<?xml version="1.0" standalone="yes"?>
<Paper uid="X98-1011">
  <Title>EXTRACTING AND NORMALIZING TEMPORAL EXPRESSIONS</Title>
  <Section position="2" start_page="0" end_page="0" type="intro">
    <SectionTitle>
1. INTRODUCTION
</SectionTitle>
    <Paragraph position="0"> As part of our TIPSTER III research program, we have enhanced the NLToolset's ~ capability to extract temporal expressions from free text and convert them into canonical form for accurate comparison, sorting, and retrieval within a database management system.</Paragraph>
    <Paragraph position="1"> The date or time that an event occurs is often a critical piece of information. Unfortunately, natural language expressions that contain this information are so numerous and varied that the interpretation of temporal expressions within free text becomes a challenging task for automatic text processing systems.</Paragraph>
    <Paragraph position="2"> This paper will look at the nature of the problem, the extraction and computation tasks, the use of a learning program, and the normalization strategy. The concluding section will discuss possible future endeavors related to time extraction.</Paragraph>
    <Paragraph position="3"> The NLToolset The NLToolset is a framework of tools, techniques, and resources designed for building text processing applications. It is a pattern based system which uses world knowledge resident in a lexicon, a location gazetteer, and lists of universal terms, such as first names and the Fortune 500 companies. This knowledge base is extensible with generic, as well as domain-specific, information. It applies lexico-semantic pattern matching in the form of basic structural patterns (possible-title firstname middle-J The NLToolset is a proprietary text processing product, owned by Lockheed Martin Corporation.</Paragraph>
    <Paragraph position="4"> initial lastname), as well as contextual knowledge (possible-name, who is X years old). The NLToolset has been applied to routing, indexing, name spotting, information extraction, and document management.</Paragraph>
    <Paragraph position="5"> It is an object-oriented system, implemented in C++ and ODBC to make it portable to both Unix and NT platforms, as well as multiple databases.</Paragraph>
  </Section>
class="xml-element"></Paper>