RecSys 07: Review for Trust-aware recommender systems
Recommendation system paper challenge (1/50)
Recommendation system paper challenge (1/50)
Why I write this blog?
The main reason is that I want to take some notes so that in the future, I can quickly recall it and do have to add so many favorite website for that.
Why this paper?
RecSys’07 and it has high citation.
They apply trust information to improve the performance of cold-start problem and discuss the trade-off utilizing distance of trust.
Moreover, they discuss local trust matrix and global trust matrix.
The data?
Epinions.com Dataset
The problem: Rating problem
MAE, MAUE, ratings coverage, users coverage
the based model
In this paper, they consider w as not only similarity weight but also trust information to deal with cold-start problem.
MT1: directly trust, trust propagation distance 1
MT2: trust propagation distance 2
MT3: trust propagation distance 3
MT4: trust propagation distance 4
Above the figure about coverage, we can clearly see the model with more trust propagation can handle cold-user more easier.
However, with too high trust propagation distance, it also hurts the accuracy. It is obvious because too far away user from trust does not mean they also share with similar taste with you.
Related:
Review for Performance of recommender algorithms on top-n recommendation tasks
Best paper in RecSys:
https://recsys.acm.org/best-papers/
My website: