ESG Sentiment Analysis and Investment Risk Early Warning Based on Big Data: A Case Study of Haidilaos Incident

This study investigates the dynamic interplay between ESG (Environmental, Social, and Governance) public sentiment and investment risk through big data analytics, using Haidilao's 2023 "man urinating in hotpot" incident as a case study. Leveraging text mining and machine learning techniquesincluding Jieba segmentation, SnowNLP sentiment analysis, LDA topic modeling, and random forest algorithmsthe research quantifies public opinion intensity, polarity, and dissemination patterns across social media platforms. Findings reveal that negative ESG incidents trigger immediate financial repercussions (e.g., a 4.2% stock price drop) but can be mitigated by rapid, technology-driven crisis management, as evidenced by Haidilao's 3.5% price rebound within a week. The study highlights the asymmetrical impact of ESG dimensions, with Environmental and Social factors exerting stronger market effects than Governance. Theoretically, it advances ESG analytics by integrating unstructured text data with standardized frameworks like GRI. Practically, it demonstrates how AI-enhanced monitoring systems reduce crisis response time by 48 hours, offering actionable insights for corporate risk mitigation and investor decision-making. Limitations, such as semantic ambiguity in sentiment classification, suggest future directions for adopting Transformer-based models and cross-industry validation.