Natural
Language Understanding
Our research on natural language understating does not just aim at
providing a natural language interface to existing programs. Rather,
we attempt to model the mechanism of natural language comprehension by
a human being, and then simulate that mechanism on a computer. Some of
the research topic are:
- Representation of Human Intentions, Beliefs
- Sentence Generation
- Natural Language Parsing Algorithms
- Discourse Grammar/Dialogue Planning
- Natural Language Dialogue Systems
For example, most dialogue systems give no thought of what their users
believe. As a result, the responses of such systems are very limited,
and they can only communicate with their users in some fixed
situations. In order to communicate more intelligently in various
situations, a dialog system should talk account of his user's belief
and intention. In one of our researches, we present a computational
model of an intelligent agent, in which the agent communicates with a
person according to his own belief about (1) the situation where he is
communicating, (2) the person's belief and intention, and (3) his own
belief and intention.
|m 93.11.26