Skip to content

Latest commit

 

History

History
28 lines (17 loc) · 683 Bytes

File metadata and controls

28 lines (17 loc) · 683 Bytes

Introduction

this repo contains my solution to ICDM 2022 : Risk Commodities Detection on Large-Scale E-commerce Graphs

it ranked 5th for session 1 and 10th for session 2.

Key methods used in this solution are:

  • NARS
  • SCR
  • Meta Pseudo Label
  • GAMLP
  • focal loss

Setup

run pip install -r requirement.txt, dgl library is used.

Run experiment

solution for seesion 1: python main.py -c config/dgl_session1.yml

solution for session 2: python main.py -c config/dgl_session2.yml

dataset for this competition:

链接:https://pan.baidu.com/s/1gE-kNJ1fr-cFd1blnchtQw 提取码:rzdy