簡易檢索 / 詳目顯示
| 研究生： |
陳昭安 |
|---|---|
| 論文名稱： |
建構試題自動分類系統之研究－以MOCC術科試題為例 The Study on Constructing System of the Autonomous Categorization for Questions - A Case Study of MOCC Skill Questions |
| 指導教授： | 戴建耘 |
| 學位類別： |
碩士 Master |
| 系所名稱： |
工業教育學系 Department of Industrial Education |
| 論文出版年： | 2002 |
| 畢業學年度： | 90 |
| 語文別： | 中文 |
| 中文關鍵詞： | 試題分類 、字詞剖析 、文件分類 、字詞出現頻率 、逆文件頻率 |
| 英文關鍵詞： | Question Categorization (QC), Text Categorization, Term Parser(TP), Term Frequency (TF), Inverse Document Frequency (IDF) |
| 論文種類： | 學術論文 |
| 相關次數： | 點閱：191 下載：4 |
| 分享至: |
| 查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報 |
文件分類之研究自1961年由Maron提出之後，無論在中文或是英文方面都已經得到很好的成果，運用這方面研究的基礎，可轉化到試題分類的機制。試題分類系統可有效簡化教學工作者對於大量不同來源的試題作分類的負擔，進而充分的準備教材。在學習者方面，可幫助了解試題所包含的能力項目及比重，而達到更有效率的學習。在試題開發者方面，可以作為分析及鑑定試題優劣的工具。
本研究的素材為「中華民國電腦教育發展協會」所發展的PreMOUS Word (MOCC Word標準級) 及MOCC Word專業級認證，研究中採擷PreMOUS Word試題的關鍵字詞構成「特徵關鍵字詞庫(FTB)」，並藉由關鍵字詞與各能力項目間的權重，構成權重向量矩陣。應用TF、TFxIDF、IDF、WIDF和TFxIDF2等字詞權重函數構成權重向量矩陣，對PreMOUS的試題作回歸測試 (Recall Testing)，並由回歸率 (Recall Rate)了解能力項目分類的正確程度，找出最佳的方法。
本研究針對PreMOUS Word試題進行回歸測試，可得到高達100%的回歸率。 針對多於24能力項目，且具複雜的綜合能力項目的MOCC Word 專業級術科題庫分類，亦有高達72%的正確率。運用此試題分類系統可有效的提高試題的分類速度，進而達到題庫品質管控的目標。
Ever since Maron’s text categorization in 1961, Chinese and English have became more efficient today. By applying this basis of study, we can convert the domain of text categorization into question categorization. The Question Categorization System (QCS) can simplify tasks about the categorization of questions from any source, and provides enough time to prepare class materials; this is a marvelous tool for educators or scholars. As for learners, The QCS assists them to comprehend each question’s ability and enhances learning efficiency. For question developers, The QCS is used to analyze the quality of questions.
This study consists PreMOUS Word (MOCC Word Core) and MOCC Word Expert certificate model, developed by “The Chinese Computer Education Association” (CCEA). In this study uses terms from the question of PreMOUS Word, and construct a Feature Term Base (FTB). By using the weight of terms in every ability item is called Weighted Vector Matrix (WVM). Finally, the PreMOUS’s Questions uses Recall Testing. The result of recall rate shows the preciseness of The QCS, and obtains an optimum method of question categorization.
Recall testing is applied for questions in PreMOUS Word, which creates a possibility of 100% recall rate. The categorizing in operational item bank of MOCC Word Expert certificate has more than 24 ability items, and has reached 72% accuracy. QCS effectively raises the speed of categorizing; therefore, it is the target study for item bank’s quality control.