<?xml version="1.0" standalone="yes"?>
<Paper uid="X98-1022">
  <Title>AN NTU-APPROACH TO AUTOMATIC SENTENCE EXTRACTION FOR SUMMARY GENERATION</Title>
  <Section position="8" start_page="166" end_page="166" type="concl">
    <SectionTitle>
6. CONCLUDING REMARKS
</SectionTitle>
    <Paragraph position="0"> This paper proposes models to generate summary for two different applications. The first is to produce generic summaries, which do not take the user's information need into account. The second is to produce summaries, while the user's information need is an important issue. That is to say, the automatic summarization system interacts with users and takes user's query as a clue to produce user-oriented summaries. In addition, our approach is extract-based, which generates summaries using the sentences extracted from original texts. For the categorization task, the positive feature vector and the negative feature vector trained from the SUMMAC-1 texts are used as the comparative basis for sentence selection to produce generic summaries.</Paragraph>
    <Paragraph position="1"> As to adhoc task, the ES of each sentence is calculated based on the interaction of nouns and verbs. Then, the nouns of a query are compared with nouns in sentences and the closely related sentences are selected to form the summary. The result shows that the NormF of the best summary and that of the fixed summary for adhoc tasks are 0.456 and 0.447, respectively. The NorrnF of the best summary and that of the fixed summary for categorization task are 0.4090 and 0.4023, respectively. Our system outperforms the average system in categorization task but does a common job in adhoc task. We think that there are many further works to be studied in the future, e.g., extending the proposed approach to other languages, optimizing parameters of the proposed model, investigating the impact of errors introduced in tagging step, and developing a appropriate method to setup the threshold for sentence selection.</Paragraph>
  </Section>
class="xml-element"></Paper>