2022_Summer_RecSys_T11_Ah_Chu

CUAI 5th Summer Conference Project Repository
Check out Paper at
Check out Presentation at Link

👪 Teammates

• Jungjoon Kim : School of Applied Statistics, Chung-Ang University
• Yejin Kwon : School of Applied Statistics, Chung-Ang University
• Gyeongbin Bak : School of Computer Science & Engineering, Chung-Ang University
• Doyeon Lim : School of Computer Science & Engineering, Chung-Ang University

📄Abstract

Amazon Review Dataset을 이용하여 추천시스템 모델을 구현한다. 그 과정에서 자연어처리 기술을 이용하여 Meta Data의 상품 description 항목을 임베딩하여 성능을 높였다. 또한 최신 모델인 BiVAE를 이용해 성능을 높일 수 있었다.

🔧Models(Paper Link)

• BiVAE(Bilateral Variational Autoencoder)
• NCF(Neural Collaborative Filtering)
• LightFM

🔍Evaluation

MODEL	NCF	LightFM	LightFM(meta)	Matrix Factorization	Matrix Factorization(meta)	BiVAE
MAP	0.004876	0.000179	0.005266	0.016408	0.016474	0.036288
NDCG	0.011440	0.000793	0.015583	0.031498	0.032077	0.085151
PRECISION@K	0.009011	0.000772	0.013203	0.021136	0.020952	0.057221
RECALL@K	0.011217	0.000858	0.017592	0.043550	0.042821	0.058966