More about Netflix Prize and recommendation

I've just found a very smart post on recommendation. It's from read/write... it starts with the Netflix Prize and reminds me of my post called Netflix Prize a new philosophy of innovation (it's here but in french, Mathieu has also a good post here & in english) and I must say that I fully agree. But the best part is about Recommandation and the difference between search and browse (extract just following).

The analysis is brilliant and takes 3 very very good examples:
-Amazon pure collaborative filtering
-Pandora pure content based
-Delicious pure library (very smart analysis of the nearly reco system from delicious, even if Google similar are another possible way to get the same kind of aswers)

And the author ends with the idea of another approach that mixes the 3 ! Neverthless, no example to illustrate it. So let me offer mine : U.[lik]

1. U.[lik] is a virtual library of your entertainement because bookmarking matters, and sharing is a huge social leverage
2. U.[lik] is content based because it enables cool browsing (please read below in red) and content based is perfect for making recommendation in the long Tail (few ratings)
3. U.[lik] is Collaborative filtering when we have enough data on items and because it's what matters for a community like people you see movies with or talk about books !

So please Alex Iskold consider taking time reviewing U.[lik] and get over the disruptive presentation. I'll be available to chat anytime if you need (we are GMT time).
PS: take a look at the widget on the left

Browsing and Recommendations

A good recommendation engine can make a difference not just for Netflix, but for any online business. This is because there are two fundamental activities online - Search and Browse. When a consumer knows exactly what she is looking for, she searches for it. But when she is not looking for anything specific, she browses. It is the browsing that holds the golden opportunity for a recommendation system, because the user is not focused on finding a specific thing - she is open to suggestions.                                                      
During browsing, the user's attention (and their money) is up for grabs. By showing the user something compelling, a web site maximizes the likelihood of a transaction. So if a web site can increase the chances of giving users good recommendations, it makes more money. Obviously this is a difficult problem, but the incentive to solve it is very big. The main approaches fall into the following categories:

• Personalized recommendation - recommend things based on the individual's past behavior
• Social recommendation - recommend things based on the past behavior of similar users
• Item recommendation - recommend things based on the thing itself
• A combination of the three approaches above

We will now explore these different approaches by looking at old-timers like Amazon and newbies like Pandora and del.icio.us.

From read/write

Comments

[Rédigé par : Alex Iskold]

Hi, thanks for your kind words. U-lik looks interesting, i think you need to work on making the UI more accessible to people who might not be techies.

Alex

[Rédigé par : Nic Brisbourne]

Hey Raphael,

Good post. I wonder if browsing breaks down into two quite different activities:

1) Killing time looking at stuff hoping to find something great, but not essential that you find anything

2) Seeing recommendations as you consume relevant content - eg news relevant to the blog you are reading, or similar music

This isn't fully thought out, but the first might be more about high volume lower relevancy and the second lower volume, higher relevancy.