<?xml version="1.0" standalone="yes"?>
<Paper uid="H05-1123">
  <Title>Proceedings of Human Language Technology Conference and Conference on Empirical Methods in Natural Language Processing (HLT/EMNLP), pages 979-986, Vancouver, October 2005. c(c)2005 Association for Computational Linguistics A Generalized Framework for Revealing Analogous Themes across Related Topics</Title>
  <Section position="2" start_page="0" end_page="979" type="intro">
    <SectionTitle>
1 Introduction
</SectionTitle>
    <Paragraph position="0"> The ability to identify analogies and correspondences is one of the fascinating aspects of intelligence. Research in cognitive science has acknowledged the significance of this ability of human thinking, particularly in learning across different situations or domains where the common base to learning is not straightforward. Several previous computational models of analogy making (e.g. Falkenhainer et al., 1989) suggested symbolic computational mechanisms for constructing detailed mappings that connect corresponding ingredients across analogized systems.</Paragraph>
    <Paragraph position="1"> This work explores the identification of thematic correspondences in texts through an extension of the well known data clustering problem. Previous works aimed at identifying - through clusters of words - concepts, sub-topics or themes that are prominent within a corpus of texts (e.g., Pereira et al., 1993; Li, 2002; Lin and Pantel, 2002). The current work deals with extending this line of research to identify corresponding themes across a corpus pre-divided to several sub-corpora, which are focused on different, yet related, topics.</Paragraph>
    <Paragraph position="2"> This research task has been defined quite recently (Dagan et al., 2002), and has not been explored extensively yet. One could think, however, of many potential applications for drawing correspondences across textual resources: comparison of related firms or products, identifying equivalencies in news published in different countries, and so on. The experimental part of our work deals with revealing correspondences between different religions: Buddhism, Christianity, Hinduism, Islam and Judaism. Given a pre-partition of the corpus to sub-corpora, one for each religion, our method exposes common aspects for all religions, such as sacred writings, festivals and suffering.</Paragraph>
    <Paragraph position="3"> The mechanism we employ directs corresponding key terms in the different sub-corpora, such as names of festivals of different religions, to be included in the same cluster. Term clustering methods in general, and in this work in particular, rely on word co-occurrence statistics: terms sharing similar words co-occurrence statistics are clustered together. Different topics, however, are characterized by distinctive terminology and typical word co-locations. Therefore, given a pre-divided corpus, similar co-occurrence patterns would typically be extracted from the same topical sub-corpus.</Paragraph>
    <Paragraph position="4"> When the terminology and typical phrases employed by each topic differ greatly (even if the top- null ics are essentially related, e.g. different religions), the tendency to form topic-specific clusters intensifies regardless of factors that otherwise could have impact this tendency, such as the co-occurrence window size. Consequently, corresponding key terms of different topics may not be assigned by a standard method to the same cluster, in contrast to our goal. The method described in this paper aims precisely at this problem: it is designed to neutralize salient co-occurrence patterns within each topical sub-corpus and to promote less salient patterns that are shared across the sub-corpora.</Paragraph>
    <Paragraph position="5"> In an earlier line of research we have formulated the above problem and addressed it within a probabilistic vector-based setting, presenting two related heuristic algorithms (Dagan et al., 2002; Marx et al., 2004). Here, we devise a general principled distributional clustering paradigm for this problem, termed cross-partition clustering, and show that the earlier algorithms are special cases of the new framework.</Paragraph>
    <Paragraph position="6"> This paper proceeds as follows: Section 2 describes in more detail the cross-partition clustering problem. Section 3 reviews distributional data clustering methods, which form the basis to our algorithmic framework described in Section 4.</Paragraph>
    <Paragraph position="7"> Section 5 presents experimental results that reveal interesting themes common to different religions and demonstrates, through an evaluation based on human expert data, that the different variants of our framework outperform alternative methods.</Paragraph>
  </Section>
class="xml-element"></Paper>