BERT Transformer와 Deep Learning을 활용한 전이학습 효과 검증 연구 :법률상담데이터 분류문제 적용

Abstract

As AI(artificial intelligence) is actively researched, it is being applied in various fields such as natural language processing, video and voice processing. However, voices pointing out the technical limitations of deep learning are spreading, and accordingly, researches for solving the technical limitations of deep learning are being actively conducted. In this paper, BERT, well known as the pre-training model of natural language processing, was applied to the classification problem of legal counseling data to verify the effect of transfer learning. Using BERT pre-trained data, the Transformer classification model was implemented and applied to the problem of legal counseling data classification, which showed higher accuracy than the traditional machine learning algorithm.