Due to the growing importance of skills for proactively finding problems and solutions, many educational institutions provide student-centered exercises such as PBL. In particular, discussion skills are some of the most important for exchanging opinions with other students and summarizing one's own thoughts. For inexperienced students, gaining the experience necessary to understand the flow of discussions and speak precisely is important but difficult. Currently, discussions on social media have become popular, and this is a promising source that should be utilized as a reference for unskilled people. In this research, we aim to develop a mechanism for helping unskilled students gain discussion experience on social media. Specifically, we initially develop methods for extracting from Twitter discussions that have topics similar to a discussion of interest. Moreover, we develop methods for extracting discussion structures from accumulated utterances on the basis of an analysis of utterance-response relationships. On the basis of these methods, we finally develop a system for helping students gain such experience by visualizing discussion cases. In this paper, we mainly describe the methods for extracting cases along with an overview of the support provided. In addition, we describe an experiment using actual tweets and discuss the characteristics of our methods on the basis of the results.