<?xml version="1.0" standalone="yes"?>
<Paper uid="P84-1037">
  <Title>MARS: A retrieval tool on the basis of Morphological Analysis. Proc. of the ACM Conference on Research and Development in</Title>
  <Section position="7" start_page="176" end_page="176" type="ackno">
    <SectionTitle>
ACKNOWLEDGEMENTS
</SectionTitle>
    <Paragraph position="0"> The work which was described here was done by A.Baumer, M.Streit, G.Thurmair (German), I.Buettel, G.Th.Niedermair, Ph.Hoole (English) and M.Meya (Spanish).</Paragraph>
    <Paragraph position="1"> Ill L~ITATIOgS Although the morphological decomposition works quite well and is useful with respect to information retrieval problems, there are some problems concerning the integration of such an algorithm into a whole natural language system. The reason is, that some information needed therefore is not easily available; this is the information which goes beyond morphology and is based on the interpretation of decomposition results. Two examples should be mentioned here. I. Parts of speech It is not easy to derive the part of speech of a word out of its decomposition. In German, the prefix VER- forms verb-derivations, but the derivation VER-TRAUEN from the verb TRAUEN is also a noun, whereas the same derivation VER-TRETEN form the verb TRETEN does not, and the derivation VER-LEGEN (from the verb LEGEN) is also an adJectiv. The past participle GE-F~I.U~ (from the verb FAI.L~N) is also a noun, the same derivation from LAUFEN (GE-LAUFEN) is not. This fact is due to the diachronic development of the language which led to a structure of the vocabulary that followed rather the needs of usage than morphological consistency.</Paragraph>
  </Section>
class="xml-element"></Paper>