The challenges of building good recommender systems

Rich MacManus’ article on the 5 Problems of Recommender systems struck a chord with me this week.  In particular he points out the challenge of sparse data in bootstrapping a good recommender system, and that the companies that subjectively seem to have the best engines are also those with the largest volumes of data.

One of the challenges in making a startup recommender technology compelling is how well it performs in the "cold start" situation, one of my current research interest, one approach for doing so is to exploit non-preference based associations with the recommendation content corpus, nicely exemplified by Gunawardana and Meek’s paper, Tied Boltzmann Machines for Cold Start Recommendations. 

Those of you who have known me for some time will remember that I’ve always said Boltzmann Machines’ would one day be useful!

This entry was posted in Computer and Internet. Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s