Extraction of causal relations based on SBEL and BERT model

Area of Science:

• Biomedical text mining
• Natural Language Processing
• Bioinformatics

Background:

• Extracting causal relations in Biological Expression Language (BEL) is complex.
• Existing methods face challenges with BEL statement complexity.

Purpose of the Study:

• To simplify BEL statement extraction using a novel intermediate form.
• To improve causal relation extraction performance using BERT.

Main Methods:

• Proposed Simplified Biological Expression Language (SBEL) as an intermediate form.
• Decomposed BEL extraction into entity relation extraction and entity function detection.
• Employed a pre-trained BERT model for improved subtask performance.

Main Results:

• Achieved state-of-the-art performance on the BioCreative-V Track 4 corpus.
• Reached F1 scores of 54.8% in Stage 2 and 30.1% in Stage 1 evaluations.
• Demonstrated the effectiveness of SBEL and BERT for BEL extraction.

Conclusions:

• The proposed SBEL approach effectively simplifies BEL extraction.
• BERT-based relation and function extraction significantly enhances performance.
• This method advances the field of biomedical text mining for causal relation extraction.