Groarke, Michele (2014) Analysis of Million Song Dataset to Create an Accurate Recommender System. Diploma thesis, Dublin, National College of Ireland.
Download (2MB) | Preview
This study proposes an alternative music recommender system which utilises users’ feelings about songs in tailoring a music recommender system. As an alternative to current recommender systems which use a combination of collaborative and content-based filtering to predict a user’s enjoyment of a song/album/artist, this project proposes that the contextual element of a users’ feelings about a sample of songs taken from the Million Song Dataset (MSD) can be used to make more accurate predictions regarding a user’s enjoyment of proposed tracks. In addition to ‘feelings’, other factors were taken into consideration , namely the users’ age, gender and whether they are influenced by lyrics, rhythm or both.
In current systems, there is a ‘cold-start’ problem which occurs when a new user has not provided enough input to tailor a profile for them. In addition, some current music recommender systems will recommend according to artist, genre or album which can be unreliable due to the non-homogeneity of artists and albums and the difficulty of assigning a genre to every song.
The methods used in this study were to perform clustering analyses (k-Means) on the survey results and on the attributes of the MSD Subset. Then the k-NN classification technique was applied to the data.
Results of initial analysis indicate that further development is necessary to obtain the desired level of accuracy in recommendations.
Post dissertation work to be carried out includes testing and finalisation of resulting recommendations to the survey participants. The results will be reported to participants upon completion of testing and analysis.
|Item Type:||Thesis (Diploma)|
|Subjects:||Q Science > QA Mathematics > Electronic computers. Computer science
T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science
|Divisions:||School of Computing > Higher Diploma in Science in Data Analytics|
|Depositing User:||CAOIMHE NI MHAICIN|
|Date Deposited:||15 Dec 2014 15:50|
|Last Modified:||15 Dec 2014 15:51|
Actions (login required)