<?xml version="1.0" standalone="yes"?>
<Paper uid="W04-2415">
  <Title>Hierarchical Recognition of Propositional Arguments with Perceptrons</Title>
  <Section position="6" start_page="0" end_page="0" type="concl">
    <SectionTitle>
5 Experiments and Results
</SectionTitle>
    <Paragraph position="0"> We have build a system which implements the presented architecture for recognizing arguments and their semantic roles. The configuration of learning functions, related to the roles in the CoNLL-2004 data, is set as follows : Five score functions for the A0-A4 types, and two shared filtering functions FANS and FANE .</Paragraph>
    <Paragraph position="1"> For each of the 13 adjunct types (AM-*), a score function and a pair of filtering functions.</Paragraph>
    <Paragraph position="2"> Three score functions for the R0-R2 types, and two filtering functions FRS and FRE shared among them.</Paragraph>
    <Paragraph position="3"> For verbs, a score function and an end filter.</Paragraph>
    <Paragraph position="4"> We ran the learning algorithm on the training set (with predicted input syntax) with a polynomial kernel of degree 2, for up to 8 epochs. Table 1 presents the obtained results on the development set, either artificial or real. The second and third rows provide, respectively, the loss suffered because of errors in the filtering and scoring layer. The filtering layer performs reasonably well, since 89.44% recall can be achieved on the top of it.</Paragraph>
    <Paragraph position="5"> However, the scoring functions clearly moderate the performance, since working with perfect start-end functions only achieve an F1 at 75.60. Finally, table 2 presents final detailed results on the test set.</Paragraph>
    <Paragraph position="6">  tions with prefix g are gold functions, providing bounds of our performance. The top row is the upper bound performance of our architecture. The bottom row is the real performance.</Paragraph>
  </Section>
class="xml-element"></Paper>