How to efficiently update the state of an application after the user corrects false data about past events?

Suppose I'm making a voice-controlled application where spoken utterances will occasionally be misrecognized. I want the user to be able to occasionally correct misrecognized utterances and have the state of application, including the state of the GUI, correct itself as efficiently as possible. Are there existing software frameworks or design patterns that would simplify this? I've looked into self-adjusting computation, but it doesn't seem to be geared towards data that streams in constantly.