<?xml version="1.0" standalone="yes"?>
<Paper uid="H93-1100">
  <Title>A Knowledge-based Machine Assisted PROJECT GOALS The Computing Research Laboratory (CRL) at New</Title>
  <Section position="2" start_page="0" end_page="0" type="abstr">
    <SectionTitle>
RECENT RESULTS
</SectionTitle>
    <Paragraph position="0"> At this p6int, a first version of the year-two analysis system has been completed and is undergoing expansion and refinement. The initial ontology has been constructed, and a large number of sense-tokens have been incorporated. A phrasal lexicon has been extracted from on-line dictionary resources and is being prepared for incorporation into the Translator's WOrkstation.</Paragraph>
    <Paragraph position="1"> The second year analysis sytem begins with a dictionary-based part-of-speech tagger, followed by a component which chunks the tagged text into small syntactic sections. These are analyzed and semantic/lexical information accessed and incorporated into the representation by a constituent parser. These smaller constituents will be grouped into predicates and arguments, which are then further grouped into clausal structures. At each level, possible readings are rated for syntactic and semantic likelihood.</Paragraph>
    <Paragraph position="2"> With respect to automatic acquisition, several advances have occurred in the past year. We have provided the ontology with a sense-disambiguated hierarchy of nominal word senses drawn from PSongman's Dictionary of Contemporary English. The hierarchy is derived from disambiguated genus terms and rooted in the semantic categories provided by Longman's. We have gathered from our bilingual dictionaries a large number of phrases correlated with translations for that phrase. Verb classes for Spanish verbs, used as the basis for the morphological analysis program that feeds the part-of-speech tagger, were provided by Collin's Spanish-English/English-Spanish Dictionary. In a final note, the CRL has contracted with EFE (Spanish newswire service) for a continuous line feed.</Paragraph>
  </Section>
class="xml-element"></Paper>