Smith, M.P.; Pollitt, S.A.: ¬A comparison of ranking formulae and their ranks (1995)
0.00
2.3021935E-4 = product of:
0.00345329 = sum of:
0.00345329 = product of:
0.00690658 = sum of:
0.00690658 = weight(_text_:information in 5802) [ClassicSimilarity], result of:
0.00690658 = score(doc=5802,freq=2.0), product of:
0.050870337 = queryWeight, product of:
1.7554779 = idf(docFreq=20772, maxDocs=44218)
0.028978055 = queryNorm
0.13576832 = fieldWeight in 5802, product of:
1.4142135 = tf(freq=2.0), with freq of:
2.0 = termFreq=2.0
1.7554779 = idf(docFreq=20772, maxDocs=44218)
0.0546875 = fieldNorm(doc=5802)
0.5 = coord(1/2)
0.06666667 = coord(1/15)
- Abstract
- Reports a study to compare the ranking produced by several well known probabilistic formulae. Values for the variables used in these formulae (collection frequency for a query term, number of relevant documents retrieved, and number of relevant documents retrieved, and number of relevant documents indexed by the query term) were derived using a random number generator, the number of documents in the collection was fixed at 500.000. This produced ranked bands for each formula using document term characteristics rather than actual documents. These rankings were compared with one another using the Spearman Rho ranked correlation co-efficient to determine how closely the algorithms rank documents. There is little difference in the rankings produced by the Expected Mutual Information measure EMIM and the simpler F4.5 weighting scheme
