Search (1 results, page 1 of 1)

0.011551105 = product of:
  0.02310221 = sum of:
    0.02310221 = product of:
      0.04620442 = sum of:
        0.04620442 = weight(_text_:systems in 4154) [ClassicSimilarity], result of:
          0.04620442 = score(doc=4154,freq=4.0), product of:
            0.16037072 = queryWeight, product of:
              3.0731742 = idf(docFreq=5561, maxDocs=44218)
              0.052184064 = queryNorm
            0.28811008 = fieldWeight in 4154, product of:
              2.0 = tf(freq=4.0), with freq of:
                4.0 = termFreq=4.0
              3.0731742 = idf(docFreq=5561, maxDocs=44218)
              0.046875 = fieldNorm(doc=4154)
      0.5 = coord(1/2)
  0.5 = coord(1/2)

Abstract

It is argued that the evaluation of information-retrieval systems should ultimately be based on the principle of rationality, the maximization of expected utility. In full generality this would involve an estimation of both the cost and value of a system, but the emphasis in this paper is on the problem of value, in terms of which the effiency of the system could be defined. One implication of the discussion is that it is not legitimate to superimpose the 2x2 contingency tables that refer to select/discarded and relevant/irrelevant, correspondending to each request,but it might be all right to superimpose them after applying a monotonic function to the entries. In particular, it is questionable whether a useful statistic is the ratio of the total number of relevant selected documents to the total number of relevant ones, over a sample of requests.