<?xml version="1.0" standalone="yes"?>
<Paper uid="P06-1115">
  <Title>Using String-Kernels for Learning Semantic Parsers</Title>
  <Section position="3" start_page="0" end_page="913" type="intro">
    <SectionTitle>
1 Introduction
</SectionTitle>
    <Paragraph position="0">  Computationalsystemsthatlearntotransformnatural language sentences into formal meaning representations have important practical applications in enabling user-friendly natural language communication with computers. However, most of the research in natural language processing (NLP) has been focused on lower-level tasks like syntactic parsing, word-sense disambiguation, information extraction etc. In this paper, we have considered the important task of doing deep semantic parsing to map sentences into their computer-executable meaning representations.</Paragraph>
    <Paragraph position="1"> Previous work on learning semantic parsers either employ rule-based algorithms (Tang and Mooney, 2001; Kate et al., 2005), or use statistical feature-based methods (Ge and Mooney, 2005; Zettlemoyer and Collins, 2005; Wong and Mooney, 2006). In this paper, we present a novel kernel-based statistical method for learning semantic parsers. Kernel methods (Cristianini and Shawe-Taylor, 2000) are particularly suitable for semantic parsing because it involves mapping phrases of natural language (NL) sentences to semantic concepts in a meaning representation language (MRL). Given that natural languages are so flexible, there are various ways in which one can express the same semantic concept. It is difficult for rule-based methods or even statistical feature-based methods to capture the full range of NL contexts which map to a semantic concept because they tend to enumerate these contexts. In contrast, kernel methods allow a convenient mechanism to implicitly work with a potentially infinite number of features which can robustly capture these range of contexts even when the data is noisy.</Paragraph>
    <Paragraph position="2"> Our system, KRISP (Kernel-based Robust Interpretation for Semantic Parsing), takes NL sentences paired with their formal meaning representationsastrainingdata. Theproductionsoftheformal MRL grammar are treated like semantic concepts. For each of these productions, a Support-Vector Machine (SVM) (Cristianini and Shawe-Taylor, 2000) classifier is trained using string similarity as the kernel (Lodhi et al., 2002). Each classifier then estimates the probability of the production covering different substrings of the sentence. This information is used to compositionally build a complete meaning representation (MR) of the sentence.</Paragraph>
    <Paragraph position="3"> Some of the previous work on semantic parsing has focused on fairly simple domains, primarily, ATIS (Air Travel Information Service) (Price,  1990)whosesemanticanalysisisequivalenttofilling a single semantic frame (Miller et al., 1996; Popescu et al., 2004). In this paper, we have tested KRISP on two real-world domains in which meaning representations are more complex with richer predicates and nested structures. Our experiments demonstrate that KRISP compares favor- null NL: &amp;quot;If the ball is in our goal area then our player 1 should intercept it.&amp;quot; CLANG: ((bpos (goal-area our)) (do our {1} intercept))</Paragraph>
  </Section>
class="xml-element"></Paper>