EXPLOITING CONCEPT MAP MINING PROCESS FOR E-CONTENT DEVELOPMENT Chara Panopoulou 1 , Natalia Spyropoulou 1 , Christos Pierrakeas 1,2 , Achilles Kameas 1 1 Hellenic Open University (HOU) (Greece) 2 Dept. of Business Administration, Technological Educational Institute (TEI) of Western Greece (Greece) Abstract E-learning has revolutionized education all over the world, defining a different and promising aspect of education, reinforcing the perception that it builds inclusive knowledge societies. Higher Education Institutes (HEI) adopted this innovative education model in order to provide to their students the option of distance education. Since the most important component of e-learning is e-content, its development is a popular research topic in the educational community. Due to the fact that e-content reusability can be increased by using an approach based on Learning Objects (LOs), many methodologies of e- content design introduce guidelines for creating LOs. LOs can make the process of e-learning effective and can offer high quality e-Learning experience to students. The Hellenic Open University (HOU) is introducing LOs in the educational process, using teaching subject domain ontologies to describe them. Ontologies provide a simple way of identifying the knowledge domains covered by LOs, while facilitating its reusability. The preliminary step to create these domain ontologies is the design of the Concept Map (CM), that is a diagram for representing knowledge in a structured form. Concept Maps foster meaningful learning and serve both as a knowledge base for building domain ontologies and as a frame for composing more detailed LOs. Concept Map Mining (CMM), a process for automatic or semi-automatic creation of Concept Maps from documents, is used to facilitate the construction and sharing of Concept Maps. In this paper, we propose a methodological framework and a semi-automatic method for Concept Map creation from unstructured text, which can even handle the morphologically rich Greek language skillfully. The proposed approach combines language processing tools and the knowledge of domain experts. In addition, a case study based on a HOU teaching domain is presented, illustrating the process of Concept Map Mining and showing encouraging results. Keywords: e-content, e-Learning, Concept Map, Concept Map Mining 1 INTRODUCTION The digital age has brought upheaval in the fields of education and educational content development. It is a widely held view that the educational content is the main driver of the educational process, especially in distance education, considering that the educational content aims to undertake the largest possible part of the instructor’s role in the distance education process. Thus, the focus of many European institutions’ interest turned towards the design and development of qualitative digital educational content [1]. Any form of digitized content that can facilitate the learning process can be defined as e-content [2]. E-Content can effectively enhance the online academic course creation offering flexible access to learning opportunities without the time and distance barriers. Higher Education Institutions (HEIs) incorporate online courses in their curriculum [3] and develop e-content in order to complement these courses [24]. As the content development plays a key role in e-learning, the e-content must be designed properly [4]. One way is the e-content to be designed and developed in smaller, modular, discrete units of learning known as Learning Objects (LOs) [24]. Therefore, HEIs have adopted instructional technologies in e-content development such as the Learning Objects. Learning Objects could enable higher education to capitalize on the promise of e-learning, as their size and manageable format allow the re-use of material and offers the possibility to be searchable and accessible from everywhere to a wide audience [5]. The specification of a standardized set of metadata contributes to improving LOs reusability [6]. With the usage of ontologies, the representation of learning resources is attributed to a finer level of detail, providing accuracy and flexibility for LOs metadata [8]. However, ontologies are used not only