1.4.3 Grounding

There are two modules in TRAINS to make sure that the input is correctly interpreted.

There are two modules in TRAINS, the post-parser and the chart parser, that are involved in making sure that the input is correctly understood and interpreted. The system offers understanding confirmation. It displays the relevant domain objects whose employment the user has asked for, as well as the ones that the system judges need to be involved in order to achieve the user's goal. That gives the opportunity to the user to intervene in case of wrongly interpreted input. Since the system is highly involved into solving the problem and tries to make relevant suggestions, any wrong interpretation would soon become apparent to the user, anyway.

See TRAINS dialogue example (Section 1.3.3); In [7] the user makes a suggestion, in [8] the system shows that it understands the suggestion (``OK''). Then it criticises the suggestion based on what it thinks the goal is, i.e., that the transport should rather be faster than slower. In [9] the user changes the original suggestion due to the criticism.


Dimitra Tsovaltzi, Stephan Walter and Aljoscha Burchardt
Version 1.2.5 (20030212)