<?xml version="1.0" standalone="yes"?>
<Paper uid="W05-0636">
  <Title>Joint Parsing and Semantic Role Labeling</Title>
  <Section position="3" start_page="0" end_page="0" type="intro">
    <SectionTitle>
1 Introduction
</SectionTitle>
    <Paragraph position="0"> Although much effort has gone into developing statistical parsing models and they have improved steadily over the years, in many applications that use parse trees errors made by the parser are a major source of errors in the final output. A promising approach to this problem is to perform both parsing and the higher-level task in a single, joint probabilistic model. This not only allows uncertainty about the parser output to be carried upward, such as through an k-best list, but also allows information from higher-level processing to improve parsing. For example, Miller et al. (2000) showed that performing parsing and information extraction in a joint model improves performance on both tasks. In particular, one suspects that attachment decisions, which are both notoriously hard and extremely important for semantic analysis, could benefit greatly from input from higher-level semantic analysis.</Paragraph>
    <Paragraph position="1"> The recent interest in semantic role labeling provides an opportunity to explore how higher-level semantic information can inform syntactic parsing. In previous work, it has been shown that SRL systems that use full parse information perform better than those that use shallow parse information, but that machine-generated parses still perform much worse than human-corrected gold parses.</Paragraph>
    <Paragraph position="2"> The goal of this investigation is to narrow the gap between SRL results from gold parses and from automatic parses. We aim to do this by jointly performing parsing and semantic role labeling in a single probabilistic model. In both parsing and SRL, state-of-the-art systems are probabilistic; therefore, their predictions can be combined in a principled way by multiplying probabilities. In this paper, we rerank the k-best parse trees from a probabilistic parser using an SRL system. We compare two reranking approaches, one that linearly weights the log probabilities, and the other that learns a reranker over parse trees and SRL frames in the manner of Collins (2000).</Paragraph>
    <Paragraph position="3"> Currently, neither method performs better than simply selecting the top predicted parse tree. We discuss some of the reasons for this; one reason being that the ranking over parse trees induced by the semantic role labeling score is unreliable, because the model is trained locally.</Paragraph>
  </Section>
class="xml-element"></Paper>