Gossiping in a graph is a topic in information dissemination. Each point in the graph has a message to be distributed to all the other points. There are several models of gossiping in a graph. Each is based on the idea of a ``call'' between points. Calls may only be made between two points which share an edge. In traditional gossiping each call consists of all information known by the sender being transmitted to the receiver(s). In cumulative gossiping each call consists of one of the messages known by the sender. Models also vary as to whether a call may involve one way or two way ( usally specified as the H and F models, respectively) transmission of information. In the F model, both participants in the call are senders and receivers. In the H model only one participant is a sender. Other variants allow a sender to send to all its neighbors, and in the F model to receive from all its neighbors at the same time.
The traditional approach is to find algorithms which successfully complete gossiping as quickly as possible and to relate these to the theoretical lower bounds for doing so in the selected model. The goal is of course to find algorithms which allow gossiping to complete in the optimal time.