Sound Tree is a free recommender system application that serves to music streaming and downloading applications (host applications). The project is conducted with AGMLab. Based on constantly updated user logs that are received from AGMLab, we are generating track recommendations for the users of our host application.
We are adapting collaborative filtering, which is very popular among social networks. What is challenging in Sound Tree is the use of big data and distributed systems. We are aiming to give the most accurate recommendations to millions of users involved in Sound Tree.
Our data set is embedded in Neo4j graph database. When dealing with big data, graph databases are approximately 1000 times faster than relational databases. User-song relationships and performer, album, song and user nodes are stored in this non-relational database.
We are using Apache Tomcat as our HTTP server to run our Java code.
Neo4jTo learn more about our project, download the demo presentation documents.
Demo presentation as PDF