Use of recommendation models to provide support to dyslexic students

TL;DR

This work addresses the challenge of providing personalized educational support to university students with dyslexia by evaluating three collaborative-filtering recommender systems (user-based, item-based, and hybrid) across three similarity metrics. The best-performing configuration is a hybrid model with blending weight $\alpha=1/4$, using Pearson correlation and $n=3$ neighbors, achieving a mean absolute error of $MAE=0.8093$ and high top-$k$ relevance ($Precision@k=0.8524$, $Recall@k=0.8278$). A real-case study with $50$ participants demonstrates that AI-guided recommendations improve dyslexic students' performance and can close the gap with non-dyslexic peers. The findings support the feasibility of AI-driven, personalized support tools for dyslexia in higher education and point to future work with larger, multilingual datasets to broaden applicability.

Abstract

Dyslexia is the most widespread specific learning disorder and significantly impair different cognitive domains. This, in turn, negatively affects dyslexic students during their learning path. Therefore, specific support must be given to these students. In addition, such a support must be highly personalized, since the problems generated by the disorder can be very different from one to another. In this work, we explored the possibility of using AI to suggest the most suitable supporting tools for dyslexic students, so as to provide a targeted help that can be of real utility. To do this, we relied on recommendation algorithms, which are a branch of machine learning, that aim to detect personal preferences and provide the most suitable suggestions. We hence implemented and trained three collaborative-filtering recommendation models, namely an item-based, a user-based and a weighted-hybrid model, and studied their performance on a large database of 1237 students' information, collected with a self-evaluating questionnaire regarding all the most used supporting strategies and digital tools. Each recommendation model was tested with three different similarity metrics, namely Pearson correlation, Euclidean distance and Cosine similarity. The obtained results showed that a recommendation system is highly effective in suggesting the optimal help tools/strategies for everyone. This demonstrates that the proposed approach is successful and can be used as a new and effective methodology to support students with dyslexia.

Figures (4)

• Figure 1: Rating matrix representation format.
• Figure 2: User-based approach (a) discovers that Student 1 and Student 3 are similar because they have chosen some similar learning methodologies, so it recommends (dashed arrow) to the Student 3 the item that was useful to Student 1 and that the latter is not using. Item-based approach (b) finds that the two last items have been rating similarly by different users (Student 1 and 2), so since Student 3 also likes one of these objects, it recommends the other one to the Student 3.
• Figure 3: Best MAE for the different hybrid RS weight configurations.
• Figure 4: Comparison among different weight-pairs of the hybrid approach according MAE.