Please use this identifier to cite or link to this item: http://hdl.handle.net/10609/120146
Title: Sistema de recomendación musical para eHealth
Author: Quijada Gomariz, Adrián
Director: Casas-Roma, Jordi  
Tutor: Parada Medina, Raúl  
Abstract: It has been observed in medical studies that listening to music, singing or playing an instrument has a positive rehabilitative effect on various neurological diseases. ParkinSons is a mobile application born as an OPENeHealth initiative between the Universitat Oberta de Catalunya, the Hospital de la Santa Creu i Sant Pau and GMV that aims to improve the quality of life of people with Parkinson's disease through personalised music, video and audio exercises. The Recommendation Systems are information filtering systems, originally intended to provide suggestions of interest to a particular user. The aim of this work is to develop a recommendation system that helps to model the relationship between the multimedia objects considered in the Parkinsons application (specifically, music) and the users to improve in the different aspects of the disease. To achieve this, the following actions are carried out: The different existing approaches in the state of the art (content-based, collaborative-filtering and hybrid) are analysed. Research is carried out on the contribution of contextual information when making recommendations. Recommendation algorithms are implemented, based on the previous analysis, with the most advanced techniques using neural networks: Autoencoder, Matrix Factorization and Factorization Machines. Finally, experiments are carried out to determine the best architecture and a comparative study between all the solutions. It is determined that the Factorization Machine technique (based on collaborative-filtering) provides the following results optimal using contextual information. As future lines of work, we propose the implementation of Factorization Machine solutions based on graphs.
Keywords: recommender system
collaborative filtering
matrix factorization
factorization machines
autoencoder
Document type: info:eu-repo/semantics/masterThesis
Issue Date: 24-Jun-2020
Publication license: http://creativecommons.org/licenses/by/3.0/es/  
Appears in Collections:Bachelor thesis, research projects, etc.

Files in This Item:
File Description SizeFormat 
TFM-AdrianQuijada.pptxPresentación TFM28,41 MBMicrosoft Powerpoint XMLView/Open
TFM_procesado_datos.ipynbCódigo desarrollado TFM207,66 kBJupyter NotebookView/Open
aquijadagTFM0620memory.pdfMemoria del TFM1,28 MBAdobe PDFThumbnail
View/Open
aquijadagTFM0620presentation.pdfPresentación en PDF del TFM2,02 MBAdobe PDFThumbnail
View/Open