<?xml version="1.0" standalone="yes"?>
<Paper uid="W05-0101">
  <Title>Teaching Applied Natural Language Processing: Triumphs and Tribulations</Title>
  <Section position="8" start_page="6" end_page="7" type="concl">
    <SectionTitle>
7 Conclusions
</SectionTitle>
    <Paragraph position="0"> Overall, I feel the main goals of the course were met.</Paragraph>
    <Paragraph position="1"> Although I am emphasizing how the course could be improved, most students were quite positive about the class, giving it an overall score of 5.8 out of 7 with a mode of 6 in their anonymous course reviews.</Paragraph>
    <Paragraph position="2"> (This is on the low side for my courses; most who gave it low scores found the programming too difficult.) null Most students found the material highly stimulating and the work challenging but not overwhelming. Several students mentioned that a lab session with a dedicated TA would have been desirable. Several suggested covering less material in more depth and several commented that the Enron exercise was a neat idea although not entirely successful in execution. Students remarked on liking reading research papers rather than a textbook (they also liked the relatively light reading load, which I feel was appropriate given the heavy assignment load). Some students  wanted more emphasis on real-world applications; I think it would be useful to have guest speakers from industry talk about this if possible.</Paragraph>
    <Paragraph position="3"> I would like to see more research tools developed to a point to which they can be applied more successfully, especially in the area of information extraction. I would also recommend to colleagues that careful control be retained over assignments and projects to ensure feasibility in the outcome. It is more difficult to get good results on class projects in NLP than in other areas I've taught. As we so often see in text analysis work, it can often be difficult to do better than simple word counts for many projects.</Paragraph>
    <Paragraph position="4"> I am interested in hearing ideas about how to accommodate both the somewhat technical and the highly technical students, especially in the early parts of the course. Perhaps the best solution is to offer an optional laboratory section, at least for the first few weeks, but perhaps for the entire term, but this solution obviously requires more resources.</Paragraph>
    <Paragraph position="5"> When designing this course I did a fairly extensive web search looking for courses that offered what I was interested in, but didn't find much. I used the proceedings of the ACL-02 workshop on teaching NLP (where I learned about NLTK) as well as the NLP Universe. I think it would be a good idea to start an archive of teaching resources; ACM SIGCHI is in the midst of creating such an educational digital library and this example is worth studying.9</Paragraph>
  </Section>
class="xml-element"></Paper>