First page Back Continue Last page Overview Text
Using finite state machine we can model the dynamics of the interaction – and unrolling the speaker state machine gives us essentially a variable duration HMM. We can model the effects of the other participants using a modified, coupled hidden markov model with influence parameters which was actually initiailly used to model cascading failures in the power grid.
These types of models will hopefully allow us to go beyond making similar correlations between office location and communication probabilities, and instead look deeper in the causes of the behavior – giving us a sense of the global sensitivities within the organization – and simulation of organizational disruption.
holding the floor
Proximity (w w/o speaking)