Linked Data for Open and Distance Learning
Mathieu d’Aquin (The Open University, UK) is one of the world’s leading experts in artificial intelligence. He had created a report “Linked Data for Open and Distance Learning” for the Commonwealth of Learning to utilize Semantic Web mechanism to help educators find the needle they want in the haystack of options (d’Aquin, 2014).
Linked data relies on the simple idea that the mechanics used nowadays to share and interlink documents on the web can be applied to share and interlink data and metadata about these documents, as well as the concepts and entities they relate to. On the Web of Linked Data, every “data object” (representing a person, a place, or a topic) is identified by a web address and characterised with web links that can connect to representations of other data objects, identified by other Web addresses, thus using the web as a giant data graph that openly draws from any contributing source. (Fig. 4) The linked data approach enables building of an ecosystem of data where each individual provider contributes to a common, open and global network, rather than limiting themselves to a silo of information. Several Linked Data initiatives in education domains for OER and personalized learning development are underway.
Linked Data, frequently described as “the Semantic Web done right” by the Inventor of the World Wide Web Tim Berners-Lee (2009), has emerged as the de facto standard for sharing semantic data on the web. ADL technical team is currently working hard to add Linked Data/RDF support for anything in xAPI with an Internationalized Resource Identifier (IRI, a web address is the most common one) such as verbs, activity types, attachments, extensions etc.. By leveraging Linked Data as the foundation for xAPI vocabularies, it could fundamentally improve the quality and semantic interoperability of xAPI data by allowing the vocabulary metadata to match schemas and interlink previously unrelated datasets. In other words, controlled vocabularies and ontology metadata for xAPI, if linked, could provide the semantic glue needed to make xAPI data become more expressive and reusable.(Haag, 2015) And, it could possibly open the door to artificial intelligence opportunities for xAPI.
Jason Haag from ADL technical team gave more elaboration to be quoted in this article:
In terms of data interoperability, the core xAPI specification primarily addresses structural interoperability (the ability of two or more applications or agents to exchange information). This focus on structural interoperability was the top priority of the community in order to enable the integration of learning experience data from diverse sources to any application or platform. In addition to structural interoperability, semantic data interoperability is needed to avoid duplication of vocabulary terms and for applications to meaningfully interpret the information being exchanged. In other words, without a semantic vocabulary model the xAPI specification requires more manual work to interpret, organize, aggregate, and generally do useful things with the data.
The ability to link data from diverse sources is a motivator for many projects, as different CoPs seek to take advantage of semantically rich data that was previously spread across disparate sources. By adopting W3C’s RDF standard as the data model for xAPI vocabulary resources and their metadata, the xAPI specification can potentially gain a whole new level of precision for machine readability and semantic interoperability. Upon implementation of this vocabulary specification (and refinement of vocabulary publishing practices) the xAPI community will benefit from exciting new opportunities and capabilities. Ideally, a vocabulary model for xAPI will open up new doors for improved learning analytics, federated search, dynamic look-up of xAPI vocabulary data within authoring applications, improved discovery and reuse of xAPI vocabularies, multilingual translation, and some basic natural language processing capabilities. (Note1)
Scholars doing social network analytics found that putting object nodes together with learner nodes can better reveal how interactions between persons happen, that’s also how Facebook uses Open Graph in the social network. So there are a network of persons and a network of resources, xAPI connects these two networks through the <Verb>, and can help content recommendations for learners. One of the object sociality comes from usage, collective usages can suggest better resources for the same kind of learners automatically, other information in xAPI statements such as <Context> and <Result> data can be added into the recommender algorithm development.
Linked Data is to the xAPI standard age what Learning Object Metadata (LOM) was to the SCORM age, only much better! According to Haag, representing ADL technical team, in near future ADL will provide some examples, practices, tools and requirements in the specification for adopters to follow.
The Importance and Potential of xAPI for OER
OER are widely used by non-profit organizations and initiatives trying to solve issues of education in poor and rural areas where the lack of learning resources and teachers are both serious problems. Some non-profit organizations bring volunteers to rural villages; however, the temporary teachers who barely know the learners won’t ever be a stable solution for the issue. If xAPI is implemented in OER to collect learners’ data no matter if they are in cities or in rural villages, then learner profiles and progresses can be built up in LRSs. XAPI tracking is possible in both online and offline mode; even the learning records of learners without the Internet can be cached and sent back to LRS whenever the devices are connected. The data could be utilized in several ways to help learners.
- Distant tutors can review data and give interventions and instructions to help individual learners according to their strengths and weaknesses. Those distant tutors could be retired teachers in other countries. Since all data will be accumulated, switching tutors will not be a problem.
- Learning paths of high performance learners can be identified and recommended to low performance learners with similar preferences and attributes.
- Through interacting with the same content or related linked content, it’s possible to connect learners from different corners of the world naturally and efficiently to facilitate peer learning.
- Leveraging Learning Analytics and semantic analysis with Linked Data, an intelligent tutoring system can recommend appropriate OER and stimuli to learners according to the individual’s status. An open education solution to support better independent-learning for more is possible!
In the past, the major limitation of Learning Analytics research is the small size of the data bound within the system generating the data. The small size limits the confidence level when applying to other systems and learners. OER implemented with xAPI can be openly utilized; interoperable learning data across ecosystems and countries can build up strong Learning Analytics research outcomes which can help all types of learners, whether they are in big cities or rural poor villages. The data size from xAPI-enabled OER usage could be very large because of the data sharing and interoperability.
(continue to read: Experience API (xAPI): Potential for Open Educational Resources – Part 4)
At this time of publishing this article, ADL has published xAPI vocabulary spec. v1.0 draft:
Vocab Spec. – gitbook: http://bit.ly/read-vocab-spec
Vocab Primer – gitbook: http://bit.ly/read-vocab-primer