<?xml version="1.0" standalone="yes"?>
<Paper uid="W06-1411">
  <Title>Sydney, July 2006. c(c)2006 Association for Computational Linguistics Group-based Generation of Referring Expressions</Title>
  <Section position="6" start_page="78" end_page="79" type="evalu">
    <SectionTitle>
4.2 Results
</SectionTitle>
    <Paragraph position="0"> Table 1 shows the results of Experiment 1. The average accuracy of target identification is 95%.</Paragraph>
    <Paragraph position="1"> Figure 10: An example stimulus of Experiment 2 This shows a good performance of the generation algorithm proposed in this paper.</Paragraph>
    <Paragraph position="2"> The expression generated for arrangement No. 20 (shown in Figure 9) resulted in the exceptionally poor accuracy. To refer to object b1, our algorithm generated expression &amp;quot;itiban temae no tama (the most front ball)&amp;quot; because b1 is the most close object to person P in terms of the vertical axis. Humans, however, chose theobject that isthe closest to P in terms of Euclidean distance. Some psychological investigation is necessary to build a more precise geometrical calculation model to solve this problem (Landragin et al., 2001).</Paragraph>
    <Paragraph position="3"> Table 2 shows the results of Experiment 2. The first row shows the rank of expressions based on their score. The second row shows the count ofhuman votes for the expression. The third row shows the ratio of the votes. The top two expressions occupy 72% of the total. This concludes that our scoring function works well.</Paragraph>
    <Paragraph position="4"> 5Conclusion This paper extended the SOG representation proposed in (Funakoshi et al., 2004) to generate refer- null ring expressions in more general situations.</Paragraph>
    <Paragraph position="5"> The proposed method was implemented and evaluated through two psychological experiments using 18 subjects. The experiments showed that generated expressions had enough discrimination ability and that the scoring function conforms to human preference well.</Paragraph>
    <Paragraph position="6"> The proposed method would be able to handle other attributes and relations as far as they can be represented in terms of features as described in section 3. Corresponding surface realization rules might be added in that case.</Paragraph>
    <Paragraph position="7"> In the implementation, we introduced rather ad hoc parameters, particularly in the scoring function. Although this worked well in our experiments, further psychological validation is indispensable. null This paper assumed a fixed reference frame is shared by all participants in a situation. However, when we apply our method to conversational agent systems, e.g., (Tanaka et al., 2004), reference frames change dynamically and they must be properly determined each time when generating referring expressions.</Paragraph>
    <Paragraph position="8"> In this paper, we focused on two dimensional situations. To apply our method to three dimensional worlds, more investigation on human perception of spatial relations are required. We acknowledge that a simple application of the current method does not work well enough in three dimensional worlds.</Paragraph>
  </Section>
class="xml-element"></Paper>