is based on a finite state machine (FSM) where each state represents a certain shot. Each transition to another state is therefore a cut, blend etc.

In contrast to other approaches ther are two major differences:

1. We introduced three contexts: standard lecture mode, question mode and answer mode.
2. Our transitions do not have fixed weights, but condition objects and belonging conditions, e.g., <Slide>annotating</Slide>.

Therefore, our approach has some advantages:

At first, the weights of the transitions or their propabilities can vary depending on the sensor determining that, for example, a slide is annotated to a certain extent.

Second, a transition, e.g., from lecturer to slide, may have different condition objects and conditions in the different contexts.

For some more details click here; to give an overview, please see this simplified example of an FSM: