---

The Future: What music do I like?

Issues

The proposed music recommendation service will aim to be simple, flexible, and dynamic, yet also specific, robust, and quantifiable. It will use multiple approaches to attempt to generate a statistically accurate understanding of a listener's preferences.

Methods

The proposed service will be a personalized music recommendation engine that will generate personalized recommendations based on a number of direct and indirect user inputs. The service will let users easily rate tracks, albums, and artists on a simple 0-5 scale, with 0.5 increments and a "don't know" option. The user will also have the option of inputting a 0-5 confidence rating in his opinion, and the option to rate tracks, albums, and artists from 0-5 based on three criteria: accessibility, creativity, and consistency. A tags system will also be implemented, more for the purpose of giving users a quick and dirty textual understanding of music when researching new music. Furthermore, in-depth information about a user's listening habits will be collected via PC media player plugin to understand what music users listen to, when they listen to it, in what succession, and why. Although this information will be gathered and playcounts displayed as soon as the service is launched, the information will all be used as part of a long-term research project with the aim of more specifically catering and tailoring music recommendations based on time of day, mood, and other factors.