<?xml version="1.0" standalone="yes"?>
<Paper uid="W06-2908">
  <Title>Semantic Role Recognition using Kernels on Weighted Marked Ordered Labeled Trees</Title>
  <Section position="3" start_page="0" end_page="53" type="intro">
    <SectionTitle>
1 Introduction
</SectionTitle>
    <Paragraph position="0"> Semantic role labeling (SRL) is a task that recognizes the arguments of a predicate (verb) in a sentence and assigns the correct role to each argument.</Paragraph>
    <Paragraph position="1"> As this task is recognized as an important step after (or the last step of) syntactic analysis, many studies have been conducted to achieve accurate semantic role labeling (Gildea and Jurafsky, 2002; Moschitti, 2004; Hacioglu et al., 2004; Punyakanok et al., 2004; Pradhan et al., 2005a; Pradhan et al., 2005b; Toutanova et al., 2005).</Paragraph>
    <Paragraph position="2"> Most of the studies have focused on machine learning because of the availability of standard datasets, such as PropBank (Kingsbury and Palmer, 2002). Naturally, the usefulness of parse trees in this task can be anticipated. For example, the recent CoNLL 2005 shared task (Carreras and M`arquez, 2005) provided parse trees for use and their usefulness was ensured. Most of the methods heuristically extract features from parse trees, and from other sources, and use them in machine learning methods based on feature vector representation. As a result, these methods depend on feature engineering, which is time-consuming.</Paragraph>
    <Paragraph position="3"> Tree kernels (Collins and Duffy, 2001; Kashima and Koyanagi, 2002) have been proposed to directly handle trees in kernel-based methods, such as SVMs (Vapnik, 1995). Tree kernels calculate the similarity between trees, taking into consideration all of the subtrees, and, thereforethereisnoneedforsuchfeature engineering.</Paragraph>
    <Paragraph position="4"> Moschitti and Bejan (2004) extensively studied tree kernels for semantic role labeling. However, they reported that they could not successfully build an accurate argument recognizer, although the role assignment was improved. Although Moschitti et al.</Paragraph>
    <Paragraph position="5"> (2005) reported on argument recognition using tree kernels, it was a preliminary evaluation because they used oracle parse trees.</Paragraph>
    <Paragraph position="6"> Kazama and Torisawa (2005) proposed a new tree kernel for node relation labeling, as which SRL can be cast. This kernel is defined on marked ordered labeledtrees, whereanodecanhaveamarktoindicate the existence of a relation. We refer to this kernel as the MOLT kernel. Compared to (Moschitti and Bejan, 2004) where tree fragments are heuristically extracted before applying tree kernels, the MOLT kernel is general and desirable since it does not require such fragment extraction. However, the evaluation conducted by Kazama and Torisawa (2005) was limited to preliminary experiments for role assignment. In this study, we first evaluated the performance of the MOLT kernel for argument recognition, andfoundthattheMOLTkernelcannotachieve a high accuracy if used in its original form.</Paragraph>
    <Paragraph position="7">  Figure1: (a)-(c): Argumentrecognitionasnoderelationrecognition. (a'): relation(a)representedasmarked ordered tree.</Paragraph>
    <Paragraph position="8"> Therefore, in this paper we propose a modification of the MOLT kernel, which greatly improves the accuracy. The problem with the original MOLT kernel is that it treats subtrees with one mark, i.e., those including only the argument or the predicate node, and subtrees with two marks, i.e., those including both the argument and the predicate nodes equally, although the latter is likely to be more important for distinguishing difficult arguments. Thus, we modified the MOLT kernel so that the marks can  beweightedinordertogivelargeweightstothesubtrees with many marks. We call the modified kernel the WMOLT kernel (the kernel on weighted marked ordered labeled trees). We show that this modification greatly improves the accuracy when the weights for marks are properly tuned.</Paragraph>
    <Paragraph position="9"> One of the issues that arises when using tree kernels is time complexity. In general, tree kernels can be calculated in O(|T1||T2|) time, where |Ti |is the number of nodes in tree Ti, using dynamic programming (DP) procedures (Collins and Duffy, 2001; Kashima and Koyanagi, 2002). However, this cost is not negligible in practice. Kazama and Torisawa (2005) proposed a method that drastically speeds up the calculation during training by converting trees into efficient vectors using a tree mining algorithm.</Paragraph>
    <Paragraph position="10"> However, the slow classification during runtime remained an open problem.</Paragraph>
    <Paragraph position="11"> We propose a method for speeding up the runtime classification for argument recognition. In argument recognition, we determine whether a node is an argument or not for all the nodes in a tree . This requires a series of calculations between a support vector tree and a tree with slightly different marking. By exploiting this property, we can efficiently update DP cells to obtain the kernel value with less computational cost.</Paragraph>
    <Paragraph position="12"> In the experiments, we demonstrated that the WMOLT kernel drastically improved the accuracy and that our speed-up method enabled more than 40 times faster argument recognition. Despite these successes, the performance of our current system is</Paragraph>
    <Paragraph position="14"> using the Charniak parse trees, which is far worse than the state-of-the-art system. We will present possible reasons and future directions.</Paragraph>
  </Section>
class="xml-element"></Paper>