Abstract:

Personalized learning is emerging as one of the most effective instruments for academic success, especially in e-learning environment. It brings learners more flexibility, effectivity, adaptation, engagement and motivation. A good mechanism to adapt both the study materials and the reinforcement activities to the individual needs of the students is the personalized recommendation based on the questions, doubts and concerns expressed by the students themselves.

In this work, we propose the use of advanced natural language processing techniques for the design and implementation of a recommender that will provide students with aditional activities to reinforce the study of the elements or topics that have raised some type of difficulty in their preparation. Given a query posed by a student in the forum facility of the e-learning environment about a given concept, our system will recommend her exercises from a repository to practice and improve understanding of such concept. To this end, we have developed a content-based recommendation system which makes use of an algorithm capable of extracting, with high precision, the keywords (key phrases) that describe the topics or concepts in the student question and those present in the reinforcement activities. The recommender takes into account the similarity of the concepts extracted from the student’s and those covered in a database of exercises to recommend those that best meet the student’s need.

Previous works have proposed the automatic recommendation of contents and activities in e-learning settings. However, such works build their recommedations on the information from the student profiles, such as past actions taken, activities already accomplished and results achieved in previous tests or activities. In contrast, our recommender does not make use of any previous information about the student, and therefore, it does not face the cold start problem.

Our hypothesis, which will be evaluated both quantitatively and qualitatively, is that the personalized recommendation of reinforcement materials to expand and strengthen the concepts studied in a course, will be of great help for students and will, predictably, translate into an improvement of the learning results, while alleviating the workload of the teaching team, by reducing the need of answering questions that are frequently repeated.

Finally, we present an application of the recomender to a course related to the study of algorithms and advanced data structures. This is a key field in Computer Science degrees and one of the most difficult, so that improvements in this course may have a great impact on the students' perception and satisfaction. Moreover, since these subjects are usually taught in the first years of computer science degrees, they are crucial when it comes to reducing dropouts and improving degree indicators.

Keywords:

Reinforcement activities, distance education, recommenders, natural language processing.