This repository contains codes and datasets to reproduce results of the Master Thesis EVALUATING HYBRID AI FOR PREDICTION OVER LUNG CANCER KNOWLEDGE GRAPHS.
Implementations for conducting link prediction using rules generated by AMIE.
- Use repository (https://github.com/dig-team/amie to mine rules) to mine rules.
- run java command "java -jar amie3.jar -const benchmark_train.tsv".
- Run "main.py --benchmark" to get right linke prediction result.
- Cloned October 2023.
- Implementations for ConvE.
- Original repository: (https://github.com/TimDettmers/ConvE).
- Test files in benchmark_Test_Rel_Type folder is used to evaluate different relation types.
- Use ConvE.ipynb to get right link prediction result.
- Cloned October 2023.
- Implementations for TransE, TransH, TransR, TransD, RotatE, DistMult, and ComplEx.
- Original repository: (https://github.com/thunlp/OpenKE).
- Test.h, Tester.py is modified to get result for right link prediction.
- Test files in Test_Rel_Type folder for each benchmark is used to evaluate different relation types.
- Use OpenKE.ipynb to get right link prediction result.
- Cloned October 2023.
- Implementations for TuckER.
- Original repository: (https://github.com/ibalazevic/TuckER).
- Modified main.py to get result for right link prediction.
- Test files in Test_Rel_Type folder for each benchmark is used to evaluate different relation types.
- Use TuckER.ipynb to get right link prediction result.