近似值測量之改良及應用--基於命題之概念圖結構分析

Abstract

Goldsmith 等人之近似值測量，原是針對一般知識結構的分析，也被建議用在概念圖的計分上，其效果亦受肯定。該測量主要在比較兩知識圖的相似程度，只要兩圖中同一節點及其鄰近節點相同，便被視為具有相同的結構。如此一來，運用此方法分析概念圖時，概念間的連結關係便會被忽略。然而，概念和連結共同組成的命題才是概念圖中構成有意義知識的最小單位，本研究因此針對 Goldsmith 等人之測量法的應用進行改良，建議將分析的基礎由「概念」轉化為「命題」節點。本研究並透過一電腦化概念構圖工具蒐集197份學生的概念圖資料，以概念圖典型的 Novak 與 Gowin 計分法所得之分數為效標，來檢驗此轉化及應用的有效性。資料分析結果顯示，以命題為主的近似值測量結果與效標分數之 Pearson 相關係數（r= .888, p< .01）顯著高於以概念為主的近似值測量結果與效標分數的相關係數（r= .712, p< .01）。

Goldsmith et al.`s closeness measure, developed for evaluating structural representation of knowledge, has been applied to assess concept maps and the effectiveness has been positively affirmed. The measure quantifies the configural similarity between two networks by examing the degree to which the same node in two graphs is surrounded by a similar neighborhood of nodes. Once it is applied to concept map scoring, it would neglect the meaning relationship between concepts. Considering a proposition is the basic unit of meaning in a concept map, this study proposed that a concept map be transformed into a proposition-based map before applying the closeness mesure. To validate the approach, 197 students` concept maps were collected and scored. The scores derived by the closeness measure on the original concept maps and the transformed proposition based maps were compared with the scores by Novak and Gowin`s scoring scheme (criterion). The statistical analyses showed that the correlation between the scores from proposition based closeness measure and the criterion scores (r= .888, p< .01) was significantly higher than that between concept based closeness measusre and the criterion scores (r= .712, p< .01).