TGACN

This repo provides a reference implementation of TGACN as described in paper "Link-aware link prediction over temporal graph by pattern recognition"

Before you run the following command,

1. download the datasets and let this folder under the root directory of this project. Name this folder as "data".
2. replace "pro_path" in utils/args.py as your own project path.

python  main.py  --data uci  --gpu 1   --recent 6  --para 0 --patience 3  --model 0   --trace_step 35 --use_timec;  
python  main.py  --data social  --gpu 1   --recent 6   --para 0  --patience 3   --model 0   --trace_step 35 --use_timec; 
python  main.py  --data enron  --gpu 1 --recent 5  --para 4   --patience 3  --model 0   --trace_step 35   --use_timec;
python  main.py  --data wikipedia  --gpu 1   --recent 5  --para 4 --patience 3  --model 0   --trace_step 35 --use_timec; 
python  main.py  --data lastfm  --gpu 1  --recent 6  --para 3  --patience 3  --model 0  --trace_step 35    --use_timec;  
python  main.py  --data mooc  --gpu 1  --recent 9 --para 0 --use_timee --use_timec --dropout 0.07  --patience 3  --model 0  --trace_step 35