개체명인식과 서브워드 임베딩 기술을 활용한 건강 문서 검색 시스템

Abstract

Due to the growth of the healthcare service and the global epidemic, the demand for personal health-related services is increasing, and these services are provided to consumers through Q&A systems. Most Q&A systems are keyword-based systems that allow users to search for desired content only when they know the exact words of the disease. However, users lack health knowledge, so when searching, they enter queries in abbreviations and typos, not exact names of words. This results in a drop in the performance of keyword-based Q&A systems. In this paper, we propose a Korean-based health document retrieval system that can recognize various terms of healthquestions that are entered from users exploitingnamed entity recognition and subword embeddingtechniques.