<?xml version="1.0" standalone="yes"?>
<Paper uid="P98-1080">
  <Title>Tagging Inflective Languages: Prediction of Morphological Categories for a Rich, Structured Tagset</Title>
  <Section position="8" start_page="488" end_page="489" type="concl">
    <SectionTitle>
6 Conclusion and Further Research
</SectionTitle>
    <Paragraph position="0"> The combined error rate results are still far below the results reported for English, but we believe that there is still room for improvement. Moreover, splitting the tags into subtags showed that &amp;quot;pure&amp;quot; part of speech (as well as the even more detailed &amp;quot;subpart&amp;quot; of speech) tagging gives actually better results than those for English.</Paragraph>
    <Paragraph position="1"> We see several ways how to proceed to possibly improve the performance of the tagger (we are still talking here about the &amp;quot;single best tag&amp;quot; approach; the n-best case will be explored separately): * Disambiguated tags (in the left context) plus Viterbi search. Some errors might be eliminated if features asking questions about the disambiguated context are being used. The disambiguated tags concentrate - or transfer - information about the more distant context. It would avoid &amp;quot;repeated&amp;quot; learning of the same or similar features for different but related disambiguation problems. The final effect on the overall accuracy is yet to be seen. Moreover, the transition function assumed by the Viterbi algorithm must be reasonably defined (approximated). null * Final re-estimation using maximum entropy.</Paragraph>
    <Paragraph position="2"> Let's imagine that after selecting all the features using the training method described here we recompute the feature weights using the usual maximum entropy objective function. This will produce better (read: more principled) weight estimates for the features already selected, but it might help as well as hurt the performance.</Paragraph>
    <Paragraph position="3"> * Improved feature pool. This is, according to our opinion, the source of major improvement.</Paragraph>
    <Paragraph position="4"> The error analysis shows that in many cases the 9No overtraining occurred here either, but the results for thresholds 2-4 do not differ significantly.</Paragraph>
    <Paragraph position="5"> ldegFor English, using the Penn 23&amp;quot;eebank data, we have always obtained better accuracy using the VTC method (and redefinition of the tag set based on 4 categories).</Paragraph>
    <Paragraph position="6">  context to be used for disambiguation has not been used by the tagger simply because more sophisticated features have not been considered for selection. An example of such a feature, which would possibly help to solve the very hard and relatively frequent problem of disambiguating between nominative and accusative cases of certain nouns, would be a question &amp;quot;Is there a noun in nominative case only in the same clause?&amp;quot; - every clause may usually have only one noun phrase in nominative, constituting its subject. For such feature to work we will have to correctly determine or at least approximate the clause boundaries, which is obviously a non-trivial task by itself.</Paragraph>
  </Section>
class="xml-element"></Paper>