包头市防震减灾知识竞赛智能出题系统的设计与应用

Design and Application of Intelligent Question Generation System for Earthquake Prevention and Disaster Reduction Knowledge Competition in Baotou City

摘要: 针对防震减灾知识竞赛人工出题效率低下、试题标准不统一、难度缺乏动态校准、题库资源复用性差等痛点，依托包头地震信息服务平台，设计实现基于B/S架构的地震科普竞赛智能出题系统。系统采用PHP + MySQL轻量化技术栈，构建数据、逻辑、服务、表现四层解耦架构，搭建包含二级知识点标签、难度系数等字段的标准化结构化题库，搭建兼具题型、难度双重约束的智能组卷引擎，实现试题自动筛选、跨卷去重与批量快速组卷，建立五级难度自适应校准机制，通过滑动窗口汇总多场答题数据，依据正确率动态迭代更新试题难度系数。本文构建了一套面向资源有限的地方基层科普场景的轻量化智能出题闭环理论框架，小规模科普赛事的简易自适应评估算法，提出基于滑动窗口的多场次试题难度动态迭代校准方案，弥补现有科普出题系统缺少本地化自适应难度调节机制的研究空白，为市县级小型应急科普活动提供低成本、可快速部署、具备长效动态评估能力的标准化智能出题解决方案。该系统目前运行结果良好，可轻量化易部署，后续拓展可适配各类应急科普竞赛，为智慧防灾数字科普提供实践参考。

Abstract: Addressing the pain points of low efficiency, inconsistent question standards, lack of dynamic difficulty calibration, and poor reusability of question bank resources in earthquake prevention and disaster reduction knowledge competitions, this paper designs and implements an intelligent question generation system for earthquake science popularization competitions based on a B/S architecture, relying on the Baotou Earthquake Information Service Platform. The system adopts a lightweight PHP + MySQL technology stack, constructing a four-layer decoupled architecture of data, logic, service, and presentation. It builds a standardized structured question bank containing fields such as secondary knowledge point tags and difficulty coefficients, and establishes an intelligent question generation engine with dual constraints of question type and difficulty. This engine enables automatic question selection, cross-question deduplication, and batch rapid question generation. A five-level adaptive difficulty calibration mechanism is established, and multiple rounds of answer data are aggregated through a sliding window, dynamically updating the question difficulty coefficient based on the accuracy rate. This paper constructs a lightweight intelligent question-generating closed-loop theoretical framework for local grassroots science popularization scenarios with limited resources, proposes a simplified adaptive evaluation algorithm for small-scale science popularization competitions, and establishes a multi-session dynamic iterative calibration scheme for question difficulty based on a sliding window. This fills the research gap in existing science popularization question-generating systems, which lack a localized adaptive difficulty adjustment mechanism. It provides a standardized intelligent question-generating solution for small-scale emergency science popularization activities at the city and county levels that is low-cost, rapidly deployable, and possesses long-term dynamic evaluation capabilities. The system is currently running well, is lightweight and easy to deploy, and can be further expanded to adapt to various emergency science popularization competitions, providing practical reference for smart disaster prevention digital science popularization.