teaching plans by LLM

Using the method of step-by-step inquiry, the prompt command components of each teaching link are constructed, and the teaching plan content is generated by combining zero sample and few sample prompt.mong them, in the teaching analysis part, the prompt command of teaching objectives adds the examples of core literacy and teaching objectives, and combines teaching key points and teaching difficulties into one inquiry. In the teaching process, each link will add the knowledge point of this lesson, the output format and the prompt instruction of the previous link. In addition, in the lead-in link before class, add the example, the problem chain of the small sample prompt example.In addition, if a result does not conform to the format requirements, we will rerun the query. After more than 1300 rounds of inquiries, we successfully generated a dataset of 21 chapters and 183 hours of high school math lesson plans, with a total of 605,377 words of text content.

采用分步探究法构建各教学环节的提示词指令组件，并结合零样本（Zero-shot）与少样本（Few-shot）提示词生成教案内容。其中，在教学分析环节，教学目标的提示词指令新增核心素养与教学目标示例，并将教学重点与难点整合为一项探究内容；教学过程中，各环节会补充本节课知识点、输出格式及前一环节的提示词指导；课前导入环节则新增少样本提示词示例及其问题链。此外，若生成结果不符合格式要求，将重新执行查询。经过1300余轮探究后，成功生成包含21章、183课时的高中数学教案数据集，文本内容总计605,377字。