I just read how team BellKor’s Pragmatic Chaos is winning the Netflix Challenge on Wired, and I'm curious about how this kind of algorithms usually work. I know team Bellkor's solution must be an innovative one on the field.. but how does the field usually work? Is it just a really detailed database with Markov chains being run over again and again or what?
but how does the field usually work?
It's a Data Mining technique. Data Mining is used as part of Business Intelligence (Data Warehouse and such) trying to find relations and information in huge amounts of data. It's an area of computer science, dealing also with machine learning in general, e.g. pattern recognition. Automatic recommendations are got by Association Mining. An association with a high support is shown as recommendation. The k-nearest-neighbor algorithm is only one of many algorithms used by machine learning/data mining people.
If you are interested in basic theory, I recommend Data Mining: Practical Machine Learning Tools and Techniques by Ian H. Witten.
For Java there is a great machine learning package, WEKA that is able to do association mining. Ian Witten is also one of the authors of WEKA.
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With