- 期刊
- OpenAccess
應用潛在狄利克雷分布的主題建模技術分析科學教師的探究教學目標及其與108課綱「探究與實作」課程目標之比較
Analyzing Science Teachers' Inquiry Teaching Goals Using Latent Dirichlet Allocation Topic Modeling and Comparing Them With the "Inquiry and Practice" Curriculum Goals of the 108 Curriculum Guidelines
摘要
本研究呼應人工智慧技術在科學教育研究方法上的精進，採用自然語言處理領域中潛在狄利克雷分布（Latent Dirichlet Allocation, LDA）的主題建模技術，來分析高中科學教師對「探究與實作」課程所持有的教學目標，並與課綱中的課程目標進行比較，同時探索教師的任教科目、專業成長經驗、科學教學年資、探究教學年資等背景因素與其教學目標的關係。「探究與實作」課程是十二年國民基本教育課程綱要（簡稱108課綱）自然領域教學中重要的革新，而教師的教學目標是行動的核心動機因素，可引導其教學實務的規劃與執行，故成為本研究的焦點。而LDA技術可以識別文本中的潛在主題，不受預先設定的分析框架所限，有助於本研究探索教師持有的想法。本研究徵募40位具有探究教學經驗的科學教師參與半結構訪談，並使用R程式語言對訪談的逐字稿進行主題建模分析。結果指出教師的教學目標分屬於10項主題，每項主題都包含數個目標。其中，讓學生投入科學實務，發現與解決問題，提升學習動機與科學態度等目標普遍存在於大部分的主題中。但發展模型、數學與運算思維、論證與建構解釋等科學實務較少被提及。相較於課綱目標，培養自主行動、高層次獨立思考能力及跨領域的知識與能力等目標也比較少見。不同背景因素的教師在教學目標上有部分交集，也有各自強調之處。研究結果盼能提供教師及師資培育單位在設計課程及專業成長活動時的參考。
並列摘要
This study responds to the advancement of research methodologies of Artificial Intelligence (AI) in science education by utilizing Latent Dirichlet Allocation (LDA) topic modeling, a natural language processing technique, to analyze the teaching goals of high school science teachers in the context of "Inquiry and Practice" curriculum. A comparison with the curriculum goals outlined in the guidelines was conducted, alongside an exploration of the relationship between teachers' background factors such as the subjects they teach, professional development experience, years of science and inquiry-based teaching, and their teaching goals. LDA topic modeling facilitates the identification of latent topics within text, unconstrained by predefined analytical frameworks. "Inquiry and Practice" curriculum represents a significant innovation in the science education under the 108 curriculum guidelines in Taiwan, with teachers' teaching goals serving as core motivational factors guiding their instructional planning and execution, thus becoming the focal point of this study. Forty science teachers with experience in inquiry teaching were recruited for semi-structured interviews, and the transcripts were subjected to topic analysis using the R programming language. Results revealed that teachers' teaching goals clustered into 10 topics, each comprising several goals. Common goals observed across the majority of topics included engaging students in scientific practices, fostering problem-solving and discovery abilities, and enhancing motivation and attitudes toward science. However, certain scientific practices such as model development, mathematical and computational thinking, argumentation, and constructing explanations received less emphasis. In comparison to curriculum objectives, goals such as cultivating autonomy, high-level independent thinking abilities, and interdisciplinary knowledge and abilities were relatively underrepresented. Teachers with different experience showed some overlap in teaching goals, as well as distinct emphases. The findings of this study aim to provide valuable insights for teachers and relevant stakeholders in designing curricula and professional development activities.