|
|
|
|
|
|
by monkeybutton
1528 days ago
|
|
|
Academia thrives on open benchmark that researchers compete against each other to get the highest score on. How would you replicate that with a recommendation system where in industry you test recommendations by variants A/B/... and observe the winner? Your benchmark is no longer static and you rely on users' feedback for the recommendations each algorithm made. |
|
|