Matrix factorization (recommender systems): Difference between revisions

Revision as of 19:12, 7 February 2022

Revision as of 19:12, 7 February 2022

Line 1:

{{about||the decomposition of matrices as studied in linear algebra|Matrix decomposition}}

{{Recommender systems}}(Auther Rahul Kumar Balyan IIT Patna)

'''Matrix factorization''' is a class of [[collaborative filtering]] algorithms used in [[recommender system]]s<!-- I agree that this is the description but I think the wording should be different. The scientific community as a whole wouldn't commonly know matrix factorization defined in this way -->. Matrix factorization algorithms work by decomposing the user-item interaction [[Matrix (mathematics)|matrix]] into the product of two lower dimensionality rectangular matrices.<ref name="Koren09">{{cite journal |last1=Koren |first1=Yehuda |last2=Bell |first2=Robert |last3=Volinsky |first3=Chris |title=Matrix Factorization Techniques for Recommender Systems |journal=Computer |date=August 2009 |volume=42 |issue=8 |pages=30–37 |doi=10.1109/MC.2009.263|citeseerx=10.1.1.147.8295 |s2cid=58370896 }}</ref> This family of methods became widely known during the [[Netflix prize]] challenge due to its effectiveness as reported by Simon Funk in his 2006 blog post,<ref name="Funkblog">{{cite web |last1=Funk |first1=Simon |title=Netflix Update: Try This at Home |url=http://sifter.org/~simon/journal/20061211.html}}</ref> where he shared his findings with the research community. The prediction results can be improved by assigning different regularization weights to the latent factors based on items' popularity and users' activeness.<ref>{{Cite journal|last1=ChenHung-Hsuan|last2=ChenPu|date=2019-01-09|title=Differentiating Regularization Weights -- A Simple Mechanism to Alleviate Cold Start in Recommender Systems|journal=ACM Transactions on Knowledge Discovery from Data (TKDD)|volume=13|pages=1–22|language=EN|doi=10.1145/3285954|s2cid=59337456}}</ref>