Could someone point to any resource for machine learning for recommendation systems?
I was thinking of trying out something myself a few days back
A classic problem statement that I thought about was movie recommendation
Let me know if you would be interested in working on this. If others are interested, they can join too
Hi @anantguptadbl, I am interested but am currently working on another project which requires a recommender system. Have you got hold of any “learning paths” for recommender system? Some resources maybe that could help me in solving the problem?
Recommender systems are very varied. The method i have used generally is
Assuming that you need to make a movie recommender
1) Find all features and their data points wherever possible
Let your imagination run wild
2) Historical selections
Do we have data for the choices that the client had made. If not, then we need to extract that information to understand the target cluster
a) Type in your favourite movie, or
b) What genre do you like,
3) Try out algorithms
You can do clustering. or
decision trees, or
For a movie recommender, I would do clustering, or decision trees
if you look for implementation of recommendation check Mahout, it is the standard if I can say so for recommendation engine, works on top of MapReduce. The architecture is interesting as well
See “Apache Mahout beyond top reduce” the reference about Mahout.
Thanks for replies everyone. Here is a link to resources which I found online.
Hope it helps somebody!
There is an excellent MOOC on Coursera (https://www.coursera.org/learn/recommender-systems) that teaches the fundamentals of Recommender Systems. Check it out.