I've reading Paul Lamere's blog, Duke Listens!, for the last couple of months. It's the smartest blog for who is interested by recommendation ((It used to be Greg Linden (who still has some great insights) but since he has retired from hard core development of findory ... he has slowed down)).
Lately, it's one of the blog from which I open the largest number of pages from my reader (I just screen articles in it and open the page so i can bookmark it - painful process!-). Paul Lamere is a researcher @ Sun. He works on the next generation of Intelligent Music Machine (And I am happy I don't fight in the same league as him). His blog is full of insights, reviews and academic news.
Today a great piece about ilike. Ilike is a plug-in for music recommendation (Paul call them remoras !!). I don't like ilike (they have a name too close from yours ;-)) so i did not gave them a lot of my time. Paul did. And here are his findings :
Last year I tried out iLike to see how well it worked. I didn't like their 'similar artists' recommendations too much. I figured it was a cold-start problem. Since they were a new site, they didn't have enough data to make good recommendations. I tried iLike again this morning to see how well they improved after having 6 months of data. I was surprised to learn that I was getting the exact same recommendations as I got back in October. It is as if they have not updated their similarity data since they have launched. Some of these recommendations are horrendous.
Read the rest of the story
THANKS, that's precious feedback & insights.
Les commentaires récents