needs a <term> computational lexicon </term> , each <term> system </term> puts different amounts
in lateral or longitudinal directions . Each <term> character </term> has its own width
</term> , performance degrades gracefully . Each of these techniques have been evaluated
<term> generalized metaphor mappings </term> . Each <term> generalized metaphor </term> contains
</term> and the <term> target speaker </term> . Each <term> reference model </term> is transformed
single <term> understandingresult </term> after each <term> user utterance </term> . By holding
</term> of its <term> search space </term> . As each new <term> edge </term> is added to the <term>
backgrounds </term> , or <term> goals </term> , at each point in a <term> conversation </term> , difficulties
properties , and relations that are salient at each point of the <term> discourse </term> . The
</term> created by the <term> system </term> during each <term> question answering session </term> .
candidate <term> antecedents </term> and to evaluate each other 's proposals . This paper discusses
mapping </term> is estimated independently for each <term> training ( reference ) speaker </term>
a set of <term> candidate parses </term> for each input <term> sentence </term> , with associated
strength of <term> antecedence recovery </term> for each of the <term> lexical substitutions </term>
discourse-relevant purposes </term> , expressed in each of the <term> linguistic segments </term> as
with similar <term> context features </term> in each author 's <term> corpus </term> tend not to
strength of potential antecedence </term> of each element in the <term> text </term> to select
</term> the system has in the correctness of each <term> extracted field </term> . The <term> information
order to take advantage of the strengths of each . Applications of <term> path-based inference
and a line length is counted by the sum of each <term> character </term> . By using commands
hide detail