ShallowBKGC: a BERT-enhanced shallow neural network model for knowledge graph completion.

Area of Science:

• Artificial Intelligence
• Data Science
• Natural Language Processing

Background:

• Knowledge graph completion (KGC) aims to infer missing links between entities.
• Knowledge graph embedding is a key technique for KGC.
• Existing methods often increase computational complexity, hindering real-time applications.

Purpose of the Study:

• To propose an efficient BERT-enhanced shallow neural network for knowledge graph completion.
• To address the computational complexity and real-time application limitations of current KGC methods.

Main Methods:

• Utilizing the pre-trained language model BERT for extracting entity text features.
• Employing an embedding layer to extract entity structure features.
• Integrating text and structure features via averaging and non-linear transformation.
• Predicting relations using multi-label modeling based on integrated entity-pair representations.

Main Results:

• The proposed ShallowBKGC model demonstrates superior performance on three benchmark datasets.
• The model effectively integrates textual and structural information for improved KGC.
• Experimental results validate the efficiency and effectiveness of the shallow network approach.

Conclusions:

• ShallowBKGC offers an effective and computationally efficient solution for knowledge graph completion.
• The integration of BERT embeddings with shallow networks provides a promising direction for KGC research.
• The model's performance suggests its suitability for real-time KGC applications.