从静态漫画到智能学伴——基于结构化标注的初中物理学习系统设计与原型实现

From Static Comics to an Intelligent Learning Companion—Design and Prototype Implementation of a Junior High School Physics Learning System Based on Structured Annotation

摘要: 学科漫画能够以连续画面和生活情境呈现抽象物理过程，但整图资源缺少分镜级知识、教学功能和题源关系，难以支持精确检索、学习诊断和反馈。本研究采用文献分析、系统设计与原型开发方法，提出“情境表征–语义结构–学习调节”范式，构建漫画单元与分镜项目两级标注模型，并据此实现初中物理智能学伴原型。系统将245张九格漫画拆解为2205个可定位分镜，组织3430道题目，使阅读、复习、闭图考察、提示、重试和薄弱分镜回看形成可追溯的数据链。研究表明，结构化标注能够把静态漫画转化为可定位、可检索和可诊断的教学资源；原型验证了数据链与业务规则的可运行性，但学习效果仍需在真实课堂中检验。

Abstract: Subject-specific comics can represent abstract physics processes through sequential images, but resources stored as whole pictures lack panel-level knowledge, pedagogical functions, and links to assessment items. This limits precise retrieval, diagnosis, and feedback. Using literature analysis, system design, and prototype development, this study proposes a framework of contextual representation, semantic structure, and learning regulation. It develops a two-level annotation model for comic units and individual panels and implements a junior high school physics learning companion. The prototype decomposes 245 nine-panel comics into 2205 addressable panels and organizes 3430 questions. Reading, review, closed-book assessment, prompts, retries, and panel-specific review are connected through traceable data relationships. Structured annotation can therefore turn static comics into locatable, retrievable, and diagnosable learning resources. The prototype demonstrates the operability of the data chain and business rules; improvements in learning outcomes still require classroom-based evaluation.