Development of an Automatic Process for Recommending Well Adapted Educational Resources in an E-learning Environment

Learners in e-learning environments often try to go beyond what can be provided to them by teachers in terms of educational resources, trying to find other external resources on the web related to their preferences and needs. Based on this observation, and with the goal of personalizing the learning process by recommending the most appropriate teaching resources to learners, we propose a model for analyzing learners' searches to determine the relevant words that reflect their interests, with the aim of enriching their profiles. Our approach is to collect the descriptions of the links returned by the search engine, which will constitute a corpus on which we will apply the TF-IDF (term frequency-inverse document frequency) method to determine the relevant words. We will then use the Word2vec technique to determine words similar to these relevant words in the description of internal educational resources, so that we can recommend those that best correspond to the learner's needs. We developed a platform and launched an experiment with a group of real students. The data generated will be analyzed and the results will allow us to evaluate our approach and also to see areas for improvement.