<?xml version="1.0" standalone="yes"?>
<Paper uid="P06-2016">
  <Title>Markov model</Title>
  <Section position="3" start_page="0" end_page="120" type="intro">
    <SectionTitle>
1 Introduction
</SectionTitle>
    <Paragraph position="0"> Hidden Markov models (HMMs) were introduced in the late 1960s, and are widely used as a probabilistic tool for modeling sequences of observations (Rabiner and Juang, 1986). They have proven to be capable of assigning semantic labels to tokens over a wide variety of input types.</Paragraph>
    <Paragraph position="1">  This is useful for text-related tasks that involve some uncertainty, including part-of-speech tagging (Brill, 1995), text segmentation (Borkar et al., 2001), named entity recognition (Bikel et al., 1999) and information extraction tasks (McCallum et al., 1999). However, most natural language processing tasks are dependent on discovering a hierarchical structure hidden within the source information. An example would be predicting semantic roles from English sentences. HMMs are less capable of reliably modeling these tasks. In contrast hierarchical hidden Markov models (HH-MMs) are better at capturing the underlying hierarchy structure. While there are several difficulties inherent in extracting information from the patterns hidden within natural language information, by discovering the hierarchical structure more accurate models can be built.</Paragraph>
    <Paragraph position="2"> HHMMs were first proposed by Fine (1998) to resolve the complex multi-scale structures that pervade natural language, such as speech (Rabiner and Juang, 1986), handwriting (Nag et al., 1986), and text. Skounakis (2003) described the HHMM as multiple &amp;quot;levels&amp;quot; of HMM states, where lower levels represents each individual output symbol, and upper levels represents the combinations of lower level sequences.</Paragraph>
    <Paragraph position="3"> Any HHMM can be converted to a HMM by creating a state for every possible observation, a process called &amp;quot;flattening&amp;quot;. Flattening is performed to simplify the model to a linear sequence of Markov states, thus decreasing processing time.</Paragraph>
    <Paragraph position="4"> But as a result of this process the model no longer contains any hierarchical structure. To reduce the models complexity while maintaining some hierarchical structure, our algorithm uses a &amp;quot;partial flattening&amp;quot; process.</Paragraph>
    <Paragraph position="5"> In recent years, artificial intelligence re- null searchers have made strenuous efforts to reproduce the human interpretation of language, whereby patterns in grammar can be recognised and simplified automatically. Brill (1995) describes a simple rule-based approach for learning by rewriting the bracketing rule--a method for presenting the structure of natural language text-for linguistic knowledge. Similarly, Krotov (1999) puts forward a method for eliminating redundant grammar rules by applying a compaction algorithm. This work draws upon the lessons learned from these sources by automatically detecting situations in which the grammar structure can be reconstructed. This is done by applying the phrase extraction method introduced by Pantel (2001) to rewrite the bracketing rule by calculating the dependency of each possible phrase. The outcome of this restructuring is to reduce the complexity of the hierarchical structure and reduce the number of levels in the hierarchy.</Paragraph>
    <Paragraph position="6"> This paper considers the tasks of identifying the syntactic structure of text chunking and grammar parsing with previously annotated text documents. It analyses the use of HHMMs--both before and after the application of improvement techniques--for these tasks, then compares the results with HMMs. This paper is organised as follows: Section 2 describes the method for training HHMMs. Section 3 describes the flattening process for reducing the depth of hierarchical structure for HHMMs. Section 4 discusses the use of HHMMs for the text chunking task and the grammar parser. The evaluation results of the HMM, the plain HHMM and the merged and partially flattened HHMM are presented in Section 5. Finally, Section 6 discusses the results.</Paragraph>
  </Section>
class="xml-element"></Paper>