An ontology-based approach to develop an individualized learning digital platform
Technology in general, and AI, in particular, enables a deeper understanding of learning processes, knowledge discovery, knowledge creation & knowledge sharing across borders. For creating an effective ‘Digital ecosystem for Learning’, learners should be empowered with the skills & knowledge that are actually relevant in real life. The education ecosystem needs to make Learners the most ‘active’ agents in the learning process. Howard Gardner’s theory of Multiple Intelligences , rooted in psychology, highlights flaws in any system that teaches and assesses every learner utilizing monolithic parameters, modes, and processes. Learners have different intellectual & emotional strengths that need to be kept in mind while helping them learn, represent concepts in their minds and demonstrate their learning and current knowledge base. Learners think logically, spatially, visually, philosophically, often in an immersive, hands-on way; and AI can be leveraged to present content to learners in a way that is interesting for the learner and will be able to engage and build the learner’s intelligence productively.
Technology can be interactive too so that learners could demonstrate their understanding in the way that is most suited to them. This individualized learning and assessment require a thorough knowledge of the Learners which be obtained through their digital footprint enriched in both educational & non-education contexts. True individualization of learning requires cognizance of Learner’s existing knowledge and complexity comprehension level to ascertain the ‘sweet spot’ to introduce the appropriate content at the appropriate complexity level to induce “Flow” in the learners.  “Flow” is described as a state of maximum focus, dedication, and immersion in a learning activity.
The framework that we propose for truly ‘Learner- or User-centric approach’ is an Ontology-based User Modelling Framework that models the user’s most meaningful actions & behavior according to the key aspects of the users  interacting with the Knowledge & Skill-building Digital Platform. UNESCO MGIEP’s indigenously developed digital Platform called FramerSpace also leverages Ontology-based models. User’s metadata along with user preferences, goals, needs, and interests, stored as a user ontology will constitute the foundation layer of the underlying solution architecture for Semantic Web to control user’s Learning Flow. Semantic Web technology is key for moving towards collaborative, semantic-based information access.
The framework explores the link between user model and user activities such as: creating knowledge, sharing knowledge, learning and getting feedback based on learner’s activity in the system. A ‘Learner-centric’ approach defined by learner’s interests, goals, needs could be the basis for achieving ‘Flow” in learning and can be used as the basis for establishing virtual collaborations. The characteristics of this state are the transformation of time perception and loss of self-consciousness, which means feeling that time flies and all the problems and ideas in a learner’s head clear away.  Ontology-based User Modelling correlates very well to Resource Description Framework (RDF) Triples linked data representations that Semantic Networks (SNNs) capture & represent very well [4, 5].
SNNs are evolved von- Neumann’s Neural networks that enable SNNs to process not just the logical values but also fuzzy values.  This is accomplished by every neuron having a unique identifier and virtual connections or Pointers between neurons. Blockchain provides an optimal structure to store these representations and various linkages between User, Knowledge & Skill Ontologies. Blockchain can give confidence to certain assertions or RDF triples, and using semantic web we can link information and map data from different chains and/or contracts .
The application of ontology to enable semantics-driven data access and processing or semantic-enhanced search is critical. Data preparation for the SNNs includes legacy knowledge sources mapping into the knowledge domain ontology and semantically enriching the sources. For effective semantic- enhanced FramerSpace platform, advanced semantic annotation tools are developed for authoring annotation with well-defined metadata for the Legacy resources. This semantically-enriched Dataset powers better Skill & knowledge indexing and searching processes and implicitly a better Information structure. An ontology-based system can be used not only to improve the precision of search/retrieval mechanism but also to reduce search time. Ontologies offer a flexible and expressive layer of abstraction, very useful for capturing the information of repositories and facilitating their retrieval either by the user or by the system to support the user tasks . For these reasons, ontology-based approach for both Users & Information is at the core of the solution architecture for the development of next-generation of a semantic-enhanced platform like FramerSpace.
Ontology-based user Framework has been designed as a four-tiered application dedicated to managing Learner’s Digital Models. The framework architecture is modular, so that it may be extended and used for any application domain.
The Framework architecture layer  includes:
• Front end UI Layer (Interface)
• Middleware — Business Logic Layer
• Ontology layer (User/Knowledge Domain/ Skill Domain)
• Object Layer (User Instances/ Knowledge & Skill Repository)
The User model gets fine-tuned both explicitly through User profile editor available in Interface Layer that helps Users exercise Self-awareness & Self-assessment while editing their own profile & implicitly through different User modeling techniques that take into account the learning from the User’s engagement with the FramerSpace platform. This User model helps provide feedback, benchmark against the social behavioral norms, compare contextual trends in peer-to-peer interactions. The Middleware layer is where the individualization services reside and provide ‘individualized’ linkage between external requests and the data layer.
Taking into account the User’s idiosyncrasies (preferences/ context/ actions/ expertise), the Middleware enables UI individualization. Information access complexity, structure & modality to increase engagement, according to the User’s individual preferences and observed behavior, is calculated based on the data extracted from the digital footprints of the user interactions with FramerSpace platform and Semantic web. The Ontology layer is powered & represented by SNNs that enable implementation of heuristics and fuzzy logic rules that allow the logging of the interaction type, interaction scale and collaboration score of the users. Domain & User semantics are mapped into the user and domain ontology. Ontology Layer of FramerSpace platform captures & represents the relationships between different ontologies — User (Behaviour, Interest, Goal, Accessibility, Activity, Competency, Qualification, Relationship)/Content (Concepts, Domain Categories, Properties, Concepts & Sub-concepts inter-linkage) & User adaptive Interactions . This layer plays a major role in developing a shared understanding of terminologies and relationships globally, diluting communication barriers especially in diverse virtual communities. The Object layer comprises current snapshots of various system objects including the semantics of the User system interactions. It captures all the transaction logs of the User actions and the triggered events in FramerSpace platform .
Fuzzy classifier systems are further leveraged in the framework to assign users to a certain category according to their level of knowledge sharing (activity log). Fuzzy logic is often used to model various types of common sense reasoning similar to a more humane way of thinking and reasoning. Fuzzy logic extends conventional Boolean logic to handle ambiguity and uncertainty or partial truth. The value between completely true and false is determined by the membership function which takes value in the [0,1]. Fuzzy reasoning was introduced by Zadeh in 1960s to handle the uncertainty of natural language . We use the principle of fuzzy classifier systems in order to assign the users in different categories according to their level of knowledge sharing.
Fuzzy classifier systems imply a two-step process :
- to create a fine-grained fuzzy partition;
- to generate fuzzy rules and calculate membership function or degree of membership;
Processing the activity log, Ontology-based user Framework captures the level of adoption of knowledge sharing practices based on two fuzzy sets — Activity type and the Activity level to codify the membership value of a user to a certain category . Change in the output value indicates a change process that brings users from their old practices to the conscious adoption of information management practices (e.g. transition from low or non-existing levels of information sharing practices to the widespread adoption of best practices in knowledge sharing) through different types of agent-based interventions.
User Data models and metadata can be used for different scenarios including individualization, collaboration, expertise/competency discovery & assessment. Further, metadata and user modeling can be leveraged to manage tacit knowledge and implicit competencies. User ontology concepts can be mapped with the concepts of the domain ontology through properties. Thus, without requiring users to constantly update their profiles (their expertise, interests), an ontology-based platform like FramerSpace could facilitate finding the relevant domain course(s), domain experts in domains of interests for the users. Furthermore, the inferred user’s expertise and interests can be used for pushing relevant knowledge, creating communities of practice or learning networks where experts and peers can collaborate, interact, communicate or share knowledge. Such mechanisms would enable to make explicit some of the competencies that a user might not be aware of and might help educators better manage their learners’ competencies and skills and thus integrate personal knowledge management features into FramerSpace platform.
Finally, User modeling in virtual educational context takes on a lot of challenges like reducing information overload, additional support in learning complex concepts, traversing through diverse collaborative learning environments, individualization, optimized tools and mechanisms to share & discover tacit knowledge and getting right expert help at the right stages. To scale the model globally, to truly enable the exchange of best practices for knowledge sharing across virtual communities, security considerations & policies also need to be proactively designed and enforced in the FramerSpace platform. The security protocols for Information exchange do not encompass only the information sources that the learners are permitted to access, but they also extend to the relevant regulations the educators are obliged to enforce. For rich user modeling, disclosure of user data that enables new forms of individualization, communication, collaboration, and social interactions. To alleviate user’s Data privacy & security concerns, we have to put the user in control of their own respective profile data. The user profile editor enables the users to enter and update personal information and thus instantiate the user ontology. It is possible to support more complex knowledge-oriented processes by exploiting the metadata and the relationships between the concepts (concept-based navigation). The advantage is the power of the relationships which enables users to navigate easily from one concept and its instances to another concept and its instances .
Metadata and ontology-based representations connect knowledge resources with people through contextual links among the various chunks of tacit knowledge within Open Education Resources. The collaborative and personal dimensions in virtual user personas and virtual learning environments are important features for the next generation of platforms like FramerSpace. The user ontology along with user modeling processes will support a more learner-centric or user-centric approach of Semantic Web .
Semantic Web can be foreseen to provide more relevant content for the users integrating different sources of information, using Individualized recommendations in order to better harness collective knowledge, to reduce information overload and support attention regulation in learners, to better support users in searching for information and make recommendations of relevant content using collective intelligence, to better support lifelong learning and personal knowledge management, and/or to better help users to achieve their goal. Individualizing learning leveraging AI will be one of the defining characteristics of the next generation of services where the semantics of data will play a key role along with specific goals or characteristics of the users stored in a user ontology.
Individualized and adaptive education technologies are attempting to deliver differentiated learning with one-on-one virtual learning tailored to individual learner needs, often used effectively with blended-learning approaches mixing in-person and online instruction. These programs can be used in conjunction with in-classroom instruction, freeing up educators’ time to deepen learners’ understanding of the material and to develop skills like problem-solving, creativity and collaboration. They can also harness the power of data to dynamically assess learning, address gaps and track outcomes.
AI-powered FramerSpace platform can provide the back-end analytics necessary to offer an adaptive experience to learners and provides an engine that allows others to build adaptive learning applications and experiences from a wide range of content, as well as to assess what works best. The FramerSpace platform enables educators to create “adaptive pathways” for the lesson materials they create. This allows educators to design a unique and differentiated experience for learners.
Further, games and interactive simulations embedded organically in pedagogy allow learners to go beyond the traditional lecture and to interact with instructional content in an engaging way. Games allow a focus on multiple skills at once: while learners work to improve their understanding of core concepts, they can also develop skills such as creativity, curiosity, and persistence in the process. These tools, along with new pedagogical approaches such as project-based learning, are therefore at the forefront of addressing skills gaps in competencies and character qualities.
This post presents an ontology-based approach to develop an individualized learning platform like FramerSpace that creates adaptive content based on learner’s abilities, learning style, level of knowledge and preferences. In the approach, ontology is used to represent the content, user and domain models. The Ontology-based User model describes learner’s characteristics required to deliver tailored content. The domain model consists of some classes and properties to define domain topics and semantic relationships between them. The content model describes the structure of courses and their components. The system recognizes changes in the learner’s level of knowledge as they progress and the ontology-based user model is updated based on learner’s progress and the passage from one disposition of learning process to the next is determined based on the updated learner’s profile. The post also details the whole-brain approach with the challenges and opportunities that use of technology, specifically Games & AI, can present in the context of Learning.
Finally, we posit that integrating training in intellectual and socio-emotional skills, specifically critical inquiry, mindfulness, empathy, and compassion, will create responsible and caring global citizens who are aware of the consequences of their choices. This integration is critical in order to achieve the desired objective of re-orienting the purpose of education to human flourishing and well-being.
 Gardner, H. (1999). Intelligence reframed: Multiple intelligences for the 21st century. New York, NY, US: Basic Books.
 Mihaly Csikszentmihályi (1990). Flow: The Psychology of Optimal Experience. Harper & Row. ISBN 978–0–06–016253–5.
 Razmerita, Liana. “An Ontology-Based Framework for Modeling User Behavior — A Case Study in Knowledge Management.” IEEE Transactions on Systems, Man, and Cybernetics — Part A: Systems and Humans 41 (2011): 772–783.
 World Wide Web Consortium (W3C). (2014, Feb). Resource Description Framework (RDF) [Online]. Available: https://www.w3.org/RDF/
 Berners-Lee, Tim, James Hendler, & Ora Lassila (2001) The Semantic Web, Scientific American, May 2001. ; Hendler, James A., & Frank van Harmelen (2008) The Semantic Web: webizing knowledge representation, in van Harmelen et al. (2008) pp. 821–839.
 B.R.Gaines, “Foundations of fuzzy reasoning” — Man-Machine Systems Laboratory, Department of Electrical Engineering Science, University of Essex, Colchester, Essex, U.K.
 Héctor E. Ugarte R. Rheinische Friedrich-Wilhelms-Universität Bonn — Germany — “A more pragmatic Web 3.0: Linked Blockchain Data”
 Yaquob Alsobhi, Aisha & Khan, Nawaz & Rahanu, Harjinder. (2015). “Personalised Learning Materials Based on Dyslexia Types: Ontological Approach”. Procedia Computer Science. 60. 113–121. 10.1016/j.procs.2015.08.110.
 Giorgos Stoilos, Giorgos Stamou, Jeff Z. Pan, Vassilis Tzouvaras, and Ian Horrocks. 2007. “Reasoning with very expressive fuzzy description logics”. J. Artif. Int. Res. 30, 1 (October 2007), 273–320.
 Stojanovic L., Maedche A., Motik B., Stojanovic N. (2002) “User- Driven Ontology Evolution Management.” In: Gómez-Pérez A., Benjamins V.R. (eds) Knowledge Engineering and Knowledge Management: Ontologies and the Semantic Web. EKAW 2002. Lecture Notes in Computer Science, vol 2473. Springer, Berlin, Heidelberg
 World Wide Web Consortium (W3C). Linked Data [Online].