Investigating the code fragments of code clones detected by code clone detection tools is a time-consuming task, especially when a large number of reference source files are available. This paper proposes (i) a method for clustering a clone class, which is detected by code clone detection tools using syntactic similarity, based on topic similarity by considering its code fragments as sequences of words and (ii) a method for assigning short tags to clusters of the clustering result. We also report an experiment of applying the proposed method to packages of an open source operating system.