<?xml version="1.0" standalone="yes"?>
<Paper uid="N06-2003">
  <Title>Museli: A Multi-Source Evidence Integration Approach to Topic Segmentation of Spontaneous Dialogue</Title>
  <Section position="3" start_page="0" end_page="9" type="intro">
    <SectionTitle>
2 Previous Work
</SectionTitle>
    <Paragraph position="0"> Existing topic segmentation approaches can be loosely classified into two types: (1) lexical cohesion models, and (2) content-oriented models. The underlying assumption in lexical cohesion models is that a shift in term distribution signals a shift in topic (Halliday and Hassan, 1976). The best known algorithm based on this idea is TextTiling (Hearst, 1997). In TextTiling, a sliding window is passed over the vector-space representation of the text. At each position, the cosine correlation between the upper and lower region of the sliding window is compared with that of the peak cosine correlation values to the left and right of the window. A segment boundary is predicted when the magnitude of the difference exceeds a threshold.</Paragraph>
    <Paragraph position="1"> One drawback to relying on term co-occurrence to signal topic continuity is that synonyms or related terms are treated as thematically-unrelated.</Paragraph>
    <Paragraph position="2"> One solution to this problem is using a dimensionality reduction technique such as Latent Semantic Analysis (LSA) (Landauer and Dumais, 1997).</Paragraph>
    <Paragraph position="3"> Two such algorithms for segmentation are described in (Foltz, 1998) and (Olney and Cai, 2005).</Paragraph>
    <Paragraph position="4"> Both TextTiling and Foltz's approach measure coherence as a function of the repetition of thematically-related terms. TextTiling looks for co-occurrences of terms or term-stems and Foltz uses LSA to measure semantic relatedness between terms. Olney and Cai's orthonormal basis approach also uses LSA, but allows a richer representation of discourse coherence, which is that coherence is a function of how much new information a discourse unit (e.g. a dialogue contribution) adds (informativity) and how relevant it is to the local context (relevance) (Olney and Cai, 2005).</Paragraph>
    <Paragraph position="5"> Content-oriented models, such as (Barzilay and Lee, 2004), rely on the re-occurrence of patterns of topics over multiple realizations of thematically similar discourses, such as a series of newspaper articles about similar events. Their approach utilizes a hidden Markov model where states correspond to topics, and state transition probabilities correspond to topic shifts. To obtain the desired  number of topics (states), text spans of uniform length (individual contributions, in our case) are clustered. Then, state emission probabilities are induced using smoothed cluster-specific language models. Transition probabilities are induced by considering the proportion of documents in which a contribution assigned to the source cluster (state) immediately precedes a contribution assigned to the target cluster (state). Using an EM-like Viterbi approach, each contribution is reassigned to the state most likely to have generated it.</Paragraph>
  </Section>
class="xml-element"></Paper>