Graph Embedding-Based Domain-Specific Knowledge Graph Expansion Using Research Literature Summary

Knowledge bases built in the knowledge processing field have a problem that experts to add rules or update them through modifications. To solve this problem, research has been conducted on graph expansion methods using deep learning technology, and recent years, many studies of generating by embedding graph’s triple information continuous vector space. In paper, literature summary, we propose domain-specific method based embedding. end, perform pre-processing process text summarization with collected data. Furthermore, extracting entity relation expanding web summarize Bidirectional Encoder Representations from Transformers for Summarization (BERTSUM) model data design Research-BERT (RE-BERT) extracts entities information, which are components graph, summarized literature. Moreover, proposed related Google news after generated graph. experiment, measured performance summarizing BERTSUM accuracy extraction model. experiment removing unnecessary sentences key sentences, result shows Classifier model’s ROUGE-1 precision is 57.86%. The was mean reciprocal rank (MRR), (MR), HIT@N rank-based evaluation metric. showed superior terms speed quality.