CRANV2 Aslib Cranfield Research Project: Factors Determining the Performance of Indexing Systems: Volume 2 Main test results 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. - 127 [OCRerr] FIGUHE 4.414T Index Language i. 6 a (S. T. Ssmonylme. Quasi-synonyms, Word forms. Coordination) Exhauatlvlty el h[OCRerr]dexlng 2 Search Rule A Document Relevance I - 4 Number of Documente in Collection . 200 (Subset 1) Number of Questions 42 (Subset 2) Number of Relevant Documents 198 Generality Number 23.6 Coord = ! Becan Ination [OCRerr] Precision Fallout Ratio Ratio Level Ratio a/a+c a/a+b x y s b/b+d 1 ! Documents Betrleved Hal. Non-tel, 193 5,693 175 97.5% 3,253 3.4% 88.4% e7.773% 42 42 42 3 ]45 1,431 fi.1% 73.2% 39.661% 8,2% 42 42 17.447% 42 41 42 42 4 113 544 5 57.1% 64 176 17.2% 32.3% 6.033% 35 41 41 8 42 40 20.7% 21,25 2.146% 51,$% 30 39 39 0.488% 20 33 33 7 21 13 8 lO.5% 5 1 81.8% 2.5% o.158% 12 27 27 8 4 0 83.3% 2.0% 1oo, o% 0,012% 2 18 0.00o% 18 2 11 11 10 ! ! o II o 0 0 7 7 o O 12 0 0 3 3 0 1 1 FIGURE 4.415T index Language 1.6.a (S,T Synonyms, Quasi-synonyms. Word forms. Exhauvtivity of Indezing I Search Rtfle A Document Relevance I - 4 Number of Documents in CoLlection 200 (Subset 1) Number of Questions 42 (Subset 2) Nu[OCRerr]nber of Relevant Documents 108 Generality Number 23.6 Coord [nat ion) Coord- Documents Ination Recall Precision I[OCRerr]etrieved FaLlout Level Ratio Hal, Non-rel, Ratio Ratio a/a+e a/a+b x y b/b+d z 1 183 2 4.127 155 2,118 92.4% 4.2% 40.130% 76.3% 42 42 42 3 I17 585 6.9% 50.1% 25.823% 14.0% 42 42 8.352% 42 38 43 42 4 81 190 5 40.0% 47 28.9% 51 2.3175 32 41 41 6 24 23.7% II 48.0% 12.1% 0.622% 88.6% 22 39 38 0,134% 14 33 33 7 I0 8 3 2 5.1% 78.8% 1.0% 0.037% 5 27 27 9 O 2 1o0. o% 0 0.000% 1.o% 1oo.o% 1 18 0.000% 18 l 11 II tO 0 tl 0 0 0 7 7 L2 0 0 0 O. 3 3 O l 1