Digital learning content is becoming more and more commonplace. However, the creation of this learning content is usually done in specialized publishing platforms, that may lock in the content. This prevents the learning content from being discovered automatically, and hinders its uptake. A more general approach for authoring discoverable learning content is needed. We built a user-friendly authoring environment for creating digital publications, enriched with multimedia and interactivity features. This authoring environment outputs the created content in open formats, independent of any publishing platform. We adjusted our authoring environment to create semantically enriched learning content using the RDFa syntax, without the need for manual intervention. In this paper, we show how well-structured and well-annotated digital learning content can be created and validated using our authoring environment. The result is a discoverable publication in an open format, ready to be distributed as is, or ready to be imported in existing publishing platforms.
Digital learning content is being widely adopted as an addition to print-based
learning content. More recently, this digital content is also being used as a
primary source for online education initiatives such as Massive Open Online
Courses (MOOCs) [
Digital learning content is usually created and published using specialized
platforms. However, these platforms (1) lock the learning content in a custom
data format and (2) o er limited support to make the learning content
machineunderstandable, hampering the interlinking and querying of learning content,
and thus its uptake [
To overcome the content lock-in, we propose an authoring environment that is independent of the publishing platform, and outputs the content in an open format. This way, the authored publication can still be imported in existing learning publication platforms, but the content remains accessible by other consumer services.
To increase the discoverability of the authored publication, learning content needs to be annotated properly to achieve machine-understandability. However, manual annotation is to be avoided, as authors of learning content should not be burdened with this task. Therefore, the aforementioned authoring environment has to be capable of automatically inserting the appropriate metadata for the given content.
After giving some background information in Section 1, we describe the built authoring environment to tackle the aforementioned limitations in Section 2. Finally, we conclude in Section 3. 1
The following paragraphs will give an overview of how currently learning content is authored and published, and how it can be made discoverable. 1.1
Learning content is usually authored in a Learning Content Management System
(LCMS) [
SCORM is an e-learning speci c solution to package and transfer learning content between Learning Management Systems (LMSs). Furthermore, the SCORM speci cation also de nes the communication between the client-side application that uses SCORM learning objects and the host system (e.g., to verify the answers to questions inside the learning object). As SCORM is LMS-oriented, the knowledge contained in a SCORM learning object cannot be transferred to systems with a di erent scope. Moreover, since the resulting learning object is a packaged format, the original learning object is not really shared, but duplicated every time it is (re-)used. When the learning content is updated, the updates cannot be disseminated automatically to the users that already have a previous version of that learning content.
An LMS is a platform where teachers can present learning content to their
students [
Although such an integrated approach does improve the ease of use for the teacher (as he or she can seamlessly create and distribute new learning content), it prevents the adoption of this learning content by other publishing environments, which further hinders the discoverability of the created learning content.
Meanwhile, Open Web standards such as HTML are being used as the de facto publication format on the Web to publish any kind of (possibly enriched) content. These HTML publications can be easily distributed and can be made accessible o ine by packaging them as EPUB 3 5, an open packaging format that is capable of packaging any kind of HTML web page as an e-book. As learning content can easily be created in HTML, publishing learning content in HTML and packaging it as EPUB 3 could thus be an open alternative to the current publishing chain of creating learning content in an LCMS, packaging it as SCORM, and distributing those packages among LMSs.
Furthermore, the EPUB 3 publications can be semantically enriched with provenance information6. This provenance information can be used to compare versions of the same resource, and di erent publications of the same learning content can be compared and discovered using the included provenance information. This allows for the automatic dissemination of updates of the original learning content to already distributed publications. 1.2
To make digital publications discoverable, machine-understandable metadata
is needed [
We can conclude that EPUB 3 can be used as a discoverable open format for publishing learning content. In fact, there are recent e orts by the World Wide Web Consortium (W3C) and the International Digital Publishing Forum (IDPF) on a speci cation for using EPUB 3 in an educational environment, called EDUPUB 9. 5 http://idpf.org/epub/301 6 http://www.w3.org/TR/prov-o/ 7 http://www.lrmi.net/the-specification 8 http://schema.org/ 9 http://www.idpf.org/epub/profiles/edu/spec/
We developed an authoring environment for creating discoverable digital learning
content, making use of Open Web standards (HTML5, EPUB 3) [
To improve discoverability and overall usage, the resulting learning content is automatically semantically annotated in the form of RDFa (Lite). Indeed, these annotations facilitate machine understandability, which makes it easier to query and (automatically) interlink learning content. Furthermore, the authoring environment natively handles all multimedia and interactivity features currently available on the Web. This means that adding audio, video, and interactivity widgets to learning content is as easy as adding textual content. The result is a user-friendly authoring environment that makes it straightforward to create wellformed learning content that can be published (possibly to an existing L/CMS), discovered, and interlinked.
The authoring environment10 is a Web Application that can export wellannotated HTML5 or EPUB 3 publications with advanced multimedia and interactivity features. When authoring new learning content, the user can manually choose between a number of pre-de ned, well-annotated snippets that can be inserted into the learning content and arranged by simple drag-and-drop actions. These snippets follow (among others) the EDUPUB speci cation, and are annotated with Schema.org terms in RDFa. Di erent snippets, that follow di erent speci cations, can be created as well.
The plug-in structure of the editor assures that custom actions can be performed for certain types of snippets. For example, when inserting the intended age for a particular exercise, the user gets a dropdown menu of available age ranges (Figure 1), instead of having to manually type the age range (and accompanying semantic annotation) in the learning content. This improves userfriendliness while also streamlining the work ow for authoring digital learning content. During and after authoring, the learning content can be validated using an XML Schema De nition (XSD) document to verify whether the content has been structured properly.
When reviewing the machine-understandability of the learning content by making use of an RDFa inspector11, we can verify that the learning content is well-structured and that it contains multiple semantic annotations, without having the user to edit any HTML5 code. As a result, the machine-understandable learning content can be used as a base for interlinking. For instance, by querying the Web for all exercises with the keyword Algebra, and with a typical age range of 11-14, a user can retrieve all Middle School algebra exercises. 10 Showcase: http://users.ugent.be/~bjdmeest/lile_demo.mp4 11 E.g., http://rdfa.info/play/ By building an authoring environment that can automatically add semantic annotations to digital learning content, we remove the cumbersome burden of having to manually insert annotations in digital learning publications. The resulting output consists of well-structured, well-annotated, and validated learning content that can be published to multiple platforms. As such, our authoring environment makes it easier for a general audience to create well-de ned learning content, thus contributing to an improved production and uptake of machineunderstandable learning content.
For now, the de nition of the snippets needs to be done manually by an expert before well-annotated content can be created. For future work, it could be possible that, based on the XSD document or another type of speci cation document, the snippets can be created automatically. This would also broaden the use of the authoring environment, as any type of publication (e.g., medical or legal) could be created, as the snippets do not need to be created manually for every di erent use case.
Acknowledgements The research activities described in this paper were funded by Ghent University, iMinds, the IWT Flanders, the FWO-Flanders, and the European Union, in the context of the project \Uitgeverij van de Toekomst" (Publisher of the Future).