智慧型電腦分組系統之設計與評估

Abstract

好的開始是成功的一半！所以設計出良好的小組是相當重要的。但由於各領域其影響小組成效的因素有所不同，因此本論文將研究範圍限定在教育上，探討如何將一個班級的學生組成許多良好的合作學習小組。 合作學習廣泛地被各級學校所採用，但課堂教師最常採用的分組方式是隨機抽籤或由學生自行組隊，很難可以如教育理論所建議的—採用學生的心理特質進行異質分組，以提供合作學習的成功機率。因此本論文提出一套嶄新的異質分組系統，希望基於教育理論、以資訊技術為工具，協助教師得以進行更精緻化的合作學習。 本系統中，提供兩種異質分組法—DIANA、FUTS。DIANA分組法基於教育公平性，以基因演算法尋找較佳的分組可能，提供組內異質、組間同質的分組建議。FUTS分組法基於心理學大師Sternberg的建議，以反覆改善演算法，企圖組合出各小組風格皆相當顯著的組合。 本論文中，以合作設計教學實驗(475人)來驗證這兩種分組法的效能，發現其在成績與學生滿意度上皆顯著優於隨機分組，且學生年級及樣本密度此二權變因素與分組法有顯著的交互作用。實驗結果建議國小教師可採用DIANA平衡分組法，國中教師可採用DIANA與FUTS分組法。

Many Computer-Supported Collaborative Learning (CSCL) systems emphasize the use of technical tools to support effective/productive social interaction, social knowledge construction, and reflection. Only a handful of CSCL mediating tools provide help for the task of composing groups with good chances of success in task achievement and learning, and even fewer offer algorithm-based recommendations for establishing groups from a large pool of students with various characteristics and/or learning capacities. The author describes her work with other designers to create two computer-supported grouping systems. The first, named Differences In And Non-differences Among groups (DIANA), uses genetic algorithms to achieve fairness, equity, flexibility, and balance. The other, named Full Thinking Styles groups (FTS), uses iterative improvement algorithms to ensure the representation of all possible thinking styles in each team. DIANA and FTS were tested with 475 students assigned to groups of three, either randomly (63 groups), using DIANA (52), or using FTS (43). The results indicate that a) DIANA and FTS groups outperformed random groups overall, b) DIANA groups outperformed FTS and random groups in elementary school settings, and c) DIANA and FTS groups outperformed random groups in junior high school settings. Suggestions for applying computer-supported group composition systems are offered.