Story ChatBot (2022-2025)
PI & Co-PI: 張嘉惠、劉晨鐘、鐘曉芳、鄭芳祥
Team member:
2021-2022 黃紫嫺、黃覺修、高愷言、陳臆玄
2022-2023 林子平、李聿鎧、黃冠傑、黃淯銘
2023-2024 黃冠傑、黃淯銘、龔若齊、丁仕杰、謝程偉
Abstract
Communication skills are an important part of academic success and healthy relationships. Teacher-centered teaching style is often a one-way knowledge transfer, leaving limited time for students to interact and discuss in the classroom. In order to improve the oral narrative ability of school-aged children, this project hopes to build a story Chatbot that can interact with primary school students (in Chinese) about the English stories they read, expressing what they read, feel and discover through conversation, establishing their viewpoints and attitude towards life. We will use information extraction techniques as the basis for robots to understand stories, and guide children to converse with robots through 4F (Fact, Fact, Feeling, Discovery, Future) guidelines, about their understanding of the story, as well as feeling, discovering, and prediction of the story. In addition to reviewing the stories children have read, the story chatbot should also be able to respond to their questions, understand their feelings and give empathetic responses. It is hoped that through interaction, school-aged children will improve their narrative skills and be encouraged express themselves.
溝通技巧是學業成功和健康關係的重要組成部分。傳統以教師為中心的教學方式通常是單向的知識傳遞, 課堂上留給學生互動討論的時間很有限。為了提高學齡兒童的口語敘述能力, 本計畫希望構建一個聊故事機器人，可以與小學生（用中文）談論他們讀到的英語故事、透過互動練習來表達他們所閱讀、感受和發現的內容，以支持他們的觀點或建立他們對生活的態度。我們將使用資訊擷取作為機器人理解故事的基礎，透過4F（Fact、Fact、Feeling、Discovery、Future）提問引導學童回答故事相關問題，以及感受、發現、與預測故事的可能發展。除了引導學童回顧所閱讀的故事，聊故事機器人也要能回應學童的問題，同時也能基於常識知識圖譜（CSKG）理解學童的感受，給予同理心回覆。希望透過互動提高學習者的敘述技巧，並鼓勵他們表達自己的觀點。
Publications:
黃冠傑、黃淯銘、張嘉惠、黃覺修、鍾曉芳、鄭芳祥, 通過實體和事件關係標記改進問題答案對生成的 控制 , NCS2023.
許志仲、朱翊瑄、劉晨鐘、張嘉惠、温采婷, 基於生成式語言模型之科學探究教學代理之提示詞框架與系統設計, NCS2023.
李聿鎧 、應用強化學習與知識圖譜於故事共述生成之研究, ROCLING 2023
林子平、張嘉惠、劉晨鐘: 基於大型語言模型應用指示詞打造無程式碼對話系統平台 - 以聊故事機器人為例, TAAI 2023
黃紫嫺、張嘉惠: 基於常識知識的移情對話回覆生成, ROCLING 2022
高愷言、張嘉惠: 應用自動資訊擷取於故事書問答之研究, ROCLING 2022