目前各國政府多透過網頁或聊天機器人進行風險溝通，系統的優化對於提高公眾風險認知及政府信任至關重要。新冠肺炎疫情期間，及時準確向公眾傳遞資訊，以便民眾做出相應的防疫行為，尤其仰賴網頁或聊天機器人作為工具。然而，系統是否完整涵括民眾疫情問答的重要資訊面向，成為提供精確資訊、減少公眾對於資訊誤解之關鍵。
本研究以風險溝通理論支持的五個訊息維度變項，分別為嚴重性、脆弱性、防疫措施、自我效能和韌性，分析防疫成效良好的臺灣情形，利用臺灣中央和地方政府的聊天機器人分類和網站問答集資料，並同樣以訊息提供有關的五項變項分析，指出國外和臺灣政府疫情資訊在脆弱性、自我效能與韌性的資訊面向都比較缺乏。而臺灣中央和地方政府之間相互補充之情況，但與政府的網站的訊息變相分布相比，聊天機器人分類和網站問答集資料的分布表現沒有網站好，顯示在風險資訊上，目前聊天機器人分類和網站問答集資料仍有改進空間。
同時，我們藉由分布紀錄、政府新聞稿、歷史事件和訪問等方式，分析訊息變項與潛在人為因素的關係。臺灣的訊息分布可能層級、溝通管道、產業和制定者的因素影響，而各政府的預算尚未證實與聊天機器人的設置有相關。研究結果在於協助盤點目前風險資訊系統上缺乏的面向，期盼能為未來政府再次面臨傳染性疾病時，貢獻風險溝通和資訊傳播效果。

Governments globally employ websites and chatbots for risk communication, crucial for public awareness and trust. In the COVID-19 pandemic, timely and accurate information is vital for guiding response efficacy. However, these systems often lack comprehensiveness in addressing key pandemic-related queries. This study, using risk communication theory, examines five message dimensions (severity, vulnerability, response efficacy, self-efficacy, and resilience). Taiwan's government chatbots and Q&A sets are analyzed, revealing insufficient emphasis on vulnerability, self-efficacy, and resilience. While Taiwan's central and local governments show synergy, chatbot and Q&A classification lags behind websites in risk information distribution. The study explores the link between message variables and human factors, including hierarchy and communication channels, in Taiwan's message distribution. The impact of government budgets on chatbot establishment remains inconclusive. This research identifies gaps in the current risk information system, aiming to enhance risk communication during future outbreaks.

[1] Abrams, E. M., & Greenhawt, M. (2020). Risk communication during COVID-19. The Journal of Allergy and Clinical Immunology: In Practice, 8(6), 1791-1794.
[2] Amiri, P., & Karahanna, E. (2022). Chatbot use cases in the Covid-19 public health response. Journal of the American Medical Informatics Association, 29(5), 1000-1010.
[3] Ara, J., & Sik-Lanyi, C. (2022). Investigation of Covid-19 vaccine information websites across Europe and Asia using automated accessibility protocols. International Journal of Environmental Research and Public Health, 19(5), 2867.
[4] Arora, A., Shrivastava, A., Mohit, M., Lecanda, L. S. M., & Aly, A. (2020). Cross-lingual transfer learning for intent detection of covid-19 utterances.
[5] Bowker, G. C., & Star, S. L. (2000). Sorting things out: Classification and its consequences. MIT press.
[6] Burton, J. (2009). A Marketer’s Guide to Understanding the Economics of Digital Compared to Traditional Advertising and Media Services, New York: American Association of Advertising Agencies, Inc.
[7] Chesser A, Drassen Ham A, Keene Woods N. (2020) Assessment of COVID-19 Knowledge Among University Students: Implications for Future Risk Communication Strategies. Health Education & Behavior. 2020;47(4):540-543. doi:10.1177/1090198120931420
[8] Crawford, K. (2021). Atlas of Ai: Power, politics, and the planetary costs of Artificial Intelligence. Yale University Press.
[9] Cortizo, J. C., Carrero, F. M., & Gómez, J. M. (2011). Introduction to the special issue: Mining social media. International Journal of Electronic Commerce, 15(3), 5-8.
[10] Demirtaş-Madran, H. A. (2021). Accepting restrictions and compliance with recommended preventive behaviors for COVID-19: a discussion based on the key approaches and current research on fear appeals. Frontiers in Psychology, 12, 558437.
[11] Ding, K., Han, H., Li, L., & Yi, M. (2021, July). Research on question answering system for COVID-19 based on knowledge graph. In 2021 40th Chinese Control Conference (CCC) (pp. 4659-4664). IEEE.
[12] Dixon, L., Li, J., Sorensen, J., Thain, N., & Vasserman, L. (2018, December). Measuring and mitigating unintended bias in text classification. In Proceedings of the 2018 AAAI/ACM Conference on AI, Ethics, and Society (pp. 67-73).
[13] Enria, L., Waterlow, N., Rogers, N. T., Brindle, H., Lal, S., Eggo, R. M., ... & Roberts, C. H. (2021). Trust and transparency in times of crisis: Results from an online survey during the first wave (April 2020) of the COVID-19 epidemic in the UK. PloS one, 16(2), e0239247.
[14] Grahl, T. (2013). The 6 types of social media. Out: think.
[15] Guidry, J. P., Carlyle, K. E., LaRose, J. G., Perrin, P., Messner, M., & Ryan, M. (2019). Using the health belief model to analyze Instagram posts about Zika for public health communications. Emerging infectious diseases, 25(1), 179.
[16] Guo, YR., Cao, QD., Hong, ZS. et al. (2020). The origin, transmission and clinical therapies on coronavirus disease 2019 (COVID-19) outbreak – an update on the status. Military Med Res 7, 11 (2020). https://doi.org/10.1186/s40779-020-00240-0
[17] Griol, D., Fernández, D. P., & Callejas, Z. (2021). Hispabot-Covid19: the official Spanish conversational system about Covid-19. In IberSPEECH.
[18] Heath, R. L., & Bryant, J. (2000). Human Communication Theory and Research: Concepts, Contexts, and Challenges. London: Lawrence Erlbaum Associates, Publishers.
[19] Hofmann, T. (2013). Probabilistic latent semantic analysis. arXiv preprint arXiv:1301.6705.
[20] Houston, D. J., & Harding, L. H. (2013). Public trust in government administrators: Explaining citizen perceptions of trustworthiness and competence. Public Integrity, 16(1), 53-76.
[21] John Field, ‘Bottom-up’ and ‘top-down’, ELT Journal, Volume 53, Issue 4, October 1999, Pages 338–339, https://doi.org/10.1093/eltj/53.4.338
[22] Júnior, V. O. D. S., Branco, J. A. C., De Oliveira, M. A., Da Silva, T. L. C., Cruz, L. A., & Magalhaes, R. P. (2021, October). A natural language understanding model COVID-19 based for 聊天機器人s. In 2021 IEEE 21st International conference on bioinformatics and bioengineering (BIBE) (pp. 1-7). IEEE.
[23] Kaplan, A. M., & Haenlein, M. (2010). Users of the world, unite! The challenges and opportunities of Social Media. Business horizons, 53(1), 59-68.
[24] Ke, G. N., Grajfoner, D., Wong, R. M. M., Carter, S., Khairudin, R., Lau, W. Y., ... & Lee, S. C. (2022). Building the positive emotion-resilience-coping efficacy model for COVID-19 pandemic. Frontiers in Psychology, 13, 764811.
[25] Kimhi, S., Marciano, H., Eshel, Y., & Adini, B. (2020). Recovery from the COVID-19 pandemic: Distress and resilience. International Journal of Disaster Risk Reduction, 50, 101843.
[26] Lin, T. E., Xu, H., & Zhang, H. (2020, April). Discovering new intents via constrained deep adaptive clustering with cluster refinement. In Proceedings of the AAAI Conference on Artificial Intelligence (Vol. 34, No. 05, pp. 8360-8367).
[27] Liu, P., Ning, Y., Wu, K. K., Li, K., & Meng, H. (2021). Open intent discovery through unsupervised semantic clustering and dependency parsing. arXiv preprint arXiv:2104.12114.
[28] Lucy R., Debie L., Lilly H., Catriona M., Becky R. (2020). https://www.bbc.com/zhongwen/resources/idt-ae57eb25-a3d3-4479-bd06-cf689e1f5f1d. Accessed July 25, 2023
[29] Ludolph, R., Schulz, P. J., & Chen, L. (2018). Investigating the effects of mass media exposure on the uptake of preventive measures by Hong Kong residents during the 2015 MERS outbreak: The mediating role of interpersonal communication and the perception of concern. Journal of Health Communication, 23(1), 1-8. doi: 10.1080/10810730.2017.1388455
[30] Maiman, L. A., & Becker, M. H. (1974). The health belief model: Origins and correlates in psychological theory. Health education monographs, 2(4), 336-353.
[31] Maddux, J. E., & Rogers, R. W. (1983). Protection motivation and self-efficacy: A revised theory of fear appeals and attitude change. Journal of experimental social psychology, 19(5), 469-479.
[32] McKillop, M., South, B. R., Preininger, A., Mason, M., & Jackson, G. P. (2021). Leveraging conversational technology to answer common COVID-19 questions. Journal of the American Medical Informatics Association, 28(4), 850-855.
[33] Moe, T. L., Gehbauer, F., Senitz, S., & Mueller, M. (2007). Balanced scorecard for natural disaster management projects. Disaster Prevention and Management: An International Journal, 16(5), 785-806.
[34] Narendra, L. W., & Setyaningsih, E. R. (2021, October). Designing a Transactional Smart Assistant in Indonesian using Rasa Framework. In 2021 7th International Conference on Electrical, Electronics and Information Engineering (ICEEIE) (pp. 1-6). IEEE.
[35] Nedelka, J.(2008). The Social Media Hype is Over. 1to1 Magazine, 10 (4), 20–24.
[36] Ng, K. H., & Lean, M. L. (2012). The Fukushima nuclear crisis reemphasizes the need for improved risk communication and better use of social media. Health physics, 103(3), 307-310.
[37] OECD (2022), "OECD Framework for the Classification of AI systems", OECD Digital Economy Papers, No. 323, OECD Publishing, Paris, https://doi.org/10.1787/cb6d9eca-en.
[38] Prati, G., Pietrantoni, L., & Zani, B. (2011). A social-cognitive model of pandemic influenza H1N1 risk perception and recommended behaviors in Italy. Risk Analysis, 31(4), 645-656. doi: 10.1111/j.1539-6924.2010.01529.x
[39] Qi, S., Ip, C., Leung, R., & Law, R. (2010, May). A new framework on website evaluation. In 2010 International conference on e-business and e-government (pp. 78-81). IEEE.
[40] Rocha, Á. (2012). Framework for a global quality evaluation of a website. Online information review, 36(3), 374-382.
[41] Rogers, R. W. (1975). A protection motivation theory of fear appeals and attitude change1. The journal of psychology, 91(1), 93-114.
[42] Rutsaert, P., Regan, Á., Pieniak, Z., McConnon, Á., Moss, A., Wall, P., & Verbeke, W. (2013). The use of social media in food risk and benefit communication. Trends in food science & technology, 30(1), 84-91.
[43] Song, C., & Lee, J. (2016). Citizens’ use of social media in government, perceived transparency, and trust in government. Public Performance & Management Review, 39(2), 430-453.
[44] Tinker, T., & Fouse, D. (2009). Expert round table on social media and risk communication during times of crisis: strategic challenges and opportunities. An American Public Health Association and partners' white paper on trends and opportunities for using social media to better prepare for and respond to emergencies.
[45] Tsai, J. Y., Shih, T. J., Tsai, T. I., Lee, S. M., & Liang, C. M. (2023). Individualism, Economic Development, and Democracy as Determinants of COVID-19 Risk Information on 132 Government 網站sites. Preventive Medicine Reports, 102242.
[46] Varghese, N. E., Sabat, I., Neumann-Böhme, S., Schreyögg, J., Stargardt, T., Torbica, A., ... & Brouwer, W. (2021). Risk communication during COVID-19: A descriptive study on familiarity with, adherence to and trust in the WHO preventive measures. PloS one, 16(4), e0250872.
[47] Visschers, V. H., Meertens, R. M., Passchier, W. W., & De Vries, N. N. (2009). Probability information in risk communication: a review of the research literature. Risk Analysis: An International Journal, 29(2), 267-287.
[48] Wang, H. W., Chen, G. W., Lee, W. L., You, S. H., Li, C. W., Jang, J. H., & Shieh, C. L. (2021). Learning From Each Other in the Management of Natural Disaster and COVID-19 Pandemic: A Case Study in Taiwan. Frontiers in public health, 2018.
[49] Warren, G. W., & Lofstedt, R. (2022). Risk communication and COVID-19 in Europe: lessons for future public health crises. Journal of Risk Research, 25(10), 1161-1175.
[50] Wei, J., Huang, C., Vosoughi, S., & Wei, J. (2020). What are people asking about covid-19? a question classification dataset. arXiv preprint arXiv:2005.12522.
[51] We Are Social & Meltwater (2020), “DIGITAL 2020: TAIWAN,” retrieved from https://datareportal.com/reports/digital-2020-taiwan
[52] We Are Social & Meltwater (2021), “DIGITAL 2021: TAIWAN,” retrieved from https://datareportal.com/reports/digital-2021-taiwan
[53] We Are Social & Meltwater (2022), “DIGITAL 2022: TAIWAN,” retrieved from https://datareportal.com/reports/digital-2022-taiwan
[54] We Are Social & Meltwater (2020), “Digital 2020 Global Overview Report,” retrieved from https://datareportal.com/reports/digital-2020-global-overview-report
[55] We Are Social & Meltwater (2021), “Digital 2020 Global Overview Report,” retrieved from https://datareportal.com/reports/digital-2021-global-overview-report
[56] We Are Social & Meltwater (2022), “Digital 2020 Global Overview Report,” retrieved from https://datareportal.com/reports/digital-2022-global-overview-report
[57] Witte, K. (1992). Putting the fear back into fear appeals: The extended parallel process model. Communications Monographs, 59(4), 329-349.
[58] World Health Organization General information on risk communication. Available from: https://www.who.int/risk-communication/background/en/. Accessed March 29, 2020
[59] Wu, D., Wu, T., Liu, Q., & Yang, Z. (2020). The SARS-CoV-2 outbreak: what we know. International journal of infectious diseases, 94, 44-48.
[60] Yen, M. Y., Yen, Y. F., Chen, S. Y., Lee, T. I., Huang, K. H., Chan, T. C., Tung, T. H., Hsu, L. Y., Chiu, T. Y., Hsueh, P. R., & King, C. C. (2021). Learning from the past: Taiwan's responses to COVID-19 versus SARS. International journal of infectious diseases : IJID : official publication of the International Society for Infectious Diseases, 110, 469–478. https://doi.org/10.1016/j.ijid.2021.06.002
[61] Zeemering, E. S. (2021). Functional fragmentation in city hall and Twitter communication during the COVID-19 Pandemic: Evidence from Atlanta, San Francisco, and Washington, DC. Government Information Quarterly, 38(1), 101539.
[62] Zhou, F., Yu, T., Du, R., Fan, G., Liu, Y., Liu, Z., ... & Cao, B. (2020). Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study. The lancet, 395(10229), 1054-1062.
[63] COVID-19防疫關鍵決策時間軸。民112年3月11日，取自:衛生福利部COVID-19防疫關鍵決策時間軸: https://covid19.mohw.gov.tw/ch/sp-timeline0-205.h
[64] H1N1新型流感大流行工作紀實
[65] 李韶曼，台灣政府疫情網站分類，未發表草稿（2023）
[66] 臺灣通傳智庫股份有限公司 (民110 年 06 月) 。數位時代下之新型態政策溝通發展趨勢與創新模式—以武漢肺炎（COVID-19）疫情為例。取自https://www.ndc.gov.tw/nc_708_35589。
[67] 行政院人事行政總處、內政部(民112年3月1日)。「政府組織」。取自https://www.ey.gov.tw/state/62879155A536D543/f3662cb7-5806-4755-8aba-ab1d44b97d7a
[68] 疾病管制署(民112年7月5)。「疾管署LINE@疾管家滿週歲！導入醫療自然語言處理，法定傳染病問他就對了！」。取自https://www.mohw.gov.tw/cp-16-43973-1.html
[69] 伊澤佑美, 根本陽平(2019)。《數位時代的公關新主張》（頁149），王美娟譯，臺北：東販出版
[70] 張四明(民2020)。臺灣2020年新冠肺炎防疫大作戰之啟示：政策工具觀點分析
[71] 施琮仁. (2021). 新冠肺炎的日常防疫行為: 媒體, 情緒與風險認知的作用. 新聞學研究, (148), 153-196.
[72] 林詠青. (2021). 從 SARS 到 COVID-19—重大傳染病防治與基本權干預.
[73] 許家瑜, 陳筱丹, 王恩慈, 劉慧蓉, & 楊靖慧. (2020). 臺灣嚴重特殊傳染性肺炎 (COVID-19) 社區防疫措施與成效. 疫情報導, 36(15), 234-44.
[74] 吳宜蓁. (2011). 運用網路社交媒體於風險溝通-以 2009-2010 年台灣政府 H1N1 防疫宣導為例. 傳播與社會學刊, (15), 125-159.
[75] 陳憶寧. (2018). 傳統媒體與社群媒體上呈現的食安危機比較: 頂新餿水油事件的個案研究. 傳播與社會學刊 (總), 45, 47-89.
[76] 宋楨堯. (2021). 應用健康信念模式探討在嚴重特殊型傳染性疾病下民眾口罩購買行為之研究.
[77] 胡幼慧、姚美華（1996），一些質性方法上的思考：信度與效度？如何抽樣？如何 收集資料、登錄與分析？載於胡幼慧主編，質性研究：理論、方法及本土女性 研究實例。台北市：巨流出版社。
[78] 周淑玫、陳昶勳、周玉民與張佳琪 (2008). 因應流感大流行防治策略綜覽. 疫情報導, 24(12), 879-894.