tech,6-1-P05-1073,ak Despite much recent progress on accurate <term> semantic role labeling </term> , previous work has largely used
tech,16-1-P05-1073,ak work has largely used independent <term> classifiers </term> , possibly combined with separate
model,22-1-P05-1073,ak </term> , possibly combined with separate <term> label sequence models </term> via <term> Viterbi decoding </term> .
tech,26-1-P05-1073,ak <term> label sequence models </term> via <term> Viterbi decoding </term> . This stands in stark contrast to
other,11-2-P05-1073,ak the linguistic observation that a <term> core argument frame </term> is a joint structure , with strong
other,21-2-P05-1073,ak is a joint structure , with strong <term> dependencies </term> between <term> arguments </term> . We
other,23-2-P05-1073,ak strong <term> dependencies </term> between <term> arguments </term> . We show how to build a joint <term>
model,7-3-P05-1073,ak </term> . We show how to build a joint <term> model </term> of <term> argument frames </term> , incorporating
other,9-3-P05-1073,ak build a joint <term> model </term> of <term> argument frames </term> , incorporating novel <term> features
other,14-3-P05-1073,ak frames </term> , incorporating novel <term> features </term> that model these interactions into
model,20-3-P05-1073,ak that model these interactions into <term> discriminative log-linear models </term> . This system achieves an <term> error
measure(ment),4-4-P05-1073,ak models </term> . This system achieves an <term> error reduction </term> of 22 % on all <term> arguments </term>
other,11-4-P05-1073,ak error reduction </term> of 22 % on all <term> arguments </term> and 32 % on <term> core arguments </term>
other,16-4-P05-1073,ak all <term> arguments </term> and 32 % on <term> core arguments </term> over a state-of-the art independent
tech,23-4-P05-1073,ak over a state-of-the art independent <term> classifier </term> for <term> gold-standard parse trees
other,25-4-P05-1073,ak independent <term> classifier </term> for <term> gold-standard parse trees </term> on <term> PropBank </term> . Previous
lr-prod,29-4-P05-1073,ak gold-standard parse trees </term> on <term> PropBank </term> . Previous work has used <term> monolingual
skrij podrobnosti