as far as the collaborators in this model i understand we have history of how they have rated movies in the past. do we know the values of the feature vector for those movies.
if greg and alice are collaborators in the model and they have given the movie “Serious Relationships” a 5 do we know that this movie is 90% comedy and 10% action, as an example.
Or are there question marks in the feature vector.
It is a given that the features of this model are comedy and action.
Hi @helloeveryone,
In collaborative filtering, all feature vectors’ component values are obtained by training the model with the ratings. In other words, before we train it, we don’t know any feature vector.
Besides, in collaborative filtering, we can’t tell the meaning of each component in the feature vector, so even after we have trained it, we couldn’t tell from the trained vectors how much comedy a movie is.
Cheers,
Raymond
That is very helpful and appreciated. Thank you for the quick accurate response!