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-9 The discussion of unclear requests in part [OCRerr]D is closely linked to the relevance decisions, since as J. O'Connor shows [1], relevance disagree- ments are often due to unclear request forms; furthermore, since many requests that are thought to be clear are not so in fact, one is led to different request interpretations and hence to different relevance decisions. Probably more examples than the five given in part [OCRerr]D exist, but by the stringent criteria for clarity suggested by O'Connor, many real user requests would be regarded as unclear also. Several of the requests deal with quite similar topics, and sometimes do not have as many relevant documents in c[OCRerr]mon as the requests suggest. Exan[OCRerr]les are requests AlS and 38, Bl and B16, B9 and Bil, and A5, B3 and 36. A clear error of judgment is seen for document 7, where the photo cam position method that is described for producing NASA's "Scientific and Technical Aerospace Reports" is thought relevant to request A7, [OCRerr]ihich demands documents on systems for producing original papers by computer. The request preparer probably did not realize that NASA STAR is not a series of original reports. No specific examples have been found of documents that should have been recognized as relevant, except in those cases where two or more requests seem very similar, as noted. 5. Request Performance A) General Performance Analysis Methods. It was intended to divide the individual requests into three groups, namely: a) Requests which perform badly on all processing options; b) Requests which perform well on some options and badly on others; c) Requests which perform weli on all options.