<?xml version="1.0" standalone="yes"?>
<Paper uid="N04-2009">
  <Title>Construction of Conceptual Graph representation of texts</Title>
  <Section position="2" start_page="0" end_page="0" type="intro">
    <SectionTitle>
1 Introduction
</SectionTitle>
    <Paragraph position="0"> The problem of automatic acquisition of knowledge is an interesting and challenging one and has been tackled by linguists for some time.</Paragraph>
    <Paragraph position="1"> This paper describes a system for automatic conceptual graph acquisition using a combination of linguistic resources, such as VerbNet and WordNet, together with semi-automatically compiled domain-specific knowledge. null Such semantic information has a number of possible applications. One possible application is in the area of information retrieval/extraction for enhancing the search methods and for providing more precise search results. Another application is in question-answering systems, allowing users to communicate with the system in natural language (English) and translating their queries/responses into a machine-understandable representation. null We use conceptual graphs (CGs) (Sowa, 1984), a knowledge-representation formalism based on semantic networks and the existential graphs of C. S. Peirce. There is a defined mapping between a conceptual graph and a corresponding first-order logical formula, although conceptual graphs also allow for representation of temporal and non-monotonic logics, thus exceeding the expressive power of FOL.</Paragraph>
    <Paragraph position="2"> One of the first systems for the generation of conceptual graph representation of text is described in (Sowa and Way, 1986). It uses a lexicon of canonical graphs that represent valid (possible) relations between concepts. These canonical graphs are then combined to build a conceptual graph representation of a sentence.</Paragraph>
    <Paragraph position="3"> Veraldi at al. (1988) describe a prototype of a semantic processor for Italian sentences. It uses a lexicon of about 850 word-sense definitions, each including 10-20 surface semantic patterns (SSPs). Each SSP represents both usage information and semantic constrains and is manually acquired.</Paragraph>
    <Paragraph position="4"> There are also systems aimed at extracting partial knowledge from texts, by either filling semantic templates (Hobbs et al., 1996) or by generation of a set of linguistic patterns for information extraction (Harabagiu and Maiorano, 2000), to name few.</Paragraph>
    <Paragraph position="5"> The following section describes the general overview of the system, together with the documents we used to test our algorithms. Section 3 describes the semantic role identification module, Section 4 outlines the algorithm for constructing the conceptual graph representation of a sentence. The experiments that we performed are described in Section 5, while in Section 6 we draw some conclusions and outline ongoing and future work.</Paragraph>
  </Section>
class="xml-element"></Paper>