MetaHIN
Source code for KDD 2020 paper "Meta-learning on Heterogeneous Information Networks for Cold-start Recommendation"
Requirements
- Python 3.6.9
- PyTorch 1.4.0
- My operating system is Ubuntu 16.04.1 with one GPU (GeForce RTX) and CPU (Intel Xeon W-2133)
- Detailed requirements
Datasets
We have uploaded the original data of DBook, Movielens and Yelp in the data/ folder.
The processed data of DBook and Movielens can be downloaded from Google Drive and BaiduYun (Extraction code: ened).
The processed data of Yelp can be generate by the code data/yelp/YelpProcessor.ipynb.
Description
MetaHIN/
βββ code
β βββ main.pyοΌthe main funtion of model
β βββ Config.pyοΌconfigs for model
β βββ Evaluation.py: evaluate the performance of learned embeddings w.r.t clustering and classification
β βββ DataHelper.py: load data
β βββ EmbeddingInitializer.py: map feature and inilitize embedding tables
β βββ HeteML_new.py: update paramerters in meta-learning paradigm
β βββ MetaLeaner_new.py: the base model
βββ data
β βββ dbook
β βββ original/: the original data without any preprocess
β βββ DBookProcessor.ipynb: preprocess data
β βββ movielens
β βββ original/: the original data without any preprocess
β βββ MovielensProcessor.ipynb: preprocess data
β βββ yelp
β βββ original/: the original data without any preprocess
β βββ YelpProcessor.ipynb: preprocess data
βββ README.md
Reference
@inproceedings{lu2020meta,
title={Meta-learning on Heterogeneous Information Networks for Cold-start Recommendation},
author={Lu, Yuanfu and Fang, Yuan and Shi, Chuan},
booktitle={Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery \& Data Mining},
pages={1563--1573},
year={2020}
}