CRANV2 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. - 186 - Section 8 Abstracts and Titles This section does not introduce any new index language device, since it deals with searches carried out on the abstracts and titles of the 200 document collection. As such it represents a variation in exhaustivity of indexing, and also, in some cases, different concept indexing to that done by the project staff. The searches are done first in single term natural language and then with word forms confounded. Figures 4.900T and 4.901T present the results of searches on the titles only, and Figures 4.902T and 4.903T present the results obtained with titles and abstracts. Figure 4.904P is a plot of the four tables. In order to make a comparison from the viewpoint of exhaustivity, Figures 4.910T and 4.911T present the results of Index Language 1.3.a (word forms) at exhaustivity levels of 2 and 1, on the 200 document collection. Plot 4.912P now gives the performance curves for the three levels of exhaustivity with the project indexing, for the search on titles alone and for the search on titles and abstracts. We were able to do this series of tests by having a print out of a concordance of the titles and abstracts of the 200 documents. This was prepared with the SMART programme, and we are indebted to Prof. Salton for making it available 4. 900T 4. 901T 4. 902T 4.903T 4. 904P 4.910T 4.911T 4.912P LIST OF FIGURES Index Language IV.l.a IV.2.a IV. 3.a IV.4.a IV.1 .a IV.2.a IV.3.a IV.4.a 1.3.a 1.3.a IV. 2.a IV.4.a 1.3.a 1.3.a 1.3.a Exhaust {vity 2 1 3 2 1 No. of Question Document Questions Subset Collection 42 2 200 42 2 200 42 2 200 42 2 2OO 42 2 200 42 2 200 42 2 200 42 2 200 Plot s 4.900T 4.901T 4.902T 4.903T 4. 901T 4.903T 4.201T 4.910T 4.911T