X5gon: Cross Modal, Cross Cultural, Cross Lingual, Cross Domain, and Cross Site Global OER Network

August 1, 2017 – Data Scientist career opportunity – Recommendations  Job Description: University of Nantes wishes to recruit an engineer for project X5-GON. The goal is to specify, develop and deploy a recommender system solution.

August 2, 2017 – X5GON to be presented @ European Commission The X5GON project on converging OER was invited to present at the workshop “H2020 Media Projects’ Workshop: Collaboration Towards the Future of Media” on 17th October 2017, at EC premises in Avenue de Beaulieu 25 in Brussels.

X5gon stands for easily implemented freely available innovative technology elements that will converge currently scattered Open Educational Resources (OER) available in various modalities across Europe and globally.

X5gon will combine content understanding, user modelling and quality assurance methods and tools to boost creating a homogenous network of (OER) sites and provides users (teachers, learners) with a common learning experience. X5gon will deploy open technologies for recommendation, learning analytics and learning personalisation services that will work across various OER sites, independent of languages, modalities, scientific domains, and cultural contexts.

It will develop services for convergence of OER media which includes full courses, course materials, modules, textbooks, streaming videos, tests, software, related events and any other tools, materials, or techniques used to support access to knowledge.

The solutions that will be offered to OER sites are fivefold:

• Cross-modal: technologies for multimodal content understanding
• Cross-site: technologies to transparently accompany and analyse users across sites
• Cross-domain: technologies for cross domain content analytics
• Cross-language: technologies for cross lingual content recommendation
• Cross-cultural: technologies for cross cultural learning personalisation

The project will collect and index OER resources, track data of users and their progress and use that to drive an analytics engine driven by state-of-the-art machine learning that can improve recommendations through better understanding of users, their progress and goals, and hence their match with knowledge resources of all types. In addition X5gon will implement innovative models and methods for OER quality assessment and assurance, including trust networks between teachers for OER creation and exchange, automatic content validation and user experience.

The project will run a series of pilot case studies that enable the measurement of the broader goals of delivering a useful and enjoyable educational experience to learners in different domains, at different levels and from different cultures. The two exploitation scenarios are planned: (i) free use of services for OER, (ii) commercial exploitation of the multimodal, big data, real-time analytics pipeline.

Objective 1: X5gon, will combine content understanding, user modelling and quality assurance methods and tools to boost creating a homogenous network of OER sites and provides users (teachers, learners) with a common learning experience independent of the OER site.

Objective 2: X5gon will create a self-sustainable and growing mode of operation based on bottom-up OER sites collaboration, easy to install open software and Ministry of Education with governmental channels and UNESCO National Commissions for wider adoption and showcase for policy makers and industries with a clear potential for adoption in the Digital Single Market.

Objective 3: X5gon will tackle several online learning research and innovation challenges related in particular to big data, real-time response and multimodality.

Objective 4: The market opportunity for a Europe-centric education platform is considerable. Two important outcomes from this project that can be viewed as separate entities for the purposes of exploitation are (i) free use for OER of cross-recommendation, cross-site learning analytics and learning personalisation functionalities (ii) the enabling technology stack. Given the different risks/reward profiles, these outcomes require different strategies for exploitation.

1. University College London
2. Institut Jozef Stefan
3. Knowledge 4 All Foundation
4. Universitat Politecnica de Valencia
5. Université de Nantes
6. Universitaet Osnabrueck
7. Slovenian Post
8. Ministry of Education of Slovenia

• WP1: Learning rich content representations (UCL) This work package will automatically learn representations of OERs so that they can later be used by the X5oerfeed and recommendation engine. In context of education, providing the learner with a high quality resource (quality assurance) that is reliable, authoritative and is topically relevant to the learner’s interests is highly important. Through this WP, we will devise methods that can automatically infer the quality and authority of each OER, as well as the topics covered by the OERs so that they can be used in recommendation. We will further devise evaluation methodologies for measuring the quality of content representation models and quality assurance models.
• WP2: Analytics Infrastructure, Services and API (JSI) is concerned with the infrastructure for representing the relevant information developing an API that all the WPs make use of. Additionally it creates (1) the software platform which will be the main link connecting the different components of the system, (2) services and products and (3) respective APIs to connect. The three services to be offered are X5oerfeed, X5analytics and X5recommend.
• WP3 Learning Analytics Engine (NA) is the analytics engine for the analysis of learning and testing aspects, links with educational theories, affective computing, etc. including cross-modal cross-lingual and cross-cultural aspects. The information generated from the tools developed in WP1 and WP2 need to be cross referenced in order to obtain high level information. It will propose personalised learning as an individualized navigation through the OERs by the different sites. A more complex issue will consist in answering the following: what resource should the system propose to a specific learner, at a specific moment? This requires to be able to predict the intent of the user. Finally, prediction and recommendation will depend on collected personal information. The ethical issues related with this collection and with what should be allowed, will be discussed, and compared with international standards. A specific effort will be made here.
• WP4 Recommendation Engine (JSI) is concerned with designing rich models of users on learning sites, and use these models for recommendation and personalization of learning material. Initially we will leverage other users but in the latter stages relying increasingly on the analytics engine developed in WP3.
• WP5 Piloting (UPV) is concerned with piloting successive versions of project components and providing feedback to other WPs.
• WP6 Studies in the wild (UCL) is concerned with the studies, their design and goals, their execution and the linking of the results with the knowledge produced by the analytics engine.
• WP7 Dissemination (K4A) is concerned with the dissemination of the results. It will be innovative in order to acquire new and unmapped users. It will be tailored to the needs of the identified target audiences, including groups beyond the project’s own community. As X5gon has a clear interdisciplinary and intersectoral dimension: it transcends several domains: education and communication and information and science, aspects of public/societal engagement on issues related to the project will also be addressed.
• WP8 Exploitation (PO) is focussed on exploitation, a key component of the project. This will focus on two scenarios and will feed technology and data-drive results into the policy making partner MIZS.
• WP9 Management (UCL) arranges for the management of the project.

X5gon is an analytic platform with open services, APIs and scripts supported with AI enabled technical pipeline to converge dispersed open educational resources (OER) media content to learners and users into a one-stop-shop data-driven learning environment.

Keywords: Open Educational Resources, machine learning, cross-site, cross-domain, cross-modal, cross-language, cross-cultural, cross-social, adaptive learning, policy making, web and information systems, database systems, communication networks, media, information society, accessibility, cultural studies, cultural diversity.

Advisory Board (AB) is formed with high level professionals in the field of artificial intelligence, educational technologies and open education. The main purpose of the Advisory Board is to provide management advice about the general direction the project should follow. The Advisory Board members are:

• Rayid Ghani (m) is the Director of the Center for Data Science and Public Policy, Chief Data Scientist at the Urban Center on Computation and Data, Research Director at the Computation Institute (a joint institute of Argonne National Laboratory and The University of Chicago), and a Senior Fellow at the Harris School of Public Policy at the University of Chicago. He is also the co-founder of Edgeflip, an analytics startup that grew out of the Obama 2012 Campaign, which was led by Rayid and focused on social media products for non-profits, advocacy groups, and charities.
• Jan Newman (m) is working as head of legal affairs and organization at the North Rhine-Westphalian Library Service Centre (hbz). He is a member of the educational advisory board of the German UNESCO chapter and blogs about Open Educational Resources (OER) on OERSYS.org. Since 2014 he is the project manager of the OER World Map project, which is funded by the William and Flora Hewlett Foundation and aims at collecting data on OER actors and activities worldwide. Jan is author of several publications on OER, frequent speaker at OER conferences and participated in the organization of the German OER conferences OERde 14, OERde 15 and OERde Festival.
• Dr Javiera Atenas (f) is the co-coordinator of the open education working group for Open Knowledge International, senior fellow of the higher education academy and associate lecturer at Aga Khan University and University de Barcelona.
• Professor Pierre Dillenbourrg (m) has been professor assistant at TECFA, University of Geneva. He joined EPFL in November 2002. He has been the director of CRAFT, the pedagogical unit for 10 years and is now the academic director of the EPFL Center for Digital Education and head of the CHILI Lab: “Computer-Human Interaction for Learning & Instruction”.
• Rose Luckin (f) is Professor of Learner Centred Design at the UCL Knowledge Lab, focusing on Artificial Intelligence and Educational Technology. Until 2011 she was a member of the board of BECTA (the British Educational Communications and Technology Agency), the body charged with implementing the UK government’s eLearning strategy. Her work is interdisciplinary and encompasses education, psychology, artificial intelligence and HCI. Rose investigates the relationship between people, their context, the concepts they are learning, and the resources at their disposal. Rose holds a 1st class BA in Computing and Artificial Intelligence and a PhD in Cognitive Science.
• TJ Bliss (m) is the Director of Development and Strategy at Wiki Education, a non-profit that connects higher education to the publishing power of Wikipedia. Bridging Wikipedia and academia creates opportunities for any learner to contribute to, and access, open knowledge. Before joining Wiki Education, TJ was a Program Officer in the Education Program at the William and Flora Hewlett Foundation. In that role, he gave $45M in grants to over 30 organizations working to expand the reach and efficacy of Open Educational Resources (OER)