/UniKER

The official Pytorch implementation of "UniKER: A Unified Framework for Combining Embedding and Definite Horn Rule Reasoning for Knowledge Graph Inference"

Primary LanguagePythonMIT LicenseMIT

UniKER

Introduction

The official Pytorch implementation of the paper UniKER: A Unified Framework for Combining Embedding and Definite Horn Rule Reasoning for Knowledge Graph Inference

Supported Models:

  • RotatE, pRotatE, TransE, ComplEx, DistMult

Evaluation Metrics:

  • MRR, MR, HITS@1, HITS@3, HITS@10 (filtered)

KG Data:

  • entities.dict: a dictionary map entities to unique ids
  • relations.dict: a dictionary map relations to unique ids
  • train.txt: the KGE model is trained to fit this data set
  • valid.txt: create a blank file if no validation data is available
  • test.txt: the KGE model is evaluated on this data set
  • MLN_rule.txt: the logical rules with the format "weight \t head_rel \t body1_rel \t body2_rel"

Usage

For example, this command train a UniKER on family dataset with the scoring function defined following TransE with GPU 0.

  python run.py family 0 family_model TransE 8 0.0 0.2

Each parameter means:

  python run.py DATASET CUDA SAVE_MODEL_NAME BASIC_KGE_MODEL INTER NOISE_THRESHOLD TOP_K_THRESHOLD IS_INIT