<?xml version="1.0" standalone="yes"?>
<Paper uid="W06-1805">
  <Title>Adjective based inference</Title>
  <Section position="2" start_page="0" end_page="0" type="intro">
    <SectionTitle>
1 Introduction
</SectionTitle>
    <Paragraph position="0"> Understanding a text is one of the ultimate goals of computational linguistics. To achieve this goal, systems need to be developed which can construct a meaning representation for any given text and which furthermore, can reason about the meaning of a text. As is convincingly argued in (Ido Dagan and Magnini, 2005), one of the major inference task involved in that reasoning is the entailment recognition task: Does text T1 entail text T2? Indeed entailment recognition can be used to determine whether a text fragment answers a question (e.g., in question answering application), whether a query is entailed by a relevant document (in information retrieval), whether a text fragment entails a specific information nugget (in information extraction), etc.</Paragraph>
    <Paragraph position="1"> Because the Pascal RTE challenge focuses on real text, the participating systems must be robust that is, they must be able to handle unconstrained We thank la R'egion Lorraine, INRIA and the University of Sarrebruecken for partially funding the research presented in this paper.</Paragraph>
    <Paragraph position="2"> input. Most systems therefore are based on statistical methods (e.g., stochastic parsing and lexical distance or word overlap for semantic similarity) and few provide for a principled integration of lexical and compositional semantics. On the other hand, one of the participant teams has shown that roughly 50% of the RTE cases could be handled correctly by a system that would adequately cover semantic entailments that are either syntax based (e.g., active/passive) or lexical semantics based (e.g., bicycle/bike). Given that the overall system accuracies hovered between 50 and 60 percent with a baseline of 50 %1, this suggests that a better integration of syntax, compositional and lexical semantics might improve entailment recognition accuracy.</Paragraph>
    <Paragraph position="3"> In this paper, we consider the case of adjectives and, building on approaches like those described in (Raskin and Nirenburg, 1995; Peters and Peters, 2000), we propose a classification of adjectives which can account for the entailment patterns that are supported by the interaction of their lexical and of their compositional semantics. We start by defining a classification schema for adjectives based on their syntactic and semantic properties.</Paragraph>
    <Paragraph position="4"> We then associate with each class a set of axioms schemas which translate the knowledge about lexical relations (i.e. antonymy) the adjectives of the class are involved in by extracting this information from WordNet (Miller, 1998) and a set of semantic construction rules and we show that these correctly predicts the observed entailment patterns.</Paragraph>
    <Paragraph position="5"> For instance, the approach will account for the following (non)-entailment cases:  The approach is implemented using Description Logic as a semantic representation language and tested on a hand-built semantic test suite of approximately 1 000 items. In the latter part of the paper we discuss this testsuite and the philosophy behind it.</Paragraph>
  </Section>
class="xml-element"></Paper>