I need a recommendation engine that suggests the best products to the users based on:
So basically I need both collaborative and content-based suggestions from my engine. The engine should be able to swarm through my data-base, analyse the best suggestions for my users and return those suggestions.
Are there any such open-source (preferably, or even proprietary would do) recommendation engines available?
I've been idly playing with PredictionIO for this in the past. It works quite well, although I've never put it into production.
It's open source
It has pretty amazing documentation
It's quite flexible
It's amazingly simple to work with. This is the code sample they provide:
cli = predictionio.Client("<my key>")
cli.identify("John")
cli.record_action_on_item("view", "HackerNews" )
# predict top preferences near a specified location
r = cli.get_itemrec_topn("myEngine", 5, {"pio_latlng":[37.9, 91.2]})
You don't give a preferred language in your question, but many people have built wrappers for it in other languages. Here's one for PHP.
There's even a prebuilt Amazon EC2 image with Prediction.IO already on it, and there aren't any hourly software fees. You can also download it directly from their website.
At Unbxd, we do exactly that. Our recommendation engine works on hybrid personalization which combines the approach of content-based and collaborative filtering to provide highly relevant recommendations.
The hybrid approach guarantees better performance with fewer drawbacks of any individual personalization techniques.
You can learn more about our recommendation engine here. unbxd.com/product-recommendations
