other,10-2-A92-1026,ak <term> TACITUS </term> ; especially in the <term> MUC-3 evaluation </term> , has shown that principled techniques
other,19-4-A92-1026,ak </term> is inherently robust , in that an <term> interpretation </term> is always possible , so that in the
other,21-3-A92-1026,ak recovery technique </term> for failed <term> parses </term> , and a new technique called <term>
other,22-1-A92-1026,ak of aiming for complete and correct <term> interpretations </term> has to be abandoned . However , our
other,30-2-A92-1026,ak bolstered with methods for achieving <term> robustness </term> . We describe three techniques for
other,32-4-A92-1026,ak that in the absence of the required <term> world knowledge </term> , performance degrades gracefully
tech,1-4-A92-1026,ak terminal substring parsing </term> . For <term> pragmatics processing </term> , we describe how the method of <term>
tech,10-4-A92-1026,ak </term> , we describe how the method of <term> abductive inference </term> is inherently robust , in that an
tech,12-3-A92-1026,ak analysis </term> more robust --- an <term> agenda-based scheduling parser </term> , a <term> recovery technique </term>
tech,17-3-A92-1026,ak agenda-based scheduling parser </term> , a <term> recovery technique </term> for failed <term> parses </term> , and
tech,19-2-A92-1026,ak shown that principled techniques for <term> syntactic and pragmatic analysis </term> can be bolstered with methods for
tech,28-3-A92-1026,ak </term> , and a new technique called <term> terminal substring parsing </term> . For <term> pragmatics processing </term>
tech,6-1-A92-1026,ak </term> . It is often assumed that when <term> natural language processing </term> meets the real world , the ideal
tech,6-3-A92-1026,ak describe three techniques for making <term> syntactic analysis </term> more robust --- an <term> agenda-based
tool,5-2-A92-1026,ak abandoned . However , our experience with <term> TACITUS </term> ; especially in the <term> MUC-3 evaluation
hide detail