<?xml version="1.0" standalone="yes"?>
<Paper uid="I05-2028">
  <Title>Modelling of a Gazetteer Look-up Component</Title>
  <Section position="6" start_page="165" end_page="165" type="concl">
    <SectionTitle>
5 Conclusions and Future Work
</SectionTitle>
    <Paragraph position="0"> In the context of modeling a compact data structure for implementing a gazetteer empirical experiments reveal that a pure-FSA approach, in which all data is converted into a single MADFSA, turns out to outperform the standard approach based on an indexing numbered automaton and an auxiliary table. At least in the case of data we are dealing with benefits are observable, since major part of the attribute values are contemporary word forms. A further investigation revealed that transition jamming reduces the size of the automata significantly. However, for storing gazetteers containing large number of (alpha)numerical data the standard approach or other techniques might be a better choice. Therefore, the evaluation results are only meant to constitute a handy guideline for selecting a solution. There are number of interesting issues that can be researched in the future, e.g. investigation of jamming paths of bounded length or deployment of finite-state transducers for handling the same task.</Paragraph>
  </Section>
class="xml-element"></Paper>