3 ms·
No need to reinvent the wheel: take a look to the Apache Mahout collaborative filtering engine - it can be used for both 'small' and 'big' data. If you have one
by seektable 8y ago
No need to reinvent the wheel: take a look to the Apache Mahout collaborative filtering engine - it can be used for both 'small' and 'big' data. If you have one-machine dataset (millions of items) I may recommend to use non-Hadoop algorithms implementation that can be used as usual java lib (in case of .NET C# port is available: https://github.com/nreco/recommender https://github.com/nreco/recommender ).
Also you can check 'Mahout in Action' book - part 1 is about recommendations and it explains everything that you need to know for building your own recommendation engine.