尽管计算机支持的协作学习（CSCL）在理论研究领域和教育实践领域已经取得了很大的进展，但研究者还不明确学习者在协作学习过程中所具有的角色，所处的地位和特征。调研发现，目前在用户角色研究领域大都用社会网络分析的方法，主观假设角色种类的多样性再验证。这种方法忽略了言论内容本身，导致角色种类有所限制。本文利用数据挖掘的方法来探索CSCL中的用户角色，以学生在《信息技术与教育》在线课程的异步讨论区中的交互作为研究对象，通过内容分析法对学生的言论数据进行编码，并运用聚类分析中的K-Means算法，挖掘出此案例的交互过程中包含咨询者、参与讨论者、创新者、贡献者、论证者五种用户角色。

Although computer supported collaborative learning (CSCL) have made great progress in the field of theoretical research and education practice areas, researchers do not explicitly know learners’ roles in the collaborative learning process, and in which the status and characteristics. At present, there are many researchers who focus on analyzing learner roles based on collaborative learning activities. But most of them subjectively presume the diversity of learners’ roles in advance and then verify it based on the statistics of interaction data. This approach ignores the content of the speech, and restricts the kinds of learners’ roles. To the contrary, this paper adopts the data mining technology to explore learners’ roles in collaborative learning. According to a complete data mining framework, data preparation and learner discourse pattern mining are depicted. Furthermore, a case study is conducted to show the mining process and finding as well as discussion on the mining results.