Agents in peer-to-peer networks may gather into virtual communities, interacting continuously with agents that represent disparate actors, each of them with different interests, needs and views, and having dissimilar ontologies. Mapping all the combination of ontologies in advance is not feasible simply because all the possible combinations cannot be foreseen. Mapping complete ontologies at run time is a computationally expensive task. The framework proposed in this paper maps only the terms encountered in a dialogue, or those needed to map them. The efficiency in the mapping process is increased by accumulating experience and exploiting it in order to reduce the number of mapping candidates to verify, focusing only on the most likely ones.
|Title of host publication||2nd International Workshop on Peer to Peer Knowledge Management|
|Number of pages||9|
|Publication status||Published - 1 Jul 2005|
|Name||CEUR Workshop Proceedings|