透過您的圖書館登入
IP:3.214.184.69
  • 學位論文

以順序理論提升貝氏網路診斷測驗之成效-以國小數學五年級領域「數列與圖形序列」為例

Applying Ordering Theory to Improve the Performances of Bayesian Network based Computerized Diagnostic Test effects - Using The “Number of Sequence and Geometrically Sequence“ in Elementary School Math in Grade 5 as an example

指導教授 : 劉湘川 曹世昌
若您是本文的作者,可授權文章由華藝線上圖書館中協助推廣。

摘要


本研究將以順序理論來提升貝氏網路診斷測驗之成效,以電腦化自動線上適性診斷測驗系統,適性補救教學系統來進行驗證,並利用貝氏網路強大的預測功能及學生試題結構之學習概念選題方式,能以最少的施測題目來預測診斷學生在學習後,還會出現的錯誤概念,並讓教學者及學生能了解學習者是否已精熟了相關的子技能,並針對學生還會犯的錯誤類型及未達精熟的子技能進行補救教學。 本研究之結果將提供五年級任課教師及學生在學習本單元後,能迅速的診斷評量出學生的個別學習狀況,節省診斷試題數及施測時間。並藉由Swish max軟體所製作補救教學動畫,配合多媒體電腦來學習,以提升學生學習興趣,降低學習時間與空間的障礙。在學生學習完這個單元後,只要有電腦、能上網就可以隨時上網進行個別化的適性診斷及補救教學,對學生能有更大的助益。本研究結果發現: 一、 在學生錯誤類型及子技能通過與否之診斷測驗中,以學生試題結構確實能提昇貝氏網路推論工具之辨識效果。(動態決斷值辨識率平均為0.9357) 二、 在系統進行補救教學後,學生在子技能的通過率有提升,錯誤類型的平均發生率能有效減少,學生的前後、測平均成績有進步。 三、 適性作答結果和完整作答結果在貝氏網路的推論比較,兩者在前、後測上相似度都達九成以上,前、後測相差1.4%。所以學生試題結構所建立的電腦適性測驗是可以用適性選題來取代完整試題全測 四、 系統在前測時平均節省題數為11.5 題,在後測時,平均節省題數為13.3題。前、後測學生實際答對和系統推估預測答對的誤差情形兩者相差0.68%,因系統適性選題的閥值定為5%,誤差率在可接受範圍,由此表顯示適性診斷測驗確實可以節省試題。

並列摘要


The research of this project is applies ordering theory to improve the performance of Bayesian network based computerized diagnostic test effects. Build computerized automatic on-line adaptive diagnostic test system and teaching-aid system to test and verify student performance. Predict the wild Bayesian network based function and study selected concepts from a student's examination. Predict with fewer questions, the diagnosis of problems that appear after learning. Teachers and students can be able to understand whether the learner is already totally familiar with relevant skills. Remedying the problems that students still have or dealing with the questions that the student isn’t able to answer. The result of this study will tell teachers and students in grade five who learned the unit, it can quickly diagnose each student’s individual learning situation and reduce the diagnosis of the question quantity and testing time. The dynamic teaching animation by Swish max software to cooperate with multimedia study, to improve students’ interest in studying. Also reduce the obstacle of study time. After the students finish this study, whenever a computer connects with internet, students can be suitable for individual diagnosis and assist teaching easily, it also helps more pupils with this issue. The detection from a result of this study: a. Whether the students can be pass diagnosis or not, the student’s item relational structure can still get results from the Bayesian Network.(The average dynamic threshold value is 0.9357) b. As a result of doing the system’s remedy teaching, the student gets a higher percentage raising t through the rate. The average incidence of wrong answers can be reduced effectively.The test of the average score is higher than the first time. c. There’s comparison between the adaptive test and the all the test questions in the Bayesian network’s inference. Both test results are similar. The average tests differ by 1.4%. Therefore, building of a computerized system with adaptive test of student’s item relational structure selected questions can replace a complete examination. d. It also economizes on the question’s quantity, 11.5 in the first test and 13.3 in t the second test.When students answer a hand written questions and questions predicted by the system mistakes in the two tests differ by 0.68%.

參考文獻


沈俊達(2004)。以知識結構為主的電腦適性診斷測驗系統之研究~以國民小學數學領域為例。臺中縣。臺中健康暨管理學院資訊科技研究所碩士論文,未出版。
黃雅鳳(2006)。以貝氏網路為基礎之能力指標測驗編製及補救教學動畫製作-以六年級數學領堿之「分數與小數」相關指標為例。臺中縣。臺中亞洲大學資訊工程研究所碩士論文,未出版。
張瑞欣(2006)。以貝氏網路為基礎之線上適性測驗系統演算法。臺中縣。臺中亞洲大學資訊工程研究所碩士論文,未出版。
洪旭亮(2006)。以貝氏網路為基礎之能力指標測驗編製及補救教學動畫製作-以四年級數學領堿之「整數」相關能力指標為例。臺中縣。臺中亞洲大學資訊工程研究所碩士論文,未出版。
李佩玟(2004)。國小六年級學童發現數列樣式的推理歷程之分析研究。臺北市。國立臺北教育大學教育心理與諮商學系碩士班碩士論文,未出版。

被引用紀錄


林美秀(2009)。數位教材與電腦適性診斷測驗融入教學之探討-以國小六年級數學「放大縮小比例尺」單元為例〔碩士論文,亞洲大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0118-1511201215461355

延伸閱讀