Can The Internet Really Recommend Books And Movies?
from the how-good-is-it? dept
For many years companies have been trying to come up with better systems to help recommend what books, music, and movies you might like to buy. In the early days of the “commercial” internet there were collaborative filtering systems like FireFly, which never really got very far. More recently, firms have started to use human recommenders to augment automated suggestions. So, despite all the failed attempts at good recommendation engines, some people are still working on the perfect recommendation system. One researcher made a very interesting point, that recommendations from computerized systems are often for something “unexpected” by the users – but recommendations from friends are often more “reminders” for things that people already thought they wanted. This is something I hadn’t heard before, but which makes sense. So, the real question, then, is which is more valuable? Sometimes I’m glad when a friend “reminds” me to buy a book I’ve been thinking about for a while, since it gives me another push to make sure I actually buy it. If a computerized system comes up with a recommendation out of the blue, I may not be at all interested (unless something about it really catches my eye). As the article points out, if this unknown recommendation doesn’t make any sense – or is clearly a “sales pitch” it’s just going to turn off shoppers from being interested.
Comments on “Can The Internet Really Recommend Books And Movies?”
No Subject Given
Didn’t Firefly get bought by Amazon? I think the Amazon reccomendation engine is based on Firefly…
I was working with a company a few years ago that had software that made reccomendations based on an analysis of the music you liked. Not song title and artist, but beats per minute, tone, bass, innstumentation, etc. The demo was very impressive, but I was never 100% convinved that the demo was not fixed. They were just starting up in late 2000 – and never could land the VC they needed to commericalize the software.
Re: No Subject Given
Firefly were bought by Microsoft. As part of the Firefly work they did stuff on digital identities, which was what Microsoft bought them for to use as the backbone of their Passport service. The original reason for the company – the recommendation work – all got dumped AFAIK.
I was impressed with the recommendation system used by NetFlix.com. A user can rate a list of movies and receive recommendations about what else they might like. The accuracy is astonishing. After I’d rated a few pages of movies, a good number of the recommendations on my list were movies I already owned.
Re: Recommendation Systems
The trouble with this is that, yes, they often do come up with things you like/own, but what is the use of that? I want a recommendation system that suggests things that I don’t know about and I have yet to come across a system that even gets close to doing this. Amazon is a good case in poitn where you buy a book and your recommendations fill up with crap you have no interest in – as an example I books about Japanese and Japan, but not travel guides, but amazon persists in giving me japan travel guides even though I keep saying “not interested”. By one SF book and you are doomed to rejecting reams of utter garbage.
Re: Re: Recommendation Systems
There are lots of good recommendation systems out there, but Amazon’s recommendation service really hasn’t impressed me at all. I suspect it’s because it hasn’t really got enough information to work with for the vast number of products they deal with. They probably don’t get many people to enter ratings data, and when people give it a go, most would give up quickly.
For a much better example of a recommender system, take a look at http://www.movielens.org/
Steel and aluminum scaffolding for construction is a very useful tool.
How’s China printing industry developed this years, who can tell us?
China based injection molding services with low costs and supeior quality