https://github.com/MaurizioFD/RecSys2019_DeepLearning_Evaluation
GitHub - MaurizioFD/RecSys2019_DeepLearning_Evaluation: This is the repository of our article published in RecSys 2019 "Are We R
This is the repository of our article published in RecSys 2019 "Are We Really Making Much Progress? A Worrying Analysis of Recent Neural Recommendation Approaches" and of several follow-u...
github.com
"Are We Really Making Much Progress? A Worrying Analysis of Recent Neural Recommendation Approaches", RecSys 2019BibTex. Full text available on ACM DL (open), ResearchGate or ArXiv, source code of our experiments and full results are available here. The slides and poster are also available.
논문의 바탕이 되는 코드들을 다 담고 있으며, MovieLens data를 비롯한 수많은 데이터에 적용가능한 추천시스템 기법들 코드를 구현해놓았다. TopPopular, CF, CBF 등의 Baseline들이 있으니 직접 따라해보기 좋다!
그런데 파이썬이 조금 서툰 분들에게는 공부가 많이 필요하다 ㅠㅠ.. 기간을 조금 길게 잡고 공부해보는 것이
좋을 것 같다.
한 파일에 모두 정리된것이 아닌 모듈별 파일이 있어서 선후관계를 잘 파악하며 공부해야 하고, csr, csc 등의 sparse matrix 표현 방식들도 미리 알아놓는 것을 추천한다.
대신 이걸 한번 따라가면 추천시스템 코딩 능력뿐만 아니라 기본적인 파이썬능력도 함께 폭풍성장할것이다!!
목록
- Random: recommends a list of random items,
- TopPop: recommends the most popular items,
- UserKNN: User-based collaborative KNN,
- ItemKNN: Item-based collaborative KNN,
- UserKNN CBF: User-based content-based KNN,
- ItemKNN CBF: Item-based content-based KNN,
- UserKNN CFCBF: User-based hybrid content-based collaborative KNN,
- ItemKNN CFCBF: Item-based hybrid content-based collaborative KNN,
- P3alpha: collaborative graph-based algorithm,
- RP3beta: collaborative graph-based algorithm with reranking,
- PureSVD: SVD decomposition of the user-item matrix,
- NMF: Non-negative matrix factorization of the user-item matrix,
- IALS: Implicit alternating least squares,
- MatrixFactorization BPR (BPRMF): machine learning based matrix factorization optimizing ranking with BPR,
- MatrixFactorization FunkSVD: machine learning based matrix factorization optimizing prediction accuracy with MSE,
- EASE_R: collaborative shallow autoencoder,
- SLIM BPR: Item-based machine learning algorithm optimizing ranking with BPR,
- SLIM ElasticNet: Item-based machine learning algorithm optimizing prediction accuracy with MSE.
다음은 좀 정신없지만 논문과 코드들을 읽어보며 파악한 것들을 나름 끄적끄적 한 것이다..
도움이 될 진 모르겠지만 정리용으로 올려본다.
