GPT-4 ?? ??? ?? ?? ?? ?? (Zero-shot Sentiment Analysis of Movie Reviews Based on GPT-4)

?? ??? ??? ??? ??? ??? ??? ??? ??? ?? ?? ???? ???? ????, ??, ??, ????? ?? ?? ???? ??? ???? ???? ??. ?? GPT-4? ?? ??? ?? ?? ??? AI?? ????? ?? ???? ??? ???? ????? ????? ? ??? ?? ???, ?? ?? ???? ???? ?? ???? ??? ??? ???? ?? ???. AI ???? ?????? ?? ??? ???, GPT-4? ??? ???? ???, ??, ??, ??, ?? ?? ??? ??? ??? ???? ??? ? ???, ?? ????? ? ??? ????? ?? ??? ? ???? ??. ? ????? ???? ?? ?? ??? ??? ???? GPT-4? ?? ?? ??? ?? ???? ?? ?????, GPT-4? ?? ??? ???? ? ??? ???? ? ???? ?????. ???, ???, F1, AUC ???? GPT-4? XLNet ? ?? ????? ?????.

Sentiment analysis is the task of categorizing subjective information and emotional states based on personal experiences into positive or negative. It is used in various domains for targets such as reviews, articles, social media conversations, and more. Recently, there has been significant interest in large language models like GPT-4 to better align with user expectations for interactions with AI. However, objective performance results are not still sufficiently provided due to various practical constraints. While not reliable in some aspects, e.g. AI hallucination, GPT-4 has demonstrated the ability to perform various tasks such as sentence generation, summarization, classification, translation, and more at a high level. Therefore, there is a need to objectively compare and evaluate its performance in sentiment analysis. This paper compares the sentiment analysis capabilities of GPT-4 and some existing models in the field using a representative movie review dataset, together with some insights on considerations for utilizing GPT-4 in sentiment analysis. GPT-4 surpassed the existing models inclusive of XLNet in the metrics of accuracy, precision, F1 and AUC.