kbogas / mythesis

My Thesis Knowledge Graph Embedding for Link Prediction

My thesis

My Thesis Knowledge Graph Embedding for Link Prediction

Contributor:

• Phan Minh Tâm - [e-mail]
• Hoàng Minh Thanh - [e-mail]

Our Contribution:

Our thesis - vietnamese version

english version update later(submiting)

We Contribution in rule base AnyBURL:

• strategy offline-to-online adding knowledge into graph.
• strategy online-to-online adding knowledge into graph.

We Contribution in deep learning method:

• apply Attention Mechanism into knowledge graph.
• implement and evaluation KBGAT.

Our result

	FB15K		FB15K-237		WN18		WN18RR
	H@1	H@10	H@1	H@10	H@1	H@10	H@1	H@10
AnyBURL	79.13	82.30	20.85	42.40	93.96	95.07	44.22	54.40
CGAT	*	*	36.06	58.32	*	*	35.12	57.01

source-anyburl

• fully testing 4 data set fully_testing.ipynb
• learn with strategy offline-to-online adding knowledge into graph. learning_extend_rule.ipynb
• strategy online-to-online adding knowledge into graph. learning_extend_rule.ipynb

source-CGAT

• FB15k_237.ipynb
• GCAT_FB15k.ipynb
• GCAT_WN18.ipynb
• GCAT_WN18RR.ipynb