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. - 141 - FIGURE 4. 600T Index Language I. I .a (S. T. Natural language. Coordination) Exhaustivity of Indexing 3 Search R ',L1 e A Document Relevance 1 - 3 Number of Documents in CoLlection 1,400 Number of Questions 35 {subect 1) Number of Relevant Documents 212 Generality Number 4.3 Coord - L'mtinn Documents Recall Precision Ratio Fallout Level Retrieved Rel. Non-tel. ala+c Ratio a/a+b Ratio b/bid x y z 1 194 2 23,755 91.5% 73.6% 0.8% 48.520% 35 35 3 156 8,151 35 112 2,014 1.9% 52.8% 19,259% 35 35 35 4 3.7% 5.865% 34 35 35 8 69 625 32.5% 18.0% 0.9% 28 35 18 35 8.0% 18.5% 1.281% 6 34 150 17 43 28.3% 0,307% 35 35 7 6 0.088% 10 7 2.8% 35 35 37.5% 0.020% 3 38 35 FIGURE 4. 601T Index Language 1,8.a (S.T. Synonyms. Quasi-synonyms. Word forms. Coordination) Exhaustlvity of Indexing 3 Search Rule A Document Relevance 1 - 3 Number of Documents in CoLlection 1,400 Number of Questions 35 (Subset l) Number of Relevant Documente 212 Genera[OCRerr]ty Numbcr 4,3 Coord- Documents Ination Retrieved Level Rcl. Non-rel. I 20s (-) 2 186 19,333" 3 146 8,126 4 83 2,483 " 5 50 593 6 28 119 7 10 22 FIGURE 4. 602T Recall Precision FaLlout Ratio Ratio Ratio a/a+c a/a+b b/b+d 96.7% (-) (_) 87.7% 1.0%* 39.455%[OCRerr]' 38.9% 1.8% 16.856% 43.9% 3,6% 5.048% 23.6% 7.8% 1.215% 13.2% 19.0% 0.244% 4.7% 31.3% 0.045% Index Language I. 1.a (S. T. Natural language. Coordination) E[OCRerr]chauetlvlty of indexL[OCRerr]g 3 Search RUle A Document Relevance 1 - 2 Number of Documents in CoLlection 1,400 Number of Questions 35 (8 questions had no relevant documents) Number of Relevant Documents 78 Generality Number 1.6 Docu[nents Retrieved Non-rel. 73 [OCRerr]3.876 55 b.25Z 43 2.063 30 664 16 t77 g 51 3 13 35 O 35 35 23* 35 38 35 35 35 35 35 30 35 35 18 35 35 8 35 35 Recall Precision FaLlout Ratio Ratio Ratio x y = a/a+c a/a+b[OCRerr] b/bl-d 82.4% 0.3% 48.726% 69.6% 0.7% 20.176% 54.4% 1.4% 0.060% 38.0% 4.3% 1.357% 20.3% 8.3% 0,361% Zl.4% 16.0% 0.104% 3.8% lg.g% 0.037% 35 35 35 35 36 35 34 85 35 28 35 36 18 36 35 7 35 35 3 38 35