CRANV2
Aslib Cranfield Research Project: Factors Determining the Performance of Indexing Systems: Volume 2
Simulated ranking and document output cut-off
chapter
Cyril Cleverdon
Michael Keen
Cranfield
An investigation supported by a grant to Aslib by the National Science Foundation.
Use, reproduction, or publication, in whole or in part, is permitted for any purpose of the United States Government.
- 195 -
By using these figures it was found possible to obtain a simulated ranking
output. This is done by assigning a rank order number to each relevant
document retrieved by means of the equations:-
cR = X [OCRerr] (n - Y ) c
n c c Yc +
where cH is the rank order number of tl]e n relevant document to be
th
n
retrieved
th
is the coordination level at which the n relevant document
is retrieved
x is the additional number of documents retrieved at coordination
c level c. (i.e. those not retr[OCRerr] higher coordination level)
Ye is the additional number of relevant documents retrieved at
coordination level c. (i.e. those not retrieved at a higher
coordination level)
X is the total number of documents retrieved before searching at
c
coordination level c. (i. e. at higgher coordination levels)
Y is the total number of relevant documents retrieved before
c
searching at coordination level c. (i. e. at higher coordination
levels)
cR is taken to the nearest whole number but if its value falls exactly
n
between two whole numbers it is taken to the lower whole number for odd
numbered questions and to the higher whole number for even numbered
questions. Two examples to illustrate the effect are taken from Fig. 5.2.
With Question 100, no documents are retrieved at a coordination level
higher than four, so for this question, the various values are as follows:
Question 100
At level c=4, then x4 = 3, Y4 = 1, X4 = 0, Y4 = 0
At level c=3, then x3 = 50, Y3 = 2, X3 = 3, Y3 = i
At level c=2, then x2 = 21, Y2 = 0, X2 = 53, Y2 = 3
At level c=l, then x1 = 97, Yl = 1, X1 = 74, Y1 = 3
", For Relevant Document 1, retrieved at level 4 :-
4R1 = 0 + ¢1 - 0) \Y--[OCRerr]-i-[OCRerr] = 0 ÷ 2 = 2
For Relevant Document 2, retrieved at level 3 :-
3R2 = 3 + ,2 - i) (50 + 1)
[OCRerr]l : 3+iv -- 20
For l[OCRerr]elevant Document 3, retrieved at level 3 :-
3R3 3 + (3 - 1) \[OCRerr]j = 3 + 34 = 37
For l[OCRerr]elevant Document 4 retrieved at level 1 :-
= 74 + {4 - 3) --]--[OCRerr]+ L/ = 74 + 49 = 123
In the next example considered, Question 123, there are actually
four relevant documents; no documents are retrieved at a coordination