<?xml version="1.0" standalone="yes"?>
<Paper uid="W04-0914">
  <Title>Semantic Forensics: An Application of Ontological Semantics to Information Assurance</Title>
  <Section position="9" start_page="0" end_page="0" type="concl">
    <SectionTitle>
8 Conclusion
</SectionTitle>
    <Paragraph position="0"> The main thrust of the paper has been not so much the establishment of a sexy application as to demonstrate that the rich resources of NLP, in general, and ONSE, in particular, are versatile enough to be extended to interesting new uses and that getting there involves theoretical and methodological developments that are generally good for the field rather than just for SF (e.g., who will refuse a microtheory of euphemisms?).</Paragraph>
    <Paragraph position="1"> Throughout, we have insinuated, ever so subtly, that the tasks in hand are not manageable by any of the past or current meaning-avoiding, nonrepresentational approaches. This is not to say that a good SF NLP system must be statistics-free: Crude measures are good to have for heuristic and other startup purposes--but it is TMR elements that such statistics will be counting. We have left out many aspects of SF, such as potential demand, which is great, and other practical considerations.</Paragraph>
    <Paragraph position="2"> As resources permit, we have been moving consistently to enrich our ONSE resources with IAS capabilities and functionalities, and SF is the latest but, very probably, not the last of those.</Paragraph>
  </Section>
class="xml-element"></Paper>