Akademik Məlumat İdarəetmə Sistemi
PROCEEDINGS OF AZERBAIJAN HIGH TECHNICAL EDUCATIONAL INSTITUTIONS, vol.50, no.2674-5224, pp.914-922, 2025 (Non Peer-Reviewed Journal)
This article explores the application of genetic algorithms in recommendation systems. Various
methods are used to provide personalized product recommendations to users on e-commerce
platforms. While traditional content-based and collaborative filtering recommendation systems
are effective, they may not always yield optimal results, particularly due to the cold start problem
and personalization limitations.[1] This study investigates how genetic algorithms can be utilized
to overcome these challenges. Using the MovieLens Tag Genome dataset, a genetic algorithm was
applied to recommend lesser-known yet relevant movies to users. The algorithm was optimized
through mutation, crossover, and selection operators.
Conventional recommendation methods rely on users' past interactions or similarities with other
users to generate recommendations. While effective, these approaches struggle with new users
and less popular items, leading to suboptimal results. This study demonstrates that genetic
algorithms can enhance recommendation systems by making them more dynamic and adaptive.
By leveraging evolutionary principles, genetic algorithms optimize individual preferences over a
broader spectrum, providing more refined and personalized recommendations.
The results demonstrate that, unlike traditional methods, genetic algorithms can recommend less
popular movies while still providing personalized recommendations. This method enhances
recommendation system effectiveness by offering a more tai