以文字探勘方法檢視社群網站使用者之態度與意見：以流感疫苗為例
No Thumbnail Available
Date
2022
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
從1918年流感大流行後，流感病毒始終無法徹底絕跡。在每年的冬、春兩季通常會到達疫情之高峰。雖然安全有效的流感疫苗已經施打了超過60年，台灣每年也會採購公費流感疫苗提供九類人施打，並且積極宣傳施打流感疫苗之重要性，但是不少家長仍因畏懼流感疫苗之副作用而不願讓子女施打流感疫苗。因此本研究為幫助政府或相關單位，能準確瞭解社群網站對於流感疫苗之看法與想法，更加精確地擬定政策，所以本研究藉由分析社群網站之相關文章，釐清流感疫苗的討論主題、內容以及情緒等，對社群網站使用者的輿論做更多的解釋與揭露。隨著社群網站的興盛，人們接收資訊與相互討論的管道也從電視、新聞等傳統媒體逐漸轉移至網路。也因此，本研究使用有別於傳統問卷調查的自動化文字探勘與情緒分析進行社群網路文本的用詞與情緒分析。研究之文本取自社群網站PTT「BabyMother」板，從2016～2021年共260篇文章、12,174則留言。根據研究結果，在社群網站PTT所討論的流感疫苗文章大致可分為五種主題，討論度最高的主題為「施打流感疫苗之考量」；不同主題以及文章和留言之議題皆有差異，例如在「流感疫苗副作用討論」的文章中「媽媽」一詞出現頻率較高，在留言中則否，顯示留言者相較於成人對幼兒之關注程度更高；在情緒分析方面則是顯示文章之情緒大致為中立，留言之情緒則較為豐富，尤其在「流感疫苗副作用討論」中使用者之負面情緒比例為五個主題中最多。根據以上研究結果，本研究提出相應的建議給醫療政策擬定者、幼兒照護者與未來可行之研究：一、對醫療政策擬定者的建議：醫療政策之擬定人員能多製作流感疫苗副作用與國光疫苗之衛教宣導，以降低幼兒照護者對流感疫苗的擔憂。此外，亦能藉由公眾人物或是在網路中具影響能力的人物於社群網站進行宣導與正面的衛教，以提升幼兒照護者對於流感疫苗的信任感。二、對幼兒照護者的建議：幼兒照護者在社群網站找尋網路流感疫苗相關資訊時，可至相關闢謠專區進行資訊可信度的確認。如欲提供資訊者，應先確實查證資訊的正確性，以避免傳播之資訊為假資訊。三、未來研究建議：（一）使用不同社群網站之文章進行更多元的分析；（二）使用問卷或訪談的方式更深入探討社群網站使用者對流感疫苗之看法；（三）利用編碼更深入研究使用者在社群網站之發文與留言之差異性。
Since the 1918 influenza pandemic, the influenza virus has never been completely eliminated. The peak of the epidemic is usually reached in winter and spring every year. Although safe and effective influenza vaccines have been administered for more than 60 years, Taiwan also purchases public-funded influenza vaccines every year to provide nine types of people to vaccinate, and actively promotes the importance of influenza vaccination, but many parents are still reluctance to have their children vaccinated against the flu because of side effects. Therefore, in order to help the government or related units, this study can accurately understand the views and thoughts on influenza vaccines on social media, and formulate policies more accurately. This study uses the technique of text mining to analysis content and emotions on social media, to explain and expose the public opinion.With the prosperity of social networking sites, the channels for people to receive information and discuss with each other have gradually shifted from traditional media such as TV and news to the Internet. Therefore, this study uses text mining and sentiment analysis, which is different from traditional questionnaires, to analyze the content and sentiment of social network texts. This study takes the social website PTT"BabyMother" board as the research object, extracts 260 articles and 12,174 comments about influenza vaccine from 2016 to 2021. According to the research results, the influenza vaccine articles discussed on the social networking site PTT can be roughly divided into five topics, and the most discussed topic is "Considerations of Influenza Vaccination"; different topics and the terms used in articles and comments are different. For example, the word "mother" appears more frequently in the articleon "Influenza Vaccine Side Effects Discussion", but not in comments, indicating that the commenters are more concerned about children than adults; in sentiment analysis. The sentiment of the article is generally neutral, while the sentiment of the comments is more abundant, especially in the "Influenza Vaccine Side Effects Discussion", the proportion of users' negative emotions is the highest among the five topics. Based on the above findings, this study proposes suggestions for medical and health organizations, parents and future researches:1. Recommendations for medical policy planners: Medical policy planners can make more health education publicity of flu vaccine side effects and AdimFlu-S vaccine to reduce the concerns of parents about flu vaccines. In addition, public figures or influential figures on the Internet can also conduct publicity and positive health education on social media, so as to enhance the trust of parents in influenza vaccines.2. Recommendations for parents: When looking for information about influenza vaccine online on social networking sites, parents can go to the relevant rumor-refuting section to confirm the credibility of the information. Those who want to provide information should verify the correctness of the information to avoid dissemination of fake information.3. Suggestions for future research: (1) Use content from different social media for more meta-analysis; (2) Use questionnaires or interviews to delve deeper into social media users' perceptions of influenza vaccine; (3) Use coding to further study the differences between users' posts and comments on social media.
Since the 1918 influenza pandemic, the influenza virus has never been completely eliminated. The peak of the epidemic is usually reached in winter and spring every year. Although safe and effective influenza vaccines have been administered for more than 60 years, Taiwan also purchases public-funded influenza vaccines every year to provide nine types of people to vaccinate, and actively promotes the importance of influenza vaccination, but many parents are still reluctance to have their children vaccinated against the flu because of side effects. Therefore, in order to help the government or related units, this study can accurately understand the views and thoughts on influenza vaccines on social media, and formulate policies more accurately. This study uses the technique of text mining to analysis content and emotions on social media, to explain and expose the public opinion.With the prosperity of social networking sites, the channels for people to receive information and discuss with each other have gradually shifted from traditional media such as TV and news to the Internet. Therefore, this study uses text mining and sentiment analysis, which is different from traditional questionnaires, to analyze the content and sentiment of social network texts. This study takes the social website PTT"BabyMother" board as the research object, extracts 260 articles and 12,174 comments about influenza vaccine from 2016 to 2021. According to the research results, the influenza vaccine articles discussed on the social networking site PTT can be roughly divided into five topics, and the most discussed topic is "Considerations of Influenza Vaccination"; different topics and the terms used in articles and comments are different. For example, the word "mother" appears more frequently in the articleon "Influenza Vaccine Side Effects Discussion", but not in comments, indicating that the commenters are more concerned about children than adults; in sentiment analysis. The sentiment of the article is generally neutral, while the sentiment of the comments is more abundant, especially in the "Influenza Vaccine Side Effects Discussion", the proportion of users' negative emotions is the highest among the five topics. Based on the above findings, this study proposes suggestions for medical and health organizations, parents and future researches:1. Recommendations for medical policy planners: Medical policy planners can make more health education publicity of flu vaccine side effects and AdimFlu-S vaccine to reduce the concerns of parents about flu vaccines. In addition, public figures or influential figures on the Internet can also conduct publicity and positive health education on social media, so as to enhance the trust of parents in influenza vaccines.2. Recommendations for parents: When looking for information about influenza vaccine online on social networking sites, parents can go to the relevant rumor-refuting section to confirm the credibility of the information. Those who want to provide information should verify the correctness of the information to avoid dissemination of fake information.3. Suggestions for future research: (1) Use content from different social media for more meta-analysis; (2) Use questionnaires or interviews to delve deeper into social media users' perceptions of influenza vaccine; (3) Use coding to further study the differences between users' posts and comments on social media.
Description
Keywords
社群網站, LDA, 詞嵌入, 情緒分析, Latent Dirichlet allocation, Word Embedding, Sentiment analysis, Social Media