人工智慧優化廢棄物清理專業技術人員訓練測驗試題工作計畫
| 中文摘要 | 本計畫主要目標是針對廢棄物清除/處理專業技術人員測驗試題進行測驗學理分析，並對各科教材以人工智慧技術進行知識圖譜分析，再進一步發展自動命題技術與自動難度預測模型，為未來國家環境研究院的考試建立AI輔助認證考試的典範。 目前已完成15科題庫之試題難度與考生能力值分析及等化，建立統一量尺的試題與受測者資料庫，含AI題幹與選項相似性特徵分析，並完成各科題庫分析與優化之摘要報告。未來建議持續蒐集考生資料，分析題庫中其他試題之數據資料，以提高測驗的效度及自動命題難度預測模型的效能。 在各科教材的知識圖譜AI分析部分，已完成結合知識譜圖與歷年試題數據，依據預設通過率進行人工智慧自動命題，生成適當難易度的練習題及選擇題題組，並輸出符合現有系統格式的題目內容。為保證試題品質，計畫亦辦理專家諮詢會議，檢視並優化生成試題的妥適性。 本計畫提升了測驗系統的科學性與效率，亦為專業測驗系統的未來發展提供了新方向。結合測驗學理、知識譜圖與人工智慧技術的創新應用，顯著改善了題庫管理與命題過程，並實現了更高效的測驗生態系統。 | ||
|---|---|---|---|
| 中文關鍵字 | 測驗學理 知識譜圖 AI 自動命題 | ||
基本資訊
| 專案計畫編號 | 經費年度 | 113 | 計畫經費 | 2980 千元 | |
|---|---|---|---|---|---|
| 專案開始日期 | 2024/10/17 | 專案結束日期 | 2024/12/31 | 專案主持人 | 林閔瑩 |
| 主辦單位 | 國環院環教認證中心 | 承辦人 | 盧素如 | 執行單位 | 財團法人商業研究院 |
成果下載
| 類型 | 檔名 | 檔案大小 | 說明 |
|---|---|---|---|
| 期末報告 | 國環院期末報告-定稿-v1.pdf | 8MB |
AI-Optimized Training Exam Questions for Waste Management Professionals
| 英文摘要 | The primary objective of this project is to perform test theory analysis on examination items for the professional certification tests in waste disposal and treatment. It incorporates artificial intelligence techniques to analyze knowledge graphs across various subject materials, developing automated item generation technologies and difficulty prediction models. This work establishes a paradigm for AI-assisted certification examinations in the National Institute of Environmental Research's environmental education certification system. To date, the project has completed difficulty analysis and ability parameter estimation for item banks across 15 subjects, establishing a unified scale database that equates items and examinee responses. This includes AI-based analysis of stem and option similarity features, resulting in detailed analytical reports and optimization summaries for each subject's item bank. Future recommendations include the continued collection of examinee data and further analysis of additional test items to enhance test validity and improve the automated difficulty prediction models. Regarding AI-based knowledge graph analysis of subject materials, the project has successfully integrated knowledge graph mapping with historical item statistics to implement AI-powered automated item generation based on predefined passing rates. This system generates practice items and multiple-choice question sets with appropriate difficulty levels and outputs content compatible with existing system formats. To ensure item quality, expert consultation meetings were held to review and optimize the appropriateness of the generated items. | ||
|---|---|---|---|
| 英文關鍵字 | Test theory Test theory Knowledge graph AI-assisted automated question generation graph | ||