Abstract:

As a research centre specializing in e-learning, the UOC’s eLearn Center (eLC) continuously analyses global sources of knowledge to detect and reflect on the trends and events that transform online higher education and lifelong learning around the world. As a result of this prospective task, the eLC periodically publishes reports on trends in the field of e-learning which require its experts to carry out repetitive extraction, analysis, synthesis, codification and classification tasks on hundreds, and in some cases thousands, of bibliographic references. Given the systematic nature of the analysis process and the large volume of articles and other documentary sources needing regular review, data analysis techniques and text analysis tools were thought to have the potential to adequately cover part of this laborious process. Under this hypothesis and based on a report on trends in current e-learning research produced by the centre’s researchers, a deep categorization model was applied, capable of extracting relevant information from hundreds of scientific articles and automatically classifying it according to the parameters defined in the study (main topic, subject category, research methodology and educational stage). The results achieved by applying the automatic classification model were promising, given the closeness shown in the comparison of the data obtained by the model and the manual task (standard deviation less than 2%).

Keywords:

knowledge management, text mining, deep categorization, text processing, data analytics.