IRS13 An Analysis of the Documentation Requests 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. X-2 was familiar with the subject field of documentation, it was suggested that some familiarity with the subject would be gained by looking at the docu- ments in the collection. The task of request preparation should then follow. Specifically, requests thought likely to be asked by workers in the field should be devised, but these requests should not build in any particular documents in the collection. No suggestions were made regarding request length. An examination of all documents in the collection should follow, and every document should be judged not relevant or relevant. Neither per- son was familiar with the SMART system. The full text of the 82 documents w[OCRerr] supplied in the form of com- puter print-out, and none of the KWIC indexes or subject categories printed in the published volume were supplied. Both preparers worked independ[OCRerr]tly, one producing 17 requests and the other 18. The task appears to have been carried out as instructed except that it is suspected that some requests may have been prompted by particular documents in the collection. This factor is not necessarily a weakness, since full relevance decisions were obtained, and thus the testis not based on the 1[OCRerr]source documentt1 technique often criticized. Some conunents on the task will be made when discussing TtUnclear Requests't(part 2')) and t1Relevance Decisionstt (part 3). 3. Characteristics of the Requests A) Length Excluding non-subject words contained in the standard "commontt word list used by SMART, the stem dictionary gives an average of 8.o stems per request, and the thesaurus dictionary reduces this to 5.1 concept numbers