말뭉치 기반 비윤리 문장의 의미 연구 - 부사어와 서술어 구성을 중심으로 = Corpus-based study of the meaning of unethical sentences - focusing on adverb-predicate structure-

다국어 초록 (Multilingual Abstract)

.

다국어 초록 (Multilingual Abstract)

This study aimed to examine the meaning aspect of unethical sentences in interactive AI learning data built as a way of an interface for humans and AI to naturally communicate. In the course of the discussion, in order for AI to communicate ethically correctly with humans, the need for a filtering process to detect and remove language expressions with unethical meanings was raised. To this end, the discussion focused on the unethical meaning of the 'adverb-predicate' relationship as a structure that represents unethically in unethical sentences.
As a result of the discussion, it was found that the unethical meaning stems from the lexical properties of adverbs and predicates, that is, contextual meaning, but the cotext, an external factor in which adverbs and predicates are realized, can have an influence. In order for AI to properly grasp the contextual information, it is possible only when the information that the domain has specified. The information held by the domain is a basic condition for AI to determine the meaning of unethical expression in the process of acquiring learning data.
Only when AI accurately recognizes unethical expressions and realizes communication with humans can true communication between humans and AI be achieved. The fundamental question of how far AI and humans will strive to communicate is only the beginning. However, in order for humans to coexist with mechanical processes, research on the meaning of unethical sentences will be necessary.