IRS13 Document Length chapter E. M. Keen Harvard University Gerard Salton Use, reproduction, or publication, in whole or in part, is permitted for any purpose of the United States Government. V-20 EVALUATION NUMBER AND PERCENTAGE* MEASURE USED OF INDIVIDUAL RE[OCRerr][OCRerr]UESTS COLLECTION DICTIONARY TO DETERMINE ABSTRACT TITLE IBOTH MERIT SUPERIOR SUPERIOR EQUAL Stem Normed. Recall 29 (85.3%) 5 (l[OCRerr].7%) 0 rRE-3 Normed. Precision 28 (82.24%) 6 (17.6%) 0 3[OCRerr] Thesaurus-3 Normed. Recall 30 (88.2%) 14 (11.8%) 0 Requests Normed. Precision 32 (914.1%) 2 ( 5.9%) 0 CRAN-l Stem Normed. Recall 30 (75.0%) 10 (25.0%) 2 Normed. Precision 30 (75.0%) 10 (25.0%) 2 142 Thesaurus-3 Normed. Recall 26 (66.7%) 13 (33.3%) 3 R[OCRerr]uests Normed. Precision 22 (57.9%) 16 (142.1%) 14 Stem Normed. Recall 214 (72.7%) 9 (27.3%) 2 ADI Normed. Precision 214 (72.7%) 9 (27.3%) 2 35 Requests Thesaurus-i Normed. Recall 25 (75.8%) 8 (214.2%) 2 Normed. Precision 26 (78.8%) 7 (21.2%) 2 * ?ercentages exclude requests when both equal Comparison of individual request merit, giving numbers and percentages of requests favoring abstracts and titles on three collections each using two dictionaries, according to merit assigned by the normalized evaluation measures. Figure 9.