Recommender Systems (RSs) are software tools and techniques providing suggestions for items to be of use to a user.
RSs are primarily directed towards individuals who lack sufficient personal experience or competence to evaluate the potentially overwhelming number of alternative items that a Web site, for example, may offer.
- https://github.com/hongleizhang/RSPapers
- https://github.com/YuyangZhangFTD/awesome-RecSys-papers
- https://github.com/daicoolb/RecommenderSystem-Paper
- https://github.com/grahamjenson/list_of_recommender_systems
- https://github.com/benfred/implicit
If we have collected user
Matrix completion is to complete the matrix
SVD is to factorize the matrix into the multiplication of matrices so that $$ \hat{R}=P^{T}Q $$
And we can predict the score
$$\hat{R}{[u][i]}=\hat{r}{u,i}=\left<P_u,Q_i\right>=\sum_f p_{u,f}q_{i,f}$$
where
$$C(P,Q) = \sum_{(u,i):Observed}(r_{u,i}-\hat{r}{u,i})^{2}=\sum{(u,i):Observed}(r_{u,i}-\sum_f p_{u,f}q_{i,f})^{2}.$$
Additionally, we can add regular term into the cost function to void over-fitting
- https://www.zhihu.com/question/47716840/answer/110843844
- https://www.cnblogs.com/Xnice/p/4522671.html
- https://blog.csdn.net/turing365/article/details/80544594
- https://en.wikipedia.org/wiki/Collaborative_filtering
- https://bugra.github.io/work/notes/2014-04-19/alternating-least-squares-method-for-collaborative-filtering/
Deep learning is powerful in processing visual and text information.