The authors formalise and extend the contribution model of Clark and Schafer 1989 \ref{Clark89}. Particularly, the adaptation they propose is not just a computational reformulation of Clark’s model. The authors propose here some adaptations and addition. The first addition is to emphatise that all contribution graphs are private models, and can represent the perspective of only one agent. The second addition is the detail of the heuristic with which a simple computational agent applies its grounding criterion to the evidence in a turn. Also, they claim thath a contribution should be embedded only when the evidence it provides fails to meet the grounding criterion of the model holder. Finally they introduced a new structure to the formalism: the exchange, that is a pair of contributions linked by their complementary roles: the first proposes and the second executes a jointly achieved task.
Using this newly defined concept they tried to map contributions into tasks using the exchange as the minimal jointly determined dialog unit.
At the end the authors raise some questions like exactly what people represent and remember about a dialog, which is important in terms of understanding the impact of the personal view on the current status of the communication and secondly how to remap the grounding criterion to the current status of the communication.