數位化學習單作業批改系統

楊智琨

講者/Speaker: 楊智琨, YANG ZHI KUN

高中教師日常面臨大量學習單批改工作,一份學習單從整理、評閱到輸出 PDF 學習歷程,往往耗時數小時以上。重複的文字回饋、立可帶修正、桌面被作業淹沒,不僅損耗教師心力,也影響學習歷程檔案的呈現品質。
本研究設計一套「數位化學習單作業批改系統」,採用 Python 3 + PyQt5 + PyMuPDF + SQLite 架構,將教師批改流程拆解為兩條主線:頁面整理(支線 A)提供五碼座號快捷標記、自動跳頁、頁面旋轉與群組合併;評改作業(支線 B)整合印章縮圖牆、罐頭評語快捷鈕,並以向量方式將 A4 原稿無損疊合至 A3 畫布輸出。
系統最大特色為「類似遊戲存檔/載入的任務管理機制」:教師可隨時中斷批改,下次雙擊載入即可續做,進度、印章設定、評語紀錄完整保留;遲交與補交作業可中途追加,新檔案附加至任務尾端、自動歸入同學號群組,徹底解決傳統批改流程「一旦中斷便難以接續」的痛點。
經 50份高中物理學習單實測,批改時間由 3 小時縮短至 1 小時,立可帶用量歸零,並穩定產出可直接上傳學習歷程的 A3 PDF 檔案。系統開發階段借助 AI 協作完成,但使用端不依賴網路、不消耗 token,任何教師雙擊即可啟用,具備零成本、零門檻、零依賴的推廣特性。

 


High school teachers routinely face a heavy worksheet grading workload. Processing a single batch — from sorting and evaluating to producing PDF learning portfolios — typically consumes several hours. Repetitive written feedback, correction fluid usage, and desk clutter not only drain teachers' energy but also compromise the visual quality of student portfolio records.
This study presents a Digital Worksheet Grading System, built on Python 3 with PyQt5, PyMuPDF, and SQLite. The grading workflow is divided into two parallel tracks: page organization (Track A) offers five-digit seat-number quick tagging, auto page-advancing, page rotation, and group merging; grading (Track B) integrates a visual stamp gallery, canned-comment shortcuts, and lossless A4-to-A3 vector composition.
The system's defining feature is its game-like save / load task management: teachers can interrupt grading at any moment and resume seamlessly with a single double-click, with progress, stamp settings, and comment records fully preserved. Late submissions and supplementary worksheets can be appended mid-task and are automatically grouped under the corresponding student ID, eliminating the long-standing pain point that "once interrupted, grading is hard to resume."
Field testing on 50 physics worksheets reduced grading time from 3 hours to 1 hour, eliminated correction fluid use, and produced consistently high-quality A3 PDFs ready for portfolio submission. While the development phase was supported by AI-assisted programming, the end-user application requires no network connection and consumes no tokens — any teacher can launch it with a single double-click, achieving zero-cost, zero-barrier, and zero-dependency adoption.

 

 

 

 

💬 大會即時客服