CONFLICTING AGENTS IN DISTRIBUTED SEARCH
Youssef Hamadi · International Journal of Artificial Intelligence Tools · 2005
We extend here the work on interleaved distributed graph based backjumping, IDIBT/GBJ7,6,5 by considering conflicting agents.13 The resulting method IDIBT/CBJ-DkC for Conflict-directed Back-Jumbing and Directed k-Consistency combines distributed and parallel explorations of search spaces with more efficient backtrack steps. We take advantage of the asynchronous framework to implement a more refined and efficient update of conflicts. A fine analysis of these conflicts allows to some extent the detection of k-inconsistent values. Our experiments show a large improvement over graph based distributed/interleaved search.