<?xml version="1.0" standalone="yes"?>
<Paper uid="W06-0504">
  <Title>Ontology Population from Textual Mentions: Task Definition and Benchmark</Title>
  <Section position="8" start_page="31" end_page="31" type="concl">
    <SectionTitle>
5 Conclusion and future work
</SectionTitle>
    <Paragraph position="0"> We have presented work in progress aiming at a better definition of the general OLP task. In particular we have introduced Ontology Population from Textual Mentions (OPTM) as a simplification of OLP, where the source textual material are already classified mentions of entities.</Paragraph>
    <Paragraph position="1"> An analysis of the data has been conducted over a OPTM benchmark manually built from a corpus of Italian news. As a result a number of indicators have been extracted that suggest the complexity of the task for systems aiming at automatic resolution of OPTM.</Paragraph>
    <Paragraph position="2"> Our future work is related to the definition and extension of the OPTM benchmark for the normalization step (see Introduction). For this step it is crucial the construction and use of a large-scale ontology, including the concepts and relations referred by mentions. A number of interesting relations between mentions and ontology are likely to emerge.</Paragraph>
    <Paragraph position="3"> The work presented in this paper is part of the ONTOTEXT project, a larger initiative aimed at developing text mining technologies to be exploited in the perspective of the Semantic Web.</Paragraph>
    <Paragraph position="4"> The project focuses on the study and development of innovative knowledge extraction techniques for producing new or less noisy information to be made available to the Semantic Web. ONTOTEXT addresses three key research aspects: annotating documents with semantic and relational information, providing an adequate degree of interoperability of such relational information, and updating and extending the ontologies used for Semantic Web annotation. The concrete evaluation scenario in which algorithms will be tested with a number of large-scale experiments is the automatic acquisition of information about people from newspaper articles.</Paragraph>
  </Section>
class="xml-element"></Paper>