<?xml version="1.0" standalone="yes"?>
<Paper uid="H01-1053">
  <Title>Monitoring the News: a TDT demonstration system</Title>
  <Section position="3" start_page="0" end_page="0" type="intro">
    <SectionTitle>
2. ARCHITECTURE
</SectionTitle>
    <Paragraph position="0"> The TDT demonstration system is based upon Lighthouse, an interactive information retrieval system developed by Leuski.[6] Lighthouse provides not only a typical ranked list search result, but a visualization of inter-document similarities in 2- or 3-dimensions.</Paragraph>
    <Paragraph position="1"> The user interface is a Java client that can run as an application or an applet. Lighthouse uses http protocols to send queries to a server and receive the ranked list, summary information about the documents, and the visualization data.</Paragraph>
    <Paragraph position="2"> The TDTLighthouse system requires a TDT system running in the background. In this version of the demonstration, the TDT system is only running the segmentation and detection tasks described above. Stories arrive and are put into clusters (bins).</Paragraph>
    <Paragraph position="3"> The TDTLighthouse client can query its server to receive up-to-date information about the clusters that the TDT system has found. The server in turn queries the TDT system to get that information and maintains state information so that changes (cluster growth, additional clusters, etc.) can be highlighted.</Paragraph>
  </Section>
class="xml-element"></Paper>