Options
"Question about collaborative filtering"
Contributor I
Hi there,
I'm new at data mining and specially the field of recommender systems.
I've a question in that field and I hope someone can help me with it.
I want to make a recommender system using collaborative filtering. As the best of my knowledge I can add attributes to collaborative filtering matrix for both items and users using KNN algorithm.
I mean, I can create additional attributes to the users matrix to get better recommendation results.
The question is: Can I add some extra attributes to the items on the user-item or item-user matrices but this time these characteristics are derived from an external source, say, another dataset.
Consider the following scenario:
A recommender system for cameras that uses a dataset for a specific website. I want to recommend to the new user the suitable camera.
For the users: During the collaborative filtering, I'll group and classify the users according to some criteria for better recommendations and will add additional attribute for that.
For the items: can I create a new attribute for the items for better recommendation, but this time that particular attribute will be calculated from another separate different dataset but for the same products?.
For example extracting additional features of the cameras from another stores. The same items "same products" will be used to calculate this attribute. But from different dataset. That will produce a general description of the items and not dependent on the users usage or previous purchases.
Can the collaborative filtering algorithm be modified to fulfill this requirement?
So sorry for my broken English.
Thanks for your time.
I'm new at data mining and specially the field of recommender systems.
I've a question in that field and I hope someone can help me with it.
I want to make a recommender system using collaborative filtering. As the best of my knowledge I can add attributes to collaborative filtering matrix for both items and users using KNN algorithm.
I mean, I can create additional attributes to the users matrix to get better recommendation results.
The question is: Can I add some extra attributes to the items on the user-item or item-user matrices but this time these characteristics are derived from an external source, say, another dataset.
Consider the following scenario:
A recommender system for cameras that uses a dataset for a specific website. I want to recommend to the new user the suitable camera.
For the users: During the collaborative filtering, I'll group and classify the users according to some criteria for better recommendations and will add additional attribute for that.
For the items: can I create a new attribute for the items for better recommendation, but this time that particular attribute will be calculated from another separate different dataset but for the same products?.
For example extracting additional features of the cameras from another stores. The same items "same products" will be used to calculate this attribute. But from different dataset. That will produce a general description of the items and not dependent on the users usage or previous purchases.
Can the collaborative filtering algorithm be modified to fulfill this requirement?
So sorry for my broken English.
Thanks for your time.
Tagged:
0
