The main idea of modern recommender systems is to recommend you items similar to your profile. Your profile is calculated by your history (your favorite movies, items bought in online shops, articles suggested by Facebook...). Sometimes it is almost ridicules, especially when you have just bought let say a hydration pack but contextual advertising proposes you other hydration packs. If you like a movie, IMDB suggests you only similar movies. You do not have an opportunity to discover something new. Do you think that is a major drawback of modern recommender systems? I would appreciate to have your opinions.
Most of the recommendation systems works based on the user profiles and their activities. The user activity history plays the major role but if you have a change in your interest instantly the recommendation system can't reflect it in the systems recommendation behavior. unless knowing anything how can you suggest recommendations, so lets know something first and then the system will suggest.
I prefer to receive unexpected results but again it is hard to distinguish relevant items from junk ones.
Actually, it is depend upon number of things, like browsing history, connect IP, sometimes MAC addresses, in case of being logged-in, depend upon user ID, and frequency of browsing and accessing particular item, then based on this behavior, RS, will reflect to that.
However, to do so, large volume of data is a need.
Liana, there is a strong area within recommender system called Complementary recommendation, which has taken special place on outfit: if you buy a shirt, rather than recommending you other shirt, you recommend pants or shoes in similar style. You can chek this KDD2015 paper by McAuley et al about "Inferring Networks of Substitutable and Complementary Products" https://cseweb.ucsd.edu/~jmcauley/pdfs/kdd15.pdf
Hi Liana, not specifically browser history, it could be user ID or the same that we have prior information regarding to that.
Then RS pointed to that and based on his domain and other users history and activity recommend.
I prefer that the system recommend items similar to my history of prefered items, but the list should be completed with new items, for examples items from the (producer/seller,....) I liked its items in the past.
And For the cold start the well-known issue of recommender systems, one can be recommended items based on (location, other preferences in other online platforms, ...).
- Badges
- Science topic
- Similar topics
- Computer Science
- Program