本文利用自然语言处理技术，通过更为精细的机器处理技术以及更大样本量以期捕捉与准确识别媒体报道语气语调，在变量刻度予以厘清的基础上，考察了媒体报道语气语调与IPO抑价之间的因果效应和内在作用机制。本文的研究背景来源于A股市场特有的打新热，其背后反映了上市公司IPO普遍的高抑价现象。实证分析部分，本文基于2018年-2021年沪深主板及创业板上市的样本公司进行实证检验，创新性地借助大语言模型ChatGPT进行情感分析，构建能有效反映新闻媒体报道语气的解释变量，并检验其对首发日抑价水平的影响。实证结果表明，IPO抑价水平与新闻媒体报道的正面语气强度之间存在显著的正相关关系，与负面语气强度存在显著的负相关关系。此外，本文对模型进行中介效应和交互效应验证，发现投资者情绪在该影响机制中发挥的作用，媒体报道所呈现的情绪影响了投资者的情绪，高涨的情绪推动了股价的上升，进而加深了抑价程度，相反较为消极的语气语调对投资者带来了消极情绪，从而作用于二级市场股票价格波动。交互效应验证说明，投资者情绪可能还反过来对新闻媒体报道语气有调节作用，进而影响了IPO抑价水平。最后，本文针对上述结论分别对上市公司、投资人、媒体中介和监管部门分别提出了相应建议。

This article employs natural language processing technology to capture and accurately identify the tone of media reports through more sophisticated machine processing techniques and a larger sample size. On the basis of clarifying the scale of variables, it examines the causal effect and internal mechanism between media tone and IPO underpricing.The research background of this paper stems from the unique enthusiasm for new stock issues in the A-share market, which reflects the widespread phenomenon of high underpricing in listed company IPOs. In the empirical analysis section, the paper conducts empirical tests based on sample companies listed on the Shanghai and Shenzhen main boards and the ChiNext board from 2018 to 2021. It innovatively uses the large language model ChatGPT for sentiment analysis, constructs an explanatory variable that effectively reflects the tone of media reports, and tests its impact on the underpricing level on the first day of listing. The empirical results show that there is a significant positive correlation between the level of IPO underpricing and the intensity of positive media tone, and a significant negative correlation with the intensity of negative tone. In addition, this paper verifies the mediating effect and interaction effect of the model and finds that investor sentiment plays a role in this mechanism of influence. The emotions presented in media reports affect the emotions of investors. High emotions drive the rise in stock prices, thereby deepening the degree of underpricing. Conversely, a more negative tone and mood bring negative emotions to investors, thus affecting the fluctuation of stock prices in the secondary market. The verification of interaction effects indicates that investor sentiment may also have a regulatory effect on the tone of media reports, thereby affecting the level of IPO underpricing.In conclusion, the study offers a series of targeted recommendations tailored for listed companies, investors, media intermediaries, and regulatory authorities, based on the insights and conclusions drawn from the research. These suggestions are designed to foster a more transparent, balanced, and informed market environment, ultimately contributing to the maturation and stability of the financial ecosystem.