Recently, chatbot support has attracted attention in the field of customer support. However, it has a heavy burden of this work because the chatbot operation needs many manual and difficult works. One of the bottlenecks of operation work is that it is difficult to extract new queries from dialogue logs due to the distribution of similar queries. In this study, we introduce the reduction of operation work when referring to the dialogue logs by clustering dialogue. This has resulted in a significant reduction in operation time.