What Language Understanding Research Suggests About Distributed Artificial Intelligence,
SRI INTERNATIONAL MENLO PARK CA ARTIFICIAL INTELLIGENCE CENTER
Pagination or Media Count:
This article speculated on the possible relevance of the artificial intelligence approach to natural language understanding to the problem of communication between processors in a distributed artificial intelligence system. While the processing required to interpret a natural language is almost certainly far more than a DAI system needs, a simplified form of this processing that performs all or most of the tasks performed in understanding natural language may well be required. This suggests that designers of DAI systems might best design their communications protocols by narrowing down the capabilities required for natural language communication rather than by building up from the traditional, individual-message-oriented approach. The key point is that, to develop a robust distributed system that is able to communicate over a noisy, relatively narrow-bandwidth channel, one must consider not just the format and content of the individual messages that are used to communicate, but also the role that sequences of messages play in the overall process of communication.