随着新一代信息技术的快速发展，快节奏的生活和多任务处理使得学习时间越来越碎片化，学习间隔时间规划也越来越重要。将学习任务分布在不同学习间隔时间内完成的学习方法被称为分散学习。由于在课程中师生很难确定最优学习间隔时间，因此研究不同知识类型的最优学习间隔时间，是应用分散学习策略提升学生学习效果的重中之重。本研究以C语言程序设计为例设计了在线编程系统，支持选择、填空和编程题，采集学生编程的完整代码轨迹，以研究分散学习的最优学习间隔时间。

分散学习的情境背景假说强调背景重构可以生成有效的线索，影响了学生在掌握情况下的提取失败概率和在未掌握情况下的猜测概率。因此本研究以隐马尔科夫模型为基础，根据学生的学习时间间隔、答案结果等可观察状态推断出学生的隐性知识掌握及其转移的状态。经课程数据验证，在C语言程序设计三类知识的AUC指标在0.6以上。对于编程初学者来说，研究发现概念型知识适合中等长度的时间间隔，程序型知识适合较短的时间间隔，综合应用型知识适合较长的时间间隔。

本研究通过滞后序列分析学习间隔时间对学生的学习行为模式的影响，以解释学习间隔时间的作用机制。根据学习间隔时间的积累效应，通过不同学习间隔组行为模式和单组行为模式发展对比，研究发现1.在练习时，学生反复查看答案和查看答案解析，更好的帮助学生建立题目与答案的联系。2.在练习时，回顾、查看答案解析与调试之间的循环。学生通过个人努力的方式不断调试以获得答案，丰富了学习情境。同时本研究进一步考察分散学习对于编程学习的促进效应。研究发现对于编程初学者，分散学习能够促进学生的计算思维概念和计算思维观念，在计算思维实践上效果不明显。

综上所述，研究认为分散学习是一种有意义的学习方式。在这种学习方式中，学习间隔时间会改变学习情境。而在学习情境中各类学习行为则在改变知识点提取难度的同时也在丰富着情景的变化，进而促进或抑制知识提取和情境重构。经过一段时间的积累，最终影响学习效果。本研究验证了分散学习培养计算思维的有效性，解释了学习间隔时间对编程学习的作用机制，解决根据不同知识点最优化学习时间探索的问题，积极推动分散学习在教育教学中的应用。

With the rapid development of the new generation of information technology, fast-paced life and multitask processing make learning time more and more fragmented, so the planning of learning interval is more and more important. The learning method that distributes learning tasks in different learning intervals is called distributed learning. Because it is difficult for teachers and students to determine the optimal learning interval in the course, it is the most important to study the optimal learning interval of different knowledge types to improve the learning effect by applying the distributed learning. In order to study the optimal learning interval, the online programming system of C language programming is designed to support the selection, filling in the blank and programming, and to collect the complete code trace of students’ programming.

The contextual hypothesis of decentralized learning emphasizes that effective learning cues can be generated through background reconstruction, which affects the extraction failure probability with mastering and the guessing probability without mastering. Therefore, based on the Hidden Markov model, this study infers the state of students’ tacit knowledge mastering and transfer according to their learning interval and answer results. After the course data verification, the AUC of the three types of knowledge in C language programming is above 0.6. For beginners in programming, conceptual knowledge is suitable for medium-length time interval. Procedural knowledge is suitable for shorter time interval. And comprehensive applied knowledge is suitable for longer time interval.

In order to explain the mechanism of learning time interval, the study analyzes the effect of learning time interval on students’ learning behavior patterns by lag sequence. According to the cumulative effect of the learning time interval, this study compares the development of different learning interval group behavior patterns and single group behavior patterns. 1. During exercises, students repeatedly view the answers and the explanation to better help students to establish the connection between the questions and the answers. 2. During exercises, students are looping between debugging, viewing explanation and review. Students continue to debug through personal efforts to find answers, which enriches the learning situation. At the same time, this study further investigates the effect of distributed learning on programming learning. For beginners of programming, distributed learning can promote students’ computational thinking concepts and computational thinking concepts, which is not obvious in the practice of computational thinking.

In summary, the study believes that distributed learning is a meaningful learning method. In this learning method, the learning interval changes the learning context. And various learning behaviors change the difficulty of extracting knowledge points while enriching the changes the context. After a period of accumulation, it will eventually affect the learning effect. This study verifies the effectiveness of distributed learning in cultivating computational thinking, and explains the mechanism of learning interval on programming learning. This study solves the problem of optimizing the learning time exploration method according to different knowledge points, and actively promotes the application of distributed learning in education and teaching.