<?xml version="1.0" standalone="yes"?>
<Paper uid="W03-1206">
  <Title>HITIQA: An Interactive Question Answering System A Preliminary Report</Title>
  <Section position="3" start_page="0" end_page="1" type="intro">
    <SectionTitle>
2 Factual vs. Analytical
</SectionTitle>
    <Paragraph position="0"> The objective in HITIQA is to allow the user to submit and obtain answers to exploratory, analytical, non-factual questions. There are very significant differences between factual, or fact-finding, and analytical question answering. A factual question seeks pieces of information that would make a corresponding statement true (i.e., they become facts): &amp;quot;How many states are in the U.S.?&amp;quot; / &amp;quot;There are X states in the U.S.&amp;quot; In this sense, a factual question usually has just one correct answer that can generally, be judged for its truthfulness. By contrast, an analytical question is when the &amp;quot;truth&amp;quot; of the answer is more a matter of opinion and may depend upon the context in which the question is asked. Answers to analytical questions are rarely unilateral, indeed, a mere &amp;quot;correct&amp;quot; answer may have limited value, and in some cases may not even be determinate (&amp;quot;Which college is the best?&amp;quot;, &amp;quot;How do I stop my baby's crying?&amp;quot;). Instead, answers to analytical questions are often judged as helpful, or useful, or satisfactory, etc. &amp;quot;Technically correct&amp;quot; answers (e.g., &amp;quot;feed the baby milk&amp;quot;) may be considered as irrelevant or at best unresponsive.</Paragraph>
    <Paragraph position="1"> The distinction between factual and analytical questions depends primarily on the intention of the person who is asking, however, the form of a question is often indicative of which of the two classes it is more likely to belong to. Factual questions can be classified into a number of syntactic formats (&amp;quot;question typology&amp;quot;) that aids in automatic processing. null Factual questions display a fairly distinctive &amp;quot;answer type&amp;quot;, which is the type of the information piece needed to fulfill the statement. Recent automated systems for answering factual questions deduct this expected answer type from the form of the question and a finite list of possible answer types. For example, &amp;quot;Who was the first man in space&amp;quot; expects a &amp;quot;person&amp;quot; as the answer, while &amp;quot;How long was the Titanic?&amp;quot; expects some length measure as an answer, probably in yards and feet, or meters. This is generally a very good strategy, that has been exploited successfully in a number of automated QA systems that appeared in recent years, especially in the context of TREC QA  evaluations (Harabagiu et al., 2000; Hovy et al., 2000; Prager at al., 2001).</Paragraph>
    <Paragraph position="2"> This process is not easily applied to analytical questions. This is because the type of an answer for analytical questions cannot always be anticipated due to their inherently exploratory character. In contrast to a factual question, an analytical question has an unlimited variety of syntactic forms with only a loose connection between their syntax and the expected answer. Given the unlimited potential of the formation of analytical questions, it would be counter-productive to restrict them to a limited number of question/answer types. Even finding a non-strictly factual answer to an otherwise simple question about Titanic length (e.g., &amp;quot;two football fields&amp;quot;) would push the limits of the answer-typing approach. Therefore, the formation of an answer should instead be guided by the topics the user is interested in, as recognized in the query and/or through the interactive dialogue, rather than by a single type as inferred from the query in a factual system.</Paragraph>
    <Paragraph position="3"> This paper argues that the semantics of an analytical question is more likely to be deduced from the information that is considered relevant to the question than through a detailed analysis of their particular form. While this may sound circular, it needs not be. Determining &amp;quot;relevant&amp;quot; information is not the same as finding an answer; indeed we can use relatively simple information retrieval methods (keyword matching, etc.) to obtain perhaps 50 or 100 &amp;quot;relevant&amp;quot; documents from a database. This gives us an initial answer space to work on in order to determine the scope and complexity of the answer. In our project, we use structured templates, which we call frames to map out the content of pre-retrieved documents, and subsequently to delineate the possible meaning of the question (Section 6).</Paragraph>
    <Paragraph position="4">  TREC QA is the annual Question Answering evaluation sponsored by the U.S. National Institute of Standards and Technology www.trec.nist.gov.</Paragraph>
  </Section>
class="xml-element"></Paper>