Keywords

1 Open Education. The Direction

1.1 A Review of Open Policies and Declarations

Open Education is based on the birth and evolution of different declarations of human rights, especially those related to Education. “Everyone has the right to education.

Education shall be directed to the full development of the human personality and to the strengthening of respect for human rights and fundamental freedoms” says the Article 26. Subsequently, there were many declarations, especially sponsored by UNESCO and different governments that called for this right to education as one of the human rights [1].

The precedent for the current SDGs were the Millennium Development Goals, in 2000. Goal 2, for Achieve universal Primary Education [2], evolved into SD4 of 2015 for Quality Education [3].

But what is fundamental here is that education is associated with the open nature of this objective.

Parallel to and converging with the right to education, the Open Education movement appears more formally and forces it with two lines:

  • In 2001, the Massachusetts Institute of Technology (MIT), in an unprecedented twist, announced the publication of almost all its courses on the Internet, accessible to the entire public. This model has been imitated and the number of institutions that offer educational materials free of charge or open to the public has grown exponentially.

  • UNESCO organized in 2002 the first world forum on open access educational resources in which the expression “open access educational resources” was adopted. OERs are teaching, learning, or research materials that are in the public domain or published under an open license that allows them to be freely used, adapted, and distributed [4].

Since then a series of declarations have expanded and specified the character of Open Education. For example, the Cape Town declaration is a multi-dimensional movement which includes widening participation, lifelong learning, open educational resources, open educational practices (OEP), Massive Open Online Courses (MOOCs), open science and open access publication. These share a common commitment to the view that “everyone should have the freedom to use, customize, improve and redistribute educational resources without constraint”[5].

​​Carrying out the actions recommended by the different statements above is only possible by exploiting the openness rights over the resources. The 5Rs of Openness, from most to least degree of openness are Retain, Reuse, Revise, Remix and Redistribute [6]. Users can adapt OERs to their needs, mix them with other materials, share them, and retain copies.

1.2 Open Repositories

Generally, sharing OERs has focused on publishing educational materials with open licenses on the internet. As the open access movement has advanced, the number and variety of OER providers and content have grown. With this expansion, a broader range of resource types and their representation as available data to inform these OERs in health and care has emerged, including device data, patient-generated data, and data related to the social determinants of OERs. Some of the most important features of the current situation of OERs worldwide include:

  • Collaboration and Community: The OER ecosystem is a community where educators, students, self-learners, and others can collaborate, share resources, and learn from each other, each with their own language and culture.

  • Decentralization: There is no central authority that controls or regulates the OER ecosystem. Each provider operates autonomously, making their own decisions on how to describe, share, and license their content.

  • Diversity of Providers and Content: The open content context includes a wide range of providers, from educational institutions to non-profit organizations, private companies, and individual educators.

  • Heterogeneity of Technologies: OERs are created and consumed using a variety of technologies, from simple text documents to sophisticated OER repositories or even online learning environments.

In line with one of the recommendations on OER from UNESCO 2019, for Building capacity of stakeholders to create, access, re-use, adapt and redistribute OER, one of the recommendations is” leveraging open licensed tools, platforms with interoperation of metadata, and standards (including national and international) to help ensure OER can be easily found, accessed, re-used, adapted and redistributed in a safe, secure and privacy-protected mode. This could include free and open source authoring tools, libraries and other repositories and search engines, systems for long-term preservation and frontier technologies for automatic OER processing and translation of languages (where appropriate or needed), such as artificial intelligence methods and tools” [7]. The works presented in this paper, an OER Ecosystem (Sect. 2) and the case of the STEMSOFT project (Sect. 3) are aligned with this recommendation.

2 OER Ecosystem: LOD4OER

2.1 Brief Description of LOD4OER. Towards a OER Semantic Interoperability Ecosystem

The term “interoperability” is widely applied to all forms of data exchange, including those related to open access. In the context of OERs, interoperability is the ability of various information systems to access, exchange, integrate, and cooperatively use data and meanings. Semantic interoperability facilitates a timely and seamless transfer of data and metadata that describe the OERs, thus optimizing the openness of educational resources worldwide.

In this environment, it is essential that the open education movement orient itself towards the creation of a semantic interoperability ecosystem in OER. This ecosystem would comprise the collection of practices, technologies, cultural aspects, and relevant legal frameworks for transactions of all kinds of digital information between different organizations. The goal of promoting an OER ecosystem lies in enhancing the exchange of information on open educational materials, facilitating access and retrieval of OER data and metadata to provide better opportunities in access, use, reuse, remix, and adaptation of OER in a timely, efficient, effective, and equitable manner, user-centered. OER providers, as well as open access policy makers, can also benefit from the creation of this ecosystem to support the analysis of the state of open access to education for different populations.

The semantic interoperability ecosystem of OER includes OER providers, information systems, technologies, and processes that seek to share, exchange, and access all forms of information related to open educational materials. Content providers, information systems, researchers, funders, and users are potential stakeholders within this ecosystem. Each of them is involved in the creation, exchange, and use of information and/or OER data. An efficient interoperability ecosystem provides an information infrastructure that uses technical standards, policies, and protocols to enable the capture, discovery, exchange, and utilization of OER data and metadata in a smooth and semantically seamless manner.

From a technology and data perspective, the Semantic Web, Semantic Knowledge Graphs, and Ontologies are powerful tools and techniques that can help overcome many of the challenges in the OER interoperability ecosystem contributing in [8,9,10,11]:

  • Discovery and Access: Semantic Knowledge Graphs and ontologies can represent complex relationships between different resources and concepts, which can help users find relevant resources more intuitively.

  • Quality and Consistency: The quality and consistency of OERs can vary considerably among different providers. By applying ontologies, it is possible to represent and maintain a quality control and consistency of resources.

  • Data/Metadata Integration: A Semantic Knowledge Graph can integrate data from multiple sources coherently, facilitating the interoperability of OERs across various platforms.

  • Adaptability and Personalization: A Semantic Knowledge Graph can help personalize OERs for individual users by providing a detailed view of how individual resources relate to topics of interest to a particular user

  • Resource Enrichment: Semantic Knowledge Graphs and Ontologies can be used to enrich OERs, adding semantic metadata that describes the content and context of resources.

  • Licensing and Copyright: Navigating the complexities of licenses and copyrights can be a challenge.

The implementation of these solutions involves working collaboratively with OER creators, educators, students, as well as technology developers, to ensure that the needs of all actors are understood and addressed through an ecosystem that allows for interoperability and integration of OER. These are the principles that guided the specification of the LOD4OER framework. It comprises a set of definitions, principles, agreements, semantic resources, and good practices that guide the efforts and activities of the members of the OER ecosystem. It is about facilitating the efficient and effective exchange of information, as well as the provision of interoperable OER services to users, developers and other providers of open education services [12].

2.2 Serendipity: A Faceted Search Engine Based on Semantic Web Technologies

As part of the ecosystem we have developed some OER Potential Semantic Apps based on LOD4OER. Serendipity [13] is a platform to discover and visualize OER from around the world. It is a faceted search engine based on Semantic Web Technologies. Moving towards a Web of Linked Data (LD), Serendipity provides a service that enables faceted exploration of large OER collections. The user, when presented with the facets, is likely to discover new facets of the query that they were not aware of before. When clicking on a facet, they will narrow down their search by expanding the original query with the suggested facet. In this way, the user get more accurate and complete results, since it locates OER using different metadata and data elements, providing the user with visible options that help clarify and refine the queries (See Fig. 1).

Fig. 1.
figure 1

Example of a search using Serendipity

Full size image

The new concept of OER under the LOD4OER ecosystem is now a resource semantically operable OER combining global identifiers, machine readable metadata, human readable content, and open license. This approach enables global discovery of OER, the potential development of open educational applications, and expands their accessibility, reuse and adaptation.

2.3 New Opportunities Based on AI for an OER Ecosystem

Data and metadata are both essential components of artificial intelligence (AI) projects. The use of AI opens new perspectives to support the interoperability of repositories, their ease of discovery or aspects of interaction with the user. However the application of AI models in repositories requires quality data, and can be applied to both data and metadata. Metadata provides context and meaning to algorithms making them better understood by AI algorithms. But to improve the results of AI models it is necessary for ensuring that data is clean, relevant consistent, representative and accurate.

On the other hand, the collection and use of a wide collection of data implies that these processes are treated by ethical guidelines. For instance the “Ethical Guidelines for Educators” deal with aspects such as explaining how AI algorithms work [14] or the “Recommendations on the Ethics of Artificial Intelligence” explains how AI modifies the current roles of teachers and students [15].

3 A Case of Application: The STEMSOFT Project

3.1 The Project

STEMSOFT [16] aims to design an open learning environment based on OERs, that facilitates the development of soft skills for people with an interest in their professional development based on STEM fields, in particular for our defined target groups.

This STEM employability skills project aims to address this by defining relevant non-technical soft skills and transversal skills, described in an EQF/ECVET (European Qualifications Framework / European Credit system for Vocational Education and Training) format, targeting the following groups:

  • Migrants with a STEM background who are unemployed or underemployed;

  • Mobile workers seeking employment opportunities in another country or region;

  • NEETs (Not in Education, Employment, or Training); both male and female;

  • Undergraduates or graduates with a STEM background who are finding it hard to get a STEM job due to ethnicity, disability or gender;

  • People in career transitions (people without a formal STEM background wanting to move to a STEM field as an alternative career) facilitating permeability and flexible learning and training pathways.

Based upon a survey to explore to what extent soft skills are integrated elements in existing STEM provisions, a competence map of STEM oriented learning outcomes covering these skills was developed in an EQF/ECVET format. This mapping corresponds to a Skillsbank [17] and ESCO system (European Skills, Competences, Qualifications and Occupations) [18]. It is a multilingual classification that identifies and categorizes skills, competences, qualifications, and occupations relevant for the EU labor market and education. ESCO has been developed by The European Commission since 2010 to secure transferability between existing training provisions and the additional STEMSOFT training. This in turn will boost the employability of candidates and their access to the labor market.

An analysis of the results of the survey allowed the selection of 7 skills for each of the target groups of the project. Each of the 7 skills represents the 7 units in which the Skillsbank system is structured (Table 1).

Table 1. STEMSOFT Skillsbank: aggregated Units of Learning Outcomes
Full size table

The Skillsbank system aggregates Learning Outcomes into logical units forming full or partial qualifications. The system offers a tool for qualification definitions with structured matrices in multilingual formats. It offers as well a Recognition of Prior Learning (RPL) self-assessment module where an individual person can do a profiling against defined Learning Outcomes and optionally supported by video documentation of skilled performance [19, 20]. (see, in Table 2, a partial view of the Skillsbank structure).

Table 2. STEMSOFT Skillsbank partial view
Full size table

3.2 OER for Soft Skills Development Empowered by Serendipity.

In the context of the STEMSOFT project, we have faced the challenge of improving the accuracy of searching for resources according to the soft skills framework developed in the project. The purpose is to assign meaningful metadata to OER content that makes soft skills OER findable using Serendipity search engine enriched with AI algorithms. This improvement has been based on the application of a method that consists of the following steps:

  1. 1.

    The first stage involved identifying specific Open Educational Resources (OER) from which metadata were extracted. OER can be found in a multitude of online repositories, both generalist and specialized. Some repositories gather resources on a variety of topics, while others focus on specific thematic areas.

Once the appropriate resources have been identified, the next step is to access them. The way this is done can vary depending on the repository (Fig. 2) Some repositories allow direct download of resources, while others require the use of an API (Application Programming Interface) to access the data. In some cases, it has been necessary to write a script to automate the process of downloading or accessing the Web resources.

Fig. 2.
figure 2

OER for Soft Skills development

Full size image

The case of study scope is 15 soft skills: Core literacy functions, Digital literacy, The ability to learn independently, High-level thinking, Teamwork, Empathy and compassion, Self-management and self-discipline, Perseverance and resilience, European citizenship values, Cultural understanding and respect, Decision-making, Planning and management, Ideas and opportunities, Negotiation techniques, Problem identification and solving.

  1. 2.

    In addition to metadata extraction tools, Natural Language Processing (NLP) techniques have also been used to extract useful information from the text of the resources. For example, NLP algorithms have been used to identify named entities (such as topics, OER categories, languages) and themes in the text, which can then be added to the metadata. Metadata extraction is a critical stage in the process of building a semantic knowledge graph, as metadata provides the basic information that will be used to represent and connect resources in the graph.

  2. 3.

    Once the metadata has been extracted from the Open Educational Resources (OER), these have been transformed so they can be semantically interpreted and integrated within the Serendipity semantic knowledge graph. This stage has involved mapping the metadata to the Serendipity OER description ontology, which is a set of concepts and relationships that form a conceptual scheme to describe this work domain.

  3. 4.

    After the metadata has been transformed into semantic data, these can be used to build a semantic knowledge graph. In this graph, the nodes represent entities, which can be anything from individual OER resources to authors, topics, and more. The edges represent relationships between the entities.

  4. 5.

    Finally, the initial knowledge graph has been built and it can be enriched by incorporating more metadata or by linking with other knowledge graphs (see Fig. 3). Metadata enrichment can involve incorporating additional information about the existing entities in the graph, or the addition of new entities and relationships. This task has been done using GPT4.

Fig. 3.
figure 3

Knowledge tree for Core Literacy Functions empowered GPT4

Full size image

Once built and enriched, the semantic knowledge graph can be used for a variety of purposes. In terms of usage, the graph can serve as a powerful tool to enhance the search and retrieval of Open Educational Resources.

Users can explore the graph, follow the connections between entities, and discover new resources that might have been overlooked in a traditional search. Moreover, the graph could feed recommendation systems, which could suggest resources based on the connections and similarities between entities in the graph. Fig. 4 shows the results of number of OER discovered applying this method and an example of the subareas identified for the “Core Literacy Functions” skill knowledge tree.

Fig. 4.
figure 4

Number of OER discovered as result of the application of this method

Full size image

4 Conclusions

The use of AI in education is opening up many opportunities to face challenges that, although identified, do not achieve satisfactory results. The open repositories around which OER ecosystems are defined are no strangers. This paper shows how the definition of an ecosystem oriented towards semantic interoperability can be enriched with AI algorithms or Large Language Models such as ChatGPT that support a more precise search and adjusted to the needs of educators and institutions.