File Information

File: 05-lr/acl_arc_1_sum/cleansed_text/xml_by_section/concl/05/w05-0104_concl.xml

Size: 1,368 bytes

Last Modified: 2025-10-06 13:54:57

<?xml version="1.0" standalone="yes"?>
<Paper uid="W05-0104">
  <Title>A Core-Tools Statistical NLP Course</Title>
  <Section position="6" start_page="26" end_page="26" type="concl">
    <SectionTitle>
5 Conclusions
</SectionTitle>
    <Paragraph position="0"> There are certainly changes I will make when I teach this course again this fall. I will likely shuffle the topics around so that word alignment comes earlier (closer to HMMs for tagging) and I will likely teach dynamic programming solutions to parsing and tagging in more depth than graph-search based methods. Some students needed remedial linguistics sections and other students needed remedial math sections, and I would hold more such sessions, and ear3There was also verbose error reporting for assignment 4, which displayed each sentence's guessed and gold alignments in a grid, butsince most studentsdidn't speak French, this didn't have the same effect.</Paragraph>
    <Paragraph position="1"> lier in the term. However, I will certainly keep the substantial implementation component of the course, partially in response to very positive student feedback on the assignments, partially from my own reaction to the high quality of student work on those assignments, and partially from how easily students with so much hands-on experience seem to be able to jump into NLP research.</Paragraph>
  </Section>
class="xml-element"></Paper>
Download Original XML