The aging population represents one of the greatest social challenges of modern times, as it requires the development of solutions that would enable older people to live a better quality of life and lead more active and independent lives.
One of the technologies that can make a significant contribution to this goal is recommendation systems, which use user data analysis to offer personalized suggestions for activities that promote physical, cognitive, and social activity among older people.
Despite progress in this area, there is still room for improvement, namely in explaining the recommendations—users often do not know why a particular recommendation was given to them, which reduces their motivation to use the system.
To address this challenge, we developed several procedures and tools as part of this work, with the aim of testing the ability of large language models (LLMs) to generate explanations for recommendations. To evaluate the generated explanations, we developed a web application that allowed expert evaluators in the field of gerontology to assess the quality of the explanations. Each explanation was rated using a five-point Likert scale in terms of relevance and comprehensibility. The collected ratings were then statistically analyzed; we calculated the mean values, standard deviations, and inter-rater reliability to verify the quality of the evaluation.
|
