<?xml version="1.0" standalone="yes"?>
<Paper uid="W00-1009">
  <Title>A Common Theory of Information Fusion from Multiple Text Sources Step One: Cross-Document Structure</Title>
  <Section position="2" start_page="0" end_page="74" type="intro">
    <SectionTitle>
1 Introduction
</SectionTitle>
    <Paragraph position="0"> The Topic Detection and Tracking model (TDT) \[Allan et al. 98\] describes news events as they are reflected in news sources. First, many sources write on the same event and, second, the same source typically produces a number of accounts of the event over a period of time.</Paragraph>
    <Paragraph position="1"> Sixteen news stories related to the same event from six news sources over a two-hour time period are represented in Figure 1.</Paragraph>
    <Paragraph position="3"> Figure 1 : Time distribution of related documents from multiple sources A careful analysis of related news articles shows that they exhibit some interesting properties \[Radev &amp; McKeown 98\]. In some cases, different sourees agree with each other, at other times, the information presented in them is contradictory. The same source sometimes adds new information as it becomes available and puts it in the context of what has already been discussed earlier. In other eases, to get a full picture of an event, one has to read stories from multiple sources as neither of them presents all relevant information. All these examples point to the existence of cross-document structure that is waiting to be exploited.</Paragraph>
    <Paragraph position="4"> Figure 2 illustrates how the same story can be told in several different ways. The six extracts are from news stories about the same event: the declaration by Bill Clinton at a press conference that millions of dollars will be handed out to low income people affected by recent surges in oil pFices.</Paragraph>
    <Paragraph position="5"> In this paper we introduce CST, a theory of cross-document structure. CST assigns labels  such as subsumption, update, and elaboration to cross-document conceptual links. We argue that CST is essential for the analysis of contradiction, redundancy, and complementarity in related documents and for multi-document summarization (MDS).</Paragraph>
    <Paragraph position="6"> FoxNews: President Clinton announced Wednesday he will release funds to help New Englanders through the heating oil &amp;quot;crisis&amp;quot; and spoke on topics ranging from gun violence to his personal life and foreign policy. The president said he ordered the release of $125 million from the Low Income Home Energy Assistance Program to help families who must rely on oil to heat their homes. He said that amount was all that is left in the fund this year. ABCN~s: President Clinton today ordered the release of millions of dollars in assistance for Northeastern families struggling with soaring fuel costs, saying Americans have together to help their fellow citizens in times of need. The release of $120 million from the Low Income Home Energy Assistance Program is to help families who must rely on oil to heat their homes, he said ...</Paragraph>
    <Paragraph position="7"> CNN: Citing rising energy costs, President Clinton said Wednesday he is releasing $120 million in funds to help families buy home heating oil in the mid-Atlantic and northeastern states. The president said lie decided the help was needed heeause steadily rising oil prices have slammed consumers dependent on home heating oil and have boosted the cost of,gasoline. MSNBC: AT A WASHINGTON, D.C. news conference, Clinton announced that he had ordered the release of the remainder of federal heating assistance money to help families in the Northeast hit hard by this year's cold weather and the soaring cost of home heating oil.</Paragraph>
    <Paragraph position="8"> Clinton said lie ordered the release of $125 million from the Low Income Home Energy Assistance Program to help families who must rely on oil to heat their homes. He said that amount was all that is left in the fund this year. USA Today: President Clinton, saying too many families are being hurt by the soaring cost of heating their homes, announced Wednesday he will ask Congress for $600 million in emergency assistance to help people meet heating oil costs. He also released an additional $125 million for more immediate help to low-income families. NY Times: WASHINGTON, Feb. 16 - President Clinton ordered the release today of another $125 million in government aid to help needy families pay the soaring costs of heating their homes this winter. The release, announced by Mr. Clinton at the start of a White House news conference this afternoon, comes only six days after the government made $130 million in home-heating aid available.</Paragraph>
  </Section>
class="xml-element"></Paper>