A SEMANTIC APPROACH FOR A UBIQUITOUS LEARNING SYSTEM
, M. Goga2
1University Politehnica Bucuresti (ROMANIA)
2Technical University of Civil Engineering of Bucharest (ROMANIA)
Once seen as a more traditional field, education is being now transformed by new trends such as personalized, student-centered learning and lifelong learning but also challenged by new lifestyles, more mobile and agile, in which the traditional classroom is becoming obsolete. Technology can help the transition to newer and more effective learning styles.
In this paper we are presenting the prototype of a ubiquitous learning (u-learning) system. The system is context aware, learner-centered, providing rich, adaptive learning experiences tailored to the learner`s personal characteristics, to the learner`s location, behavior and device specifications. The objective is to provide a pervasive learning environment for lifelong learning. The learning domains offered by the application are varied. In order to maintain such a volume of information and to properly choose the most relevant information to display to the learner, the system uses a semantic approach, the content being structured by ontologies corresponding to the domains of interest (history, art, geography, culture, entertainment, tourism, healthcare, etc.). Since ontologies are not available for all the targeted domains, and since manual construction of ontologies is a very labor-intensive process, we are using ontology learning techniques, especially learning from unstructured texts in order to build our ontologies. The ontology learning component uses natural language processing algorithms such as shallow parsing and statistical approaches to extract concepts, instances, subclasses and general relations using regular expressions.
Once these ontologies are built they can be further refined or reengineered by domain experts. Starting from the generated ontologies, we use specialized crawlers to extract text from the Internet based on the concepts and relations in the ontologies. The texts are added to the electronic library which is the source for the presented learning content. The u-learning system`s interface with the user is multi-modal with applications for smartphone, tablet and web.
The localization module is used for both outdoor localization using the reading device (smartphone, tablet) GPS or GSM module and for indoor localization using wireless technologies such as Wi-Fi or Bluetooth Low Energy. The outdoor localization, combined with the personal profile data can be used to predict the duration of the daily commute and provide a learning material fitting those time parameters. Similarly, the indoor localization can be used to deduce the activity context of the person, also correlated with her/his daily routines.
Currently our system is in the prototype phase, with some modules having a greater maturity status such as the ontology learning and the localization components. The paper describes the state-of-the-art, the design of the proposed solution, preliminary implementation and first results.