<?xml version="1.0" standalone="yes"?>
<Paper uid="W04-3209">
  <Title>Comparing and Combining Generative and Posterior Probability Models: Some Advances in Sentence Boundary Detection in Speech</Title>
  <Section position="7" start_page="0" end_page="0" type="concl">
    <SectionTitle>
6 Conclusions and Future Work
</SectionTitle>
    <Paragraph position="0"> We have described two different approaches for modeling and integration of diverse knowledge sources for automatic sentence segmentation from speech: a state-of-the-art approach based on HMMs, and an alternative approach based on posterior probability estimation via maximum entropy. To achieve competitive performance with the max-ent model we devised a cumulative binary coding scheme to map posterior estimates from auxiliary submodels into features for the maxent model.</Paragraph>
    <Paragraph position="1"> The two approaches have complementary strengths and weaknesses that were reflected in the results, consistent with the findings for text-based NLP tasks (Klein and Manning, 2002). The maxent model showed much better accuracy than the HMM with lexical information, and a smaller win after combination with prosodic features. The HMM made more effective use of prosodic information and degraded less with errorful word recognition.</Paragraph>
    <Paragraph position="2"> A interpolation of posterior probabilities from the two systems achieved 2-7% relative error reduction compared to the baseline (significant at p &lt; 0:05 for the reference transcription condition). The results were consistent for two different genres of speech.</Paragraph>
    <Paragraph position="3"> In future work we hope to determine how the individual qualitative differences of the two models (estimation methods, model structure, etc.) contribute to the observed differences in results. To improve results overall, we plan to explore features that combine multiple knowledge sources, as well as approaches that model recognition uncertainty in order to mitigate the effects of word errors. We also plan to investigate using a conditional random field (CRF) model. CRFs combine the advantages of both the HMM and the maxent approaches, being a discriminatively trained model that can incorporate overlapping features (the maxent advantages), while also modeling sequence dependencies (an advantage of HMMs) (Lafferty et al., 2001).</Paragraph>
  </Section>
class="xml-element"></Paper>