This paper describes the current version of state-event transformation system to transform story lines and story worlds each other using a knowledge base and a conceptual dictionary. Moreover, we extend the basic framework to a circulative generation mechanism with a simple mutation function. Through the preliminary performance checks, we confirmed that the transformation of story worlds and story lines is approximately logically adequate and the circular process produces the diversity of stories. The proposed system is a module in an integrated narrative generation system and the pilot version is already implemented. In the context of the narrative generation system, the proposed system plays roles for expanding the variation of discourse to be generated and limiting possible narrative elements at any given time in a narraive generation process.