<?xml version="1.0" standalone="yes"?>
<Paper uid="W03-1017">
  <Title>Towards Answering Opinion Questions: Separating Facts from Opinions and Identifying the Polarity of Opinion Sentences</Title>
  <Section position="10" start_page="0" end_page="0" type="concl">
    <SectionTitle>
9 Conclusions
</SectionTitle>
    <Paragraph position="0"> We presented several models for distinguishing between opinions and facts, and between positive and negative opinions. At the document level, a fairly straightforward Bayesian classifier using lexical information can distinguish between mostly factual and mostly opinion documents with very high precision and recall (F-measure of 97%). The task is much harder at the sentence level. For that case, we described three novel techniques for opinion/fact classification achieving up to 91% precision and recall on the detection of opinion sentences. We also examined an automatic method for assigning polarity information to single words and sentences, accurately discriminating between positive, negative, and neutral opinions in 90% of the cases.</Paragraph>
    <Paragraph position="1"> Our work so far has focused on characterizing opinions and facts in a generic manner, without examining who the opinion holder is or what the opinion is about. While we have found presenting information organized in separate opinion and fact classes useful, our goal is to introduce further analysis of each sentence so that opinion sentences can be linked to particular perspectives on a specific subject. We intend to cluster together sentences from the same perspective and present them in summary form as answers to subjective questions.</Paragraph>
  </Section>
class="xml-element"></Paper>