For recommender systems based on matrix factorization techniques the recommendation step scales linearly with the number of objects in the catalog. This leads to a serious bottleneck in large-scale applications that have a strict time budget and in which there may be millions of items. In this work it is developed a probabilistic model for the recommender system that exploiting some constraints allows to give high quality suggestions in sublinear time.

Constrained models for optimized recommender systems

Fraccaro, Marco
2014/2015

Abstract

For recommender systems based on matrix factorization techniques the recommendation step scales linearly with the number of objects in the catalog. This leads to a serious bottleneck in large-scale applications that have a strict time budget and in which there may be millions of items. In this work it is developed a probabilistic model for the recommender system that exploiting some constraints allows to give high quality suggestions in sublinear time.
2014-10-14
machine learning, recommender systems, bayesian inference
File in questo prodotto:
File Dimensione Formato  
Thesis_UNIPD_Fraccaro.pdf

accesso aperto

Dimensione 3.92 MB
Formato Adobe PDF
3.92 MB Adobe PDF Visualizza/Apri

The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/20.500.12608/18241