<?xml version="1.0" standalone="yes"?>
<Paper uid="W06-3806">
  <Title>Similarity between Pairs of Co-indexed Trees for Textual Entailment Recognition</Title>
  <Section position="2" start_page="0" end_page="33" type="intro">
    <SectionTitle>
1 Introduction
</SectionTitle>
    <Paragraph position="0"> Recently, a remarkable interest has been devoted to textual entailment recognition (Dagan et al., 2005).</Paragraph>
    <Paragraph position="1"> The task requires to determine whether or not a text T entails a hypothesis H. As it is a binary classification task, it could seem simple to use machine learning algorithms to learn an entailment classifier from training examples. Unfortunately, this is not. The learner should capture the similarities between different pairs, (Tprime,Hprime) and (Tprimeprime,Hprimeprime), taking into account the relations between sentences within a pair.</Paragraph>
    <Paragraph position="2"> For example, having these two learning pairs:</Paragraph>
    <Paragraph position="4"> T1 &amp;quot;At the end of the year, all solid companies pay dividends&amp;quot; H1 &amp;quot;At the end of the year, all solid insurance companies pay dividends.&amp;quot; T1 notdblarrowright H2 T1 &amp;quot;At the end of the year, all solid companies pay dividends&amp;quot; H2 &amp;quot;At the end of the year, all solid companies pay cash dividends.&amp;quot; determining whether or not the following implication holds:</Paragraph>
    <Paragraph position="6"> T3 &amp;quot;All wild animals eat plants that have scientifically proven medicinal properties.&amp;quot; null H3 &amp;quot;All wild mountain animals eat plants that have scientifically proven medicinal properties.&amp;quot; requires to detect that: 1. T3 is structurally (and somehow lexically) similar to T1 and H3 is more similar to H1 than to H2; 2. relations between the sentences in the pairs (T3,H3) (e.g., T3 and H3 have the same noun  governing the subject of the main sentence) are similar to the relations between sentences in the pairs (T1,H1) and (T1,H2).</Paragraph>
    <Paragraph position="7"> Given this analysis we may derive that T3 = H3. The example suggests that graph matching tecniques are not sufficient as these may only detect the structural similarity between sentences of textual entailment pairs. An extension is needed to consider also if two pairs show compatible relations between their sentences.</Paragraph>
    <Paragraph position="8"> In this paper, we propose to observe textual entailment pairs as pairs of syntactic trees with co-indexed nodes. This shuold help to cosider both the structural similarity between syntactic tree pairs and the similarity between relations among sentences within a pair. Then, we use this cross-pair similarity with more traditional intra-pair similarities (e.g., (Corley and Mihalcea, 2005)) to define a novel kernel function. We experimented with such kernel using Support Vector Machines on the Recognizing Textual Entailment (RTE) challenge test-beds. The comparative results show that (a) we have designed an effective way to automatically learn entailment rules  from examples and (b) our approach is highly accurate and exceeds the accuracy of the current state-of-the-art models.</Paragraph>
    <Paragraph position="9"> In the remainder of this paper, Sec. 2 introduces the cross-pair similarity and Sec. 3 shows the experimental results.</Paragraph>
  </Section>
class="xml-element"></Paper>