<?xml version="1.0" standalone="yes"?>
<Paper uid="P02-1021">
  <Title>Semi-Supervised Maximum Entropy Based Approach to Acronym and Abbreviation Normalization in Medical Texts</Title>
  <Section position="6" start_page="7" end_page="7" type="metho">
    <SectionTitle>
4 Testing
</SectionTitle>
    <Paragraph position="0"> To summarize the goals of this study, one of the main questions in this study is whether local sentence-level context can be used successfully to disambiguate abbreviation expansion. Another question that naturally arose from the structure of the data used for this study is whether more global section-level context indicated by section headings such as &amp;quot;chief complaint&amp;quot;, &amp;quot;history of present illness&amp;quot; , etc., would have an effect on the accuracy of predicting the abbreviation expansion. Finally, the third question is whether it is more beneficial to construct multiple ME models limited to a single abbreviation. To answer these questions, 4 sets of tests were conducted:  1. Local Context Model and Set A 2. Combo Model and Set A 3. Local Context Model and Set B 4. Combo Model and Set B</Paragraph>
  </Section>
class="xml-element"></Paper>