NEU-IDKE / ConvEICF

Knwoledge Graph Completion

Multi-aspect Enhanced Convolutional Neural Networks for Knowledge Graph Completion

This is the code of paper Multi-aspect Enhanced Convolutional Neural Networks for Knowledge Graph Completion.

Requirements

• python 3.8
• torch 1.9
• dgl 0.8

Datasets

Available datasets are:

FB15k-237
WN18RR

Reproduce the Results

To run a model execute the following command :

• FB15k-237 python run.py --data FB15k-237 --th1 0.3 --th2 0.2 --conve_hid_drop 0.2 --feat_drop 0.2 --input_drop 0.2 --num_filt 200 --x_ops p.b.d --temperature 0.007
• WN18RR python run.py --data wn18rr --th1 0.2 --th2 0.1 --conve_hid_drop 0.4 --feat_drop 0.1 --input_drop 0.2 --num_filt 250 --x_ops p.b.d --r_ops p.b.d --temperature 0.001

Acknowledgement

We refer to the code of LTE and DGL. Thanks for their contributions.