- Le plus récent
- Le plus de votes
- La plupart des commentaires
According to the Amazon Personalize documentation "In the absence of sufficient user behavior data for an item, this recipe recommends popular items.". You could change the popularity_discount_factor hyperparameter close to 1 to push the model to pay less attention to an items' popularity during calculation of similarity.
I'm having the exact same issue. SIMS is only recommending the most popular items for that recently added items (i.e. new items with, I presume, a very small number of interactions). That behavior is not much sense to me. Using HPO does not solve the issue, and adjusting popularity_discount_factor neither. Also, tuning the rest of the solution hyperparamenters changes nothing for that new items getting only popularity items as recommendations. Please, if you have any suggestion it will be nice to hear. I've used SIMS in other projects without problems.
Contenus pertinents
- demandé il y a un an
- demandé il y a 13 jours
- demandé il y a un an
- AWS OFFICIELA mis à jour il y a 2 ans
- AWS OFFICIELA mis à jour il y a 2 ans
- AWS OFFICIELA mis à jour il y a 2 ans