CRANV2 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. - 216 - DOCUMENT OUTPUT CUT-OFF III-1 R P III- 2 III- 3 II1-4 Ill - 5 III- 6 R P R P R P R P R P 1 13 62 13 62 11 50 10 48 10 48 8 38 2 21 50 21 50 16 37 16 37 15 36 13 30 3 31 46 31 48 24 38 25 40 23 36 22 34 4 36 42 35 42 29 36 29 36 28 33 28 33 5 39 37 38 37 35 33 35 33 31 29 33 31 6-7 46 31 45 31 43 29 43 29 38 26 39 27 8-10 53 25 52 25 49 23 51 24 46 22 46 22 11-15 62 20 63 20 59 18[OCRerr] 60 19 54 17 55 17 16-20 67 16 68 16 63 14 65 15 62 14 65 15 72 ii 79 7 85 5 88 4 93 3 96 3 21-30 31-50 51-75 76-100 101-125 126-150 151-175 I00 2 176-200 I00 2 NORMALISED RECALL 74 12 70 10 71 11 70 10 74 12 79 7 78 7 79 7 80 7 82 8 86 5 85 5 86 5 87 5 89 6 88 4 89 4 90 4 91 4 '93 4 93 3 92 3 93 3 95 3 95 3 95 3 95 3 95 3 98 3 98 3 100 2 99 2 99 2 99 2 100 2 100 2 100 2 100 2 100 2 100 2 Search E 63.58 63.64 61.00 61.58 60.41 61.17 Search A 61.76 61.76 60.11 59,70 59.58 59.17 FIGURE 5.21T RECALL AND PRECISION FIGURES FOR INDEX LANGUAGES III. 1 - III.6 . FOR SEARCH E BY DOCUMENT OUTPUT CUT-OFF METHOD, TOGETHER WITH NORMALISED RECALL FOR SEARCH E AND SEARCH A. (R = Recall Ratio, P = Precision Ratio)