EUM-POIRecSys

Welcome! This repository is dedicated to the development of a general Position of Interest (POI) Recommendation System in collaboration with the EUM-Project.

How to use

• For API endpoint, check main.py
• For scoring and planning model, check scoring_models/ and planning_models/
• For datasets, check datasets/ and to download dataset from AI-HUB, follow the instruction at datasets/README.md

Roadmap

• Setup Github
• Review RS papers & Choose candidate models
• Choose evaluation method
• Make demo scoring model & test platform
• Make demo planning model & test platform
• Test three scoring model
• Test three planning model
• Conclude the result
• Make FastAPI interface
• Deploy to AWS EC2
• Model serving optimization
• Make a MLOps system

Papers & References

Github & Paper

Type	Description	Link	License
Github	Best Practices on Recommendation Systems	https://github.com/recommenders-team/recommenders/tree/main?tab=readme-ov-file	MIT
Github	파이썬을 활용한 추천 시스템 구현	https://github.com/lsjsj92/recommender_system_with_Python	NA
Github	A curated list of awesome Recommender System	https://github.com/jihoo-kim/awesome-RecSys	MIT
Github	Neural Collaborative Filtering	https://github.com/DAC-KHUPID/seoul-date-course-recommendation?tab=readme-ov-file	NA
Github	Implementation of 'Personalized POI Recommendation:Spatio-Temporal Representation Learning with Social Tie'	https://github.com/dsj96/PPR-master	MIT
Github	Implementation of 'A Diffusion model for POI recommendation'	https://github.com/Yifang-Qin/Diff-POI	Not Specified
Github	POI Paper Archive	https://github.com/hubojing/POI-Recommendation	NA
Github	Next-POI Paper Archive	https://github.com/kevin-xuan/Next-POI-Recommendation	NA
POI Paper	Translating Embeddings for Modeling Multi-Relational Data	NA	NA
POI Paper	LightGCN: Simplifying and Powering Graph Convolution Network for Recommendation	http://arxiv.org/abs/2002.02126	NA
POI Paper	Neural Collaborative Filtering	http://arxiv.org/abs/1708.05031	NA
POI Paper	Large Language Models Are Zero-Shot Rankers for Recommender Systems	http://arxiv.org/abs/2305.08845	NA
POI Paper	Large Language Models Meet Collaborative Filtering: An Efficient All-Round LLM-Based Recommender System	http://arxiv.org/abs/2404.11343	NA
POI Paper	Recent Developments in Recommender Systems: A Survey	http://arxiv.org/abs/2306.12680	NA
POI Paper	A Survey of Graph Neural Networks for Social Recommender Systems	https://doi.org/10.1145/3661821	NA
POI Paper	An Attentive Inductive Bias for Sequential Recommendation beyond the Self-Attention	http://arxiv.org/abs/2312.10325	NA
POI Paper	Neighborhood-Enhanced Supervised Contrastive Learning for Collaborative Filtering	https://doi.org/10.1109/TKDE.2023.3317068	NA
POI Paper	Recommender Systems in the Era of Large Language Models (LLMs)	http://arxiv.org/abs/2307.02046	NA
POI Paper	Factorization Machines	https://doi.org/10.1109/ICDM.2010.127	NA

Candidate Models and Test Methods

Candidates

Scoring Model

• Amazon Personalize [https://aws.amazon.com/ko/personalize/]
• BPR(Cornac/Bayesian Personalized Ranking) [https://github.com/recommenders-team/recommenders/blob/main/examples/02_model_collaborative_filtering/cornac_bpr_deep_dive.ipynb]
• TBD (Maybe NCF?)

Planning Model

• Solve as TSP using Dynamic Programming (with Held-Karp Algorithm) [https://en.wikipedia.org/wiki/Held%E2%80%93Karp_algorithm]
• TBD

Test Methods

Scoring Model

• NDCG@k [https://velog.io/@whdgnszz1/%EC%B6%94%EC%B2%9C-%EC%8B%9C%EC%8A%A4%ED%85%9C-NDCG]
• Recall@k
• Precision@k

Planning Model

• NLL (Negative Log Likelihood)
• TBD

Test Results

Scoring Model (To be updated)

Name	NDCG@k	Recall@k	Precision@k
A	1.0	1.0	1.0
B	1.0	1.0	1.0
C	1.0	1.0	1.0

Planning Model

Name	NLL	TBD
A	1.0	1.0
B	1.0	1.0
C	1.0	1.0