Search (1 results, page 1 of 1)

0.009550241 = product of:
  0.019100482 = sum of:
    0.019100482 = product of:
      0.038200963 = sum of:
        0.038200963 = weight(_text_:u in 6686) [ClassicSimilarity], result of:
          0.038200963 = score(doc=6686,freq=4.0), product of:
            0.14932987 = queryWeight, product of:
              3.2744443 = idf(docFreq=4547, maxDocs=44218)
              0.045604642 = queryNorm
            0.25581595 = fieldWeight in 6686, product of:
              2.0 = tf(freq=4.0), with freq of:
                4.0 = termFreq=4.0
              3.2744443 = idf(docFreq=4547, maxDocs=44218)
              0.0390625 = fieldNorm(doc=6686)
      0.5 = coord(1/2)
  0.5 = coord(1/2)

Abstract

In networked information environments, using server-browser architectures, nearly all information finding episodes become extended interactions between the user and the system. In this setting the system needs some way to "understand" what the user is seeking, as this goal adapts and is modified during a session or a series of sessions. We describe a formal model, in which the model of the user's quest is represented as a generalized abstract "response function" representing the user's response to the information delivered by the system. Representing this response as u(n) = Q(S(n - 1)) shows that the user's utterance u(n) at a time step n is determined according to the user's "response function" Q by the materials S(n - 1) that had been presented up through the previous time step n - 1. The entire history of materials presented thus plays a role in determining the user's response, providing a very rich probe into the precise nature of the user's information quest, here represented by the rule Q. We show how this gives rise naturally to a new model for assimilating relevance feedback information, and to the concept of itineraries in the information network. Finally the concept of an information quest Q, provides a natural framework for considering the time dependence of information about the user's needs, and for various models of information aging. The use and effectiveness of this concept are illustrated with data collected in the Ant World Project at Rutgers