Severe Disease-Named Entity Recognition to Support Analysis of Korean Emergency Calls

Named entity recognition (NER) helps to extract information from emergency calls (e.g., 119 calls in South Korea), thereby enabling the analysis of emergency situations. If nouns or clauses related to severe diseases can be recognized from emergency calls, it helps in assessing the severity of the emergency situations. This paper proposes severe disease-named entity recognition (SD-NER) that can detect nouns or clauses related to severe diseases and symptoms from Korean emergency calls. This paper also demonstrates the implementation of SD-NER using bidirectional encoder representations from transformers (BERT) embeddings and cosine similarity. SD-NER supports emergency call analysis by assessing the severity of the emergency situations, thus potentially mitigating the deterioration of emergency situations.