<?xml version="1.0" standalone="yes"?>
<Paper uid="P05-1053">
  <Title>Exploring Various Knowledge in Relation Extraction</Title>
  <Section position="7" start_page="432" end_page="433" type="concl">
    <SectionTitle>
6 Discussion and Conclusion
</SectionTitle>
    <Paragraph position="0"> In this paper, we have presented a feature-based approach for relation extraction where diverse lexical, syntactic and semantic knowledge are employed. Instead of exploring the full parse tree information directly as previous related work, we incorporate the base phrase chunking information first. Evaluation on the ACE corpus shows that base phrase chunking contributes to most of the performance improvement from syntactic aspect while further incorporation of the parse tree and dependence tree information only slightly improves the performance. This may be due to three reasons: First, most of relations defined in ACE have two mentions being close to each other.</Paragraph>
    <Paragraph position="1"> While short-distance relations dominate and can be resolved by simple features such as word and chunking features, the further dependency tree and parse tree features can only take effect in the remaining much less and more difficult long-distance relations. Second, it is well known that full parsing  is always prone to long-distance parsing errors although the Collins' parser used in our system achieves the state-of-the-art performance. Therefore, the state-of-art full parsing still needs to be further enhanced to provide accurate enough information, especially PP (Preposition Phrase) attachment. Last, effective ways need to be explored to incorporate information embedded in the full parse trees. Besides, we also demonstrate how semantic information such as WordNet and Name List, can be used in feature-based relation extraction to further improve the performance.</Paragraph>
    <Paragraph position="2"> The effective incorporation of diverse features enables our system outperform previously best-reported systems on the ACE corpus. Although tree kernel-based approaches facilitate the exploration of the implicit feature space with the parse tree structure, yet the current technologies are expected to be further advanced to be effective for relatively complicated relation extraction tasks such as the one defined in ACE where 5 types and 24 subtypes need to be extracted. Evaluation on the ACE RDC task shows that our approach of combining various kinds of evidence can scale better to problems, where we have a lot of relation types with a relatively small amount of annotated data. The experiment result also shows that our feature-based approach outperforms the tree kernel-based approaches by more than 20 F-measure on the extraction of 5 ACE relation types.</Paragraph>
    <Paragraph position="3"> In the future work, we will focus on exploring more semantic knowledge in relation extraction, which has not been covered by current research.</Paragraph>
    <Paragraph position="4"> Moreover, our current work is done when the Entity Detection and Tracking (EDT) has been perfectly done. Therefore, it would be interesting to see how imperfect EDT affects the performance in relation extraction.</Paragraph>
  </Section>
class="xml-element"></Paper>