chapter  6
32 Pages

Probabilistic Message Passing

Semantic Gossiping (Chap. 4) has introduced novel methods relying on the analysis of mapping closures in order to detect semantic disagreement as well as to selectively forward queries in a network of heterogeneous parties. Such methods suffer, however, from certain issues that limit their applicability and performance in practice:

(1) In Semantic Gossiping, a peer performs its analysis autonomously and independently of the other peers. Peers that form a semantic cycle all examine the cycle in isolation without exchanging or reusing the computations of the other peers in the cycle.