We realize a recommender system based on multi agent approach and Galois lattice, we need a benchmark to evaluate it (medical domain, CRM or other).
I congratulate you for this original idea. In my humble opinion, to evaluate your recommendation system, I need to read a little more about recommendation system based on multi-agents and concept lattice.
Would you send me some of your work for me to give an objective opinion?
thank you
Benchmarking recommender systems is a large subject because there can be many different things you can benchmark and they give different answers. The two most common benchmarks is the average number of "correctly predicted" items of a top N (e.g. 10) and the Root Mean Square error of ratings.
There was a whole session on benchmarking at the leading recommender system conference recsys this year http://recsys.acm.org/recsys14/. First tutorial session may also be of interest.
- Badges
- Science topic
- Similar topics
- Books