کنفرانس مهندسی برق ایران

صفحه اصلی / سی امین کنفرانس بین المللی مهندسی برق

Designing Music Recommendation System based on music Genre by using Bi-LSTM

نویسندگان :

Saman Mesghali¹ Javad Askari²

1- دانشگاه صنعتی اصفهان 2- دانشگاه صنعتی اصفهان

کلمات کلیدی :

recommender systems،music،Bi-LSTM،data analysis،content-based approach

چکیده :

Many people consider music as an integral part of their lives and use it in their daily activities. Therefore, suggesting a music track based on users' previous searches and tastes can be a pleasant experience for them. The music recommendation system also helps users filter and search for songs based on their interests, age, region, and language. To do so, processing to get to best recommendations is done based on artist style and searched words. The approaches and mechanisms applied for data analysis in recommender systems change based on their type. The goal of using recommender systems in these contexts is to encourage users to have higher application utilization. One of the latest recommender systems, which is mentionable, is the content-based approach. In these systems, data content is used instead of user scoring patterns for recommendation generation. The present study implemented a music recommendation system is using an approach based on Bi-LSTM. The “Million songs” dataset is used for performing this research. Comparison of results of the proposed method with previous works, according to the F1 score, shows an improvement of three percent.