Original Data for Sentiment of Smoking Cessation Discussions on Social Media in the Context of EVALI: A Hybrid Machine-Learning-Based Content Analysis Approach, by Yuqi Zhang and Yingning Wang

Paper Abstract:Objective:This study explores the public discourse on Twitter regarding smoking cessation before, during, and after the EVALI outbreak, aiming to discern any changes in sentiment, topics, and content and help with contemporary smoking cessation efforts.Methods:Using snscrape, English tweets from September 1, 2018, to January 31, 2020, were collected and filtered for smoking-cessation-related keywords. Sentiments were evaluated with VADER, classifying tweets into positive, negative, or neutral. Topics were identified through Latent Semantic Analysis, and LexRank extracted representative sentences for content analysis.Results:There was a significant increase in smoking cessation discussions in September 2019, coupled with a decline in average sentiment score. The "Vaping" theme dominated, characterized by a drop in sentiment. Opinions on vaping were divided; some promoted e-cigarettes as a tool for smoking cessation while others expressed negativity, largely criticizing regulations on non-tobacco-flavored e-cigarette products.Conclusions:These findings highlight the urgency for policymakers to intervene on social media, aiming to amplify targeted smoking cessation content and improve the long-lasting less effective communication around e-cigarette policies, ensuring that public discourse is informed and constructive.

论文摘要：
研究目的：本研究针对电子烟相关肺损伤（EVALI）暴发的前、中、后期推特平台上的戒烟相关公共讨论展开探究，旨在剖析其情感倾向、讨论主题与内容的变化，助力当代戒烟工作开展。
研究方法：本研究借助snscrape工具，采集2018年9月1日至2020年1月31日期间的英文推文，并通过戒烟相关关键词进行筛选。采用VADER情感分析工具对推文进行情感评分，将其划分为积极、消极与中性三类；通过潜在语义分析（Latent Semantic Analysis）识别讨论主题，并借助LexRank算法提取代表性语句开展内容分析。
研究结果：2019年9月，戒烟相关讨论量显著上升，同时平均情感评分有所下降。"雾化（Vaping）"主题占据主导地位，其情感评分呈下降趋势。公众对雾化产品的态度存在分歧：部分人群将电子烟（e-cigarettes）作为戒烟辅助工具加以推广，而另一部分则持否定态度，主要批评针对非烟草味电子烟产品的监管政策。
研究结论：本研究结果凸显了政策制定者介入社交媒体平台的紧迫性，以期强化针对性戒烟内容的传播，并改善电子烟监管政策长期以来效果欠佳的宣传沟通工作，确保公众讨论具备科学性与建设性。