TPRG: Multi-Hop Temporal Knowledge Graph Reasoning with Temporal Path Rules Guidance This repository contains the implementation of the TPRG architectures described in the paper.
- JDK1.8
- IDEA
- Python 3.x (tested on Python 3.6)
- pycharm
- xshell
- xftp
After installing the requirements, run the following command to reproduce results for TPRG:
$ java -jar HitBest.jar -train yago1/train.txt -valid yago1/valid.txt -test yago1/test.txt -rule yago1/groundings2.txt -m 10 -n 10585 -k 300 -d 0.01 -c 0.01 -ne 10 -ge 0.025 -gr 0.025 -# 500 -skip 25
$ java -jar 1-Hit1Test.jar -train gdeltROAN/train.txt -valid gdeltROAN/valid.txt -test gdeltROAN/test.txt -rule gdeltROAN/two.txt -m 20 -n 500 -k 300 -d 0.01 -c 0.01 -ne 10 -ge 0.1 -gr 0.1 -# 500 -skip 20
$ java -jar 2-Hit13510Test.jar -train gdeltROAN/train.txt -valid gdeltROAN/valid.txt -test gdeltROAN/test.txt -rule gdeltROAN/two.txt -m 20 -n 500 -k 300 -d 0.01 -c 0.01 -ne 10 -ge 0.1 -gr 0.1 -# 500 -skip 20
Baselines | Code |
---|---|
MINERVA | link |
TTransE | link |
HyTE | link |
TA-TransE / TA-DistMult | link |
DE-TransE / DE-DistMult / DE-SimplE | link |