<?xml version="1.0" standalone="yes"?>
<Paper uid="W05-0205">
  <Title>Towards Intelligent Search Assistance for Inquiry-Based Learning</Title>
  <Section position="5" start_page="26" end_page="27" type="concl">
    <SectionTitle>
4 Preliminary Results and Discussion
</SectionTitle>
    <Paragraph position="0"> OLISA is under development. While thorough evaluation is needed, our preliminary results demonstrate its effectiveness. We conducted field studies with middle school students for OIBL projects using IdeaKeeper. Fig.1 shows a case of using OLISA search function in IdeaKeeper. By video  taping some students' search session, we found that enhanced search functions of OLISA significantly saved students' effort and improve their experience on search. The term suggestions were frequently used in these sessions.</Paragraph>
    <Paragraph position="1"> Fig. 1 Using OLISA function in IdeaKeeper Our initials results also demonstrate that calculation on the snippets returned by search engines is simple and efficient. Therefore, we don't need to retrieve each full document behind. We want to point out that in our feature vector calculation each past query is combined into previous context. So the learning context is interactively changing.</Paragraph>
    <Paragraph position="2"> Previous research has found that in OIBL projects, students often spend considerable time searching for sites due to their limited search skills. Consequently, students have little time on higher-order cognitive and metacognitive activities, such as evaluation, sense making, synthesis, and reflection. By supporting students' search, OLISA helps student focus more on higher-order activities, which provide rich opportunities for deep learning to occur.</Paragraph>
    <Paragraph position="3"> Our future work includes fine-tuning the parameters in our algorithms and conducting more evaluation of each component of OLISA. We are also considering taking into account the snippets or documents users selected, because they also represent user feedback. How to determine the relative weight of words in selected documents, and how to disambiguate polysemies using WordNet or other resources are topics of future research.</Paragraph>
  </Section>
class="xml-element"></Paper>