With the advancement of generative AI, research on dialogue generation has rapidly progressed. This dialogue generation not only enables interactions with fictional AIs but also facilitates dialogues with historical people, providing new avenues for communication. When generating dialogues with historical people, it is essential to consider their philosophies, which reflect their thoughts and life views. Currently, dialogue generation that considers personas has been realized; however, personas merely represent the personality or character of the dialogue partner and differ from philosophies. Therefore, this study proposes a method for dialogue generation that considers historical people’s philosophies. As an initial step, this paper proposes a method for constructing training data that considers these philosophies.