<?xml version="1.0" standalone="yes"?>
<Paper uid="W02-0508">
  <Title>A MORPHOLOGICAL, SYNTACTIC AND SEMANTIC SEARCH ENGINE FOR HEBREW TEXTS. UZZI ORNAN Visiting Professor, Computer Science, Technion - I.I.T. Scientific Director, Multitext, Multidimensional Publishing Systems</Title>
  <Section position="4" start_page="173" end_page="184" type="concl">
    <SectionTitle>
6. Conclusion
</SectionTitle>
    <Paragraph position="0"> The analysis and identification of correct readings of words in Hebrew script is far from being a simple task. The correct reading of any word is achieved only as a result of reading the rest of the words of the complete clause and sentence. We had to invest much work to solve this problem and to build a complex system of programs before we could have achieved a high-grade search engine, which is far better than other existing suggestions. See appendix for comparison.</Paragraph>
    <Paragraph position="1"> As noted above we succeeded in producing this engine only on the assumption that a Hebrew word must be read not on its own but in accordance with the reading of a complete sentence. We found that we achieved a powerful program, constituting a comprehensive infrastructure, for processing the Hebrew language for the computer. This infrastructure has already produced other results (the reading machine for the blind, see note 12), and it enables us to begin to work on automatic translation from Hebrew to other languages -- a task which has never yet been attempted.</Paragraph>
    <Paragraph position="2">  It seems that this method of translation by computer may be suitable to any language, and could be a contribution to translations from other languages.</Paragraph>
  </Section>
class="xml-element"></Paper>