Invited Panel and Special Session
Special Session
Application of Digital Technology in Teaching, Research and Tutoring : Chatbots in Education
Chair: Chien-Ping Chung (National Taiwan University of Science and Technology)
Session Abstract
隨著人工智慧與網際網路技術的快速發展，不但加速了人類知識的快速累積，也改變了人類的學習型態。尤其是智慧型手機的普及，帶來許多學生學習方式的轉變。行動科技最受人矚目之特有優勢即在於「促進個人化學習」、「提供即時回饋與評量」、「促進任何時間、任何地點的學習」、「支持情境式的學習」與「最大化經濟效益」。此外，隨著社群平台使用的普及化，建構了許多網路社群，也拓展了許多以社群平台為中心的生活圈，學生在社群平台參與程度也是逐年遞增。因此，若能透過數位工具—聊天機器人導入將有助於提升學生學習意願、社群參與程度、教學評量及資訊不對稱改善等效益提升。本場次共有四篇文章發表，依序為：1. 應用資訊系統改善媒合之資訊不對稱與焦慮；2. 建置以訓練為導向之專業英語課程聊天機器人方法論；3. 智能聊天機器人學術導師於自我評量及教學評量之應用；及4. 社群與對話式介面設計對理財學習網路社群參與意願之影響。
Presented Papers
1.應用資訊系統改善媒合之資訊不對稱與焦慮 Information Systems in Matchmaking for Improving
Information Asymmetry and Anxiety
吳恩昕 En-Hsin Wu（國立臺灣科技大學 National Taiwan University of Science and Technology）
呂志豪 Shih-Hao Lu （國立臺灣科技大學 National Taiwan University of Science and Technology）
在招募過程中，企業資訊揭露程度影響求職者決策。本研究將招募情境設置
於校園，以 A 大學 109 學年度企業管理系之碩士班甄試生與考試生為受測對象，
透過使用媒合系統與聊天機器人，大幅降低研究生找尋指導教授所需時間與精力、
資訊不對稱與焦慮感，並進一步配對到最適合的教授，而指導教授亦能收到最符
合期待之學生。
本研究共有兩項研究，研究一：透過媒合系統的使用與發問來降低資訊不對
稱與焦慮感。媒合系統是透過收集教授選擇學生的條件與配分建置而成。後續學
生填答系統後即能算出與教授的適配度，成為學生找指導教授的參考指標。
研究一採單組前後測實驗設計，於使用媒合系統前後與發問前後，施予資訊
不對稱與焦慮問卷量表測驗，共回收 53 份有效問卷。研究結果發現：（1）使用
媒合系統對降低資訊不對稱與焦慮有正向影響。（2）發問行為對降低資訊不對
稱與焦慮有正向影響。（3）資訊不對稱程度影響資訊收集行為的多寡，而愈焦
慮的人，愈傾向做出資訊收集行為。
根據研究一結果，研究生之發問行為有效降低資訊不對稱與焦慮程度，因此
發展研究二：建立聊天機器人，收集研究生之常見問題。該研究目前正在進行中。
關鍵詞：決策支援系統、資訊收集、發問行為、聊天機器人
2.建置以訓練為導向之專業英語課程聊天機器人方法論 A Methodology for Constructing Chatbots for Training-
Oriented ESP Courses
陳庭涓 Ting-Juan Chen（國立臺灣科技大學 National Taiwan University of Science and Technology）
呂志豪 Shih-Hao Lu （國立臺灣科技大學 National Taiwan University of Science and Technology）
摘要/Abstract
Research on the field of training and development has lasted for a long
time and is mainly linked with human resource management. It was found
that the basis of training is learning and training can be applied to diverse
scenarios not only in the workplace but also in education. Moreover, as the
generation Z was born and has been growing up, the trend of implementing
technology into education is rocketing up in order to fulfill their need of
multitasking. Furthermore, in the end of January 2020, the outbreak of
COVID-19 has forced schools to shut down, and students can only stay at
home to have distant learning. The demand for learning by technological
approach is booming. Besides the video communicator, another
frequently-used approach is “chatbot” on mobile devices which are
connected with instant communication applications. According to the
research on the review and quality assessment of educational chatbots by
Smutny and Schreiberova (2020), the most used module were quizzes or
the question and answer and the broadcast function can help learners stay
engaged. However, these educational chatbots exist shortly, which led the
difficulty for other educators to find and use.
Although several researches have put an emphasis on the
conversational chatbot for a variety of learning use. Research on the
application of Natural Language Processing (NLP) into educational
chatbots and the method of constructing and designing a pedagogical
chatbot for English for Specific Purposes (ESP) courses were relatively
scant. Therefore, the aim of this study is to develop a methodology of
building up a customized chatbot by the technique of NLP to assist ESP
lecturers’ teaching in class and even after class.
Four-stage study is designed to explore the students’ and teachers’
needs and the effect of chatbot on teaching, learning and skill training, and
most importantly, to propose a practical methodology for educators to
build up chatbot for their specific purpose on their own. In order to delve
into the influence of chatbot on training-oriented courses, this study
involves four rounds of data collection by the means of historical document
and records of teaching feedback, teaching expert interview, statistics on
the chatbot developing platform, and questionnaire of user experience
(accessibility, helpfulness and efficiency), and two rounds of data analysis
via content analysis and descriptive statistics analysis.
The expected outcome will be a proposed methodology for educators
in higher education to efficiently construct an effective chatbot for trainingoriented ESP courses including the suggested methods of data collection,
data analysis and platform selection, steps of construction and the design
of user manual. The researcher believes that this proposed methodology
for construction of chatbot will be beneficial for both training-oriented
courses in schools and training department in organizations from the
perspective of technology adoption and can be applied in different fields
of training.
Keywords:training and development, English for Specific Purposes, technology adoption,educational chatbots, natural language processing
3.智能聊天機器人學術導師於自我評量及教學評量之應用 Application of Academic Advisors AI Chatbots for
Student Self-Evaluation and Teaching Evaluations in Taiwan
王緁妶 Chieh-Hsuan Wang（世新大學 Shih Hsin University）
鍾建屏 Chien-Ping Chung （國立臺灣科技大學 National Taiwan University of Science and Technology）
摘要/Abstract
本研究結合通訊App Line並建置智能聊天機器人對學生進行自我評量及教學評量，首先打造出課程專屬的學生端及教師 (管理) 端「智能聊天機器人學術導師」，並採自然語言處理模式，持續增進機器人的語句理解能力，以支援教師教學，並提升學生的主動學習的動機，來獲得即時的教學回饋。其次，透過建置「聊天機器人智能反饋教學評量系統」，藉由人工智慧的引入讓評量介面與學生之間的互動更有溫度，以達到提高學生填答率及填答精確度，並產生學習成效的統計報表方便供學校教學發展中心進行檢核。最後對透過人工智慧教學評量反饋系統之系統態度問卷施測來了解學生對這套評量系統的填達意願、填達精確度、系統接受度及滿意度等。
在本研究的校園問卷調查中，有74%的學生受測者認為互動式填答的問題描述式容易理解的，其中許多人認為藉由聊天機器人來進行互動填答會讓人更專注在理解問題上，減少了亂填的情形；有將近70%的學生受測者認為互動式填答比傳統的紙本以及線上填答更具備互動性、親切度、方便度和易用性，同時也提升了填答的動機並且更願意給予教學意見上的文字回饋，而這些問題項持負向看法的受測者幾乎佔不到10%。
關鍵詞：聊天機器人、智能反饋教學評量系統、學術導師
4.社群與對話式介面設計對理財學習網路社群參與意願之影響 The Influence of Investment Communities and
Conversational User Interface Design on Participation in Online Financial Advising Groups
羅歆慈（國立臺北商業大學 National Taipei University of Business）
范晴虹 （國立臺北商業大學 National Taipei University of Business）
溫明輝 Ming-Hui Wen（國立臺北商業大學 National Taipei University of Business）
摘要/Abstract
聊天通訊軟體(Instant Messenger)在人們生活中扮演重要的角色，其用來與使用者溝通的介面--對話式介面(CUI: Conversational User Interface)提供了一種新的互動方式促進人與電腦之間的互動。本研究針對金融學習為主題，應用對話式介面做為廣告內容呈現的媒介，探討介面設計元素以及社群元素對年輕族群參與理財學習意圖之影響。具體而言，本研究設計一 3 x 2 的因子實驗，透過操弄介面設計元素（對話以圖片形式 / 推薦性語氣 / 中性語氣）以及社群元素（廣告含社群元素 / 廣告不含社群元素）探討受測者對理財社群廣告內容互動情形及參與意願的影響。實驗的廣告訊息由一擁有30萬使用者的理財機器人推送給族群，每個因子實驗組合各有 200 位參與實驗者，男女各半，合計 1,200 位。實驗者對廣告內容的點擊次數、及對理財學習社群的加入次數等數據皆透過系統進行收集。本研究之結果預期可找到一能提高對話式軟體中的金融廣告內容點擊與參與意願之介面設計元素組合參考。
關鍵詞：聊天通訊軟體、對話式人機介面、金融教育、廣告設計
