at a much larger scale. These include the GNOSS platform in Spain, that acts as a social network of students and teachers, to exchange, share and describe (educational) resources; or the VIVO platform (Borner et al., 2012) which exploits linked data technologies to collect information within universities, that can then be opened through linked data to others, forming a network of resources from many different universities.
Of course, a way in which linked data can help is also in providing easy access to information for learners about the subject matter of their study. While it is hard to find a generic application of this sort, they are many examples of visualizations and information exploration tools that use web, open data to support learners in a particular domain. For example, Globe Town (Townsend et al., 2013) focuses on sustainable development and uses open data to “explore the intersections, tensions and trade-offs between the ‘three pillars’ of sustainable development: the environment, the economy and society”. PoliMedia (Kleppe et al., 2013) is another example where linked data is used not necessarily in a way that is explicitly related to education, but that presents information about a certain specific topic online, through aggregating and combining information available as linked data, and therefore helps learners in their study of this particular topic. It is designed to facilitate large-scale, cross-media analysis of the coverage of political events, focusing on the meetings of the Dutch parliament. It achieves this by automatically generating links between the transcripts of those meetings, newspaper articles and radio bulletins.
However, the benefits of using linked data in learning analytics are just starting to be considered, mostly within the research community dedicated to learning analytics (see d’Aquin et al, 2014a; d’Aquin and Jay, 2013). Indeed, while linked data simplifies access to information, integration with the tools required for analytics is still preliminary. Interesting examples can be demonstrated, including for example using R to do statistical analysis on linked data sources (see Figure 8a for an example) or, with ad-hoc development of visualisations (see Figure 8b for another example). In many cases however, linked data can be used to enrich the data to analyse or to provide background information for its interpretation (as for example in d’Aquin and Jay, 2013).
A REPORT ON: Linked Data for Open and Distance Learning
Second Edition July 2014
Prepared for the Commonwealth of Learning
by: Mathieu d’Aquin Research Fellow Knowledge Media Institute The Open University, UK