沙海潮1，馮 鈞1，杭婷婷1，成建國2

(1. 河海大學計算機與信息學院，江蘇 南京 210098;2. 水利部信息中心，北京 100053)

　　摘 要：針對突發水事件的各類數據信息對分析事件起因、發展和后續影響有著相當重要的價值。在研究應用現狀分析的基礎上，利用事件熱度與Bi-Lstm情感分析2個模型，對2021年嘉陵江鉈污染突發事件進行分析，分析結果說明：事件熱度方面，華東、西北地區的公眾較為活躍，微博熱度具有聚集性和滯后性的特點，新聞平臺則體現出熱度的延續性特點;事件輿情方面，微博輿情表現出比新聞平臺輿情更為負面的情緒，兩者都缺乏相對理性客觀的評論。從后續影響角度看，部分氣象水文因素導致鉈污染二次發生，其次疫情前后，煉鋼、冶鋅行業利潤存在大幅度波動態勢，產量受高利潤驅動而提高，一定程度誘發此類污染事件。突發涉水事件輿情傳播迅速，影響范圍較廣，事件發展程度與輿情走勢息息相關。

　　關鍵詞：突發水事件;輿情;大數據分析;情感分析;事件熱度分析;后續影響

　　Analysis on public sentiment and subsequent impact of water-related emergency——A case study of thallium pollution in Jialing River

　　SHA Haichao1, FENG Jun1, HANG Tingting1, CHENG Jianguo2

　　(1. College of Computer and Information, Hohai University, Nanjing 210098, China;

　　2. Information Center, Ministry of Water Resources, Beijing 100053, China)

　　Abstract: All kinds of data and information for water-related emergencies are of great value to analyze the cause, development and impact of the events. Based on the analysis of the research and application status, the event heat model and Bi-LSTM sentiment analysis model are used to comprehensively analyze the Jialing River thallium pollution emergency in 2021. Results show that: in terms of event popularity, the public in East China and Northwest China are more active, the popularity of microblog has the characteristics of aggregation and lag, and the news platform reflects the continuity of the popularity. In terms of event public opinion, microblog public opinion shows more negative emotions than news platform public opinion, both of which lack relatively rational and objective comments. From the perspective of the follow-up impact of public opinion, some meteorological and hydrological factors lead to the secondary occurrence of thallium pollution. Before and after the epidemic, the profits of steel-making and zinc smelting industries are fluctuated greatly, and the output is driven by high profits, which, in a certain degree, induces such pollution events. Public opinion of sudden water-related incidents spreads rapidly and has a wide range of influence. The development degree of incidents is closely related to the trend of public opinion.

　　Key words: water-related emergency; public opinion; big data analysis; sentiment analysis; event heat analysis;subsequent impact