<?xml version="1.0" standalone="yes"?>
<Paper uid="C00-1081">
  <Title>A Stochastic Parser Based on a Structural Word Prediction Model Shinsuke MORI, Masafumi NISHIMURA, Nobuyasu ITOH,</Title>
  <Section position="2" start_page="0" end_page="558" type="intro">
    <SectionTitle>
1 Introduction
</SectionTitle>
    <Paragraph position="0"> The stochastic language modeling, imported fl:om the speech recognition area, is one of the snccessflfl methodologies of natural language processing. In fact, all language models for speech recognition are, as far&amp;quot; a.s we know, based on an n-gram model and most practical part-of-speech (POS) taggers are also based on a word or POS n-gram model or its extension (Church, 1.988; Cutting et el., 1992; Merialdo, 1994; l)ennatas and Kokkinakis, 1.995). POS tagging is the first step of natural language processing, and stochastic taggers have solved this problem with satisfying accuracy for many applications. The next step is parsing, or that is to say discovering the structure of a given sentence. Recently, many parsers based on the stochastic approach have been proposed. Although their reported accuracies are high, they are not accurate enough for many applications at this stage, and more attempts have to be made to improve them fm:ther.</Paragraph>
    <Paragraph position="1"> One of the major applications of a parser is to parse the spoken text recognized by a speech recognizer. This attempt is clearly aiming at spoken language understanding. If we consider how to con&gt; bine a parser and a speech recognizer~ it is better if the parser is based on a generative stochastic model, as required for the language model of a speech recognizer. Here, &amp;quot;generative&amp;quot; means that the sum of probabilities over all possible sentences is equal to or less than 1. If the language model is generative, it allows a seamless combination of the parser and the speech recognizer. This means that the speech recognizer has the stochastic parser as its language model and benefits richer information than a normal n-gram model. Even though such a Colnbiimtion is not possible in practices , the recognizer outputs N-best sentences with their probabilities, and the parser, taking them as input, parses all of them and outputs the sentence with its parse tree that has the highest probability of all possible combinations. As a resnlt, a parser based on a generative stochastic language model may hell) a speech recognizer to select the most syntactically reasonable sentence among candidates. Therefore, it is better if the language model of a parser is generative.</Paragraph>
    <Paragraph position="2"> In this paper, taking Japanese as the object language, we propose a generative stochastic language model and a parser based on it. This model treats a sentence as a word sequence and predicts each word from left to right. The history at each step of prediction is a sequence of partial parse trees covering the preceding words. To predict a word, our model first predicts which of the partial parse trees at this stage have dependency relation with the word, and then predicts the word fi'om the selected partial parse trees. In Japanese each word depends on a subsequent word, that is to say, each dependency relation is left to right, it is not necessary to predict the direction of each dependency relation. So in order to extend our model to other languages, the model may have to predict the direction of each dependency. We built a parser based on this model, whose parameters are estimated fl:om 1,072 sentences in a financial newspaper, and tested it on 1.19 sentences fl:om the same newspaper:. The accuracy of the depen- null dency relation was 89.9%, the highest obt.ained by any aapa.nese stochastic parsers.</Paragraph>
  </Section>
class="xml-element"></Paper>