IRS13 Thesaurus, Phrase and Hierarchy Dictionaries 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. VII-20 Evaluation Number and Percentage* Input and Measure Used of Individual Requests To Determine Thesaurus Stem Both Collection Thesaurus Type Merit Superior Superior Equal IRE-3 Abstract Normed. Recall 28 82.4% 6 17.6% 0 34 Requests (Thesaurus-3) Normed. Precision 26 76.5% 8 23.5% 0 Abstract Normed. Recall 24 61.5% 15 48.5% 3 CRAN-l (Thesaurus-3) Normed. Precision 22 56.4% 17 43.6% 3 42 Indexing Normed. Recall 21 53.8% 18 46.2% 3 Requests (Thesaurus-2) Normed. Precision 25 64.1% 14 35.9% 3 Text Normed. Recall 22 64.7% 12 35.3% 1 ADI (Thesaurus-i) Normed. Precision 23 67.6% 11 32.4% 1 35 Abstract Normed. Recall 20 62.5% 12 37.5% 3 Requests (Thesaurus-i) Normed. Precision 24 70.6% 10 29.4% 1 *Percentages do not include cases where dictionaries have equal merit Comparisons of Individual Request Merit giving the Number of Requests Favoring Stem and Thesaurus for Five Runs in Three Collections, according to Merit Assigned by[OCRerr]Norma1ized Recall and Precision Fig. 13