ISR11 Design Consideration for Time Shared Automatic Documentation Centers chapter M. E. Lesk Harvard University Gerard Salton Use, reproduction, or publication, in whole or in part, is permitted for any purpose of the United States Government. X-15 in question will be solely occupied by SMART data, it is highly unlikely that anyone else will want access to them. But the use of mLlltiple requests, or of the multiplexor channel by some other 1-0 device, might cause difficulty. Since internal operations are fast compared with the read rate of the documents, time-shared programs which do not require 1-0 devices should not affect the speed of the SMART programs seriously. The major prob][OCRerr]em that arises in connection with this timin.[OCRerr] estimate 10 that the time-sharin[OCRerr] supervisor may 510' do'.[OCRerr]n the programs excessively. One well-publicized present-day time-sharing system slows down programs by a factor of 6o. It is expected that future systems will be more efficient. Other means of saving time are: 1) increased segmentation of the collection; 2) correlation algorithms that begin with the heavily weighted concepts, and do not go on to process the lightly weighted concepts unless the correlation has a chance of being above the cutoff; 3) use of higher-speed input-output equipment (this need not cost more; the system could make a quick, rough, scan of the request during the first phase of the look-up and transfer the clusters that are likely to be needed from the data cell to the disk or drum); [OCRerr]) type out correlations above cutoff as they occur, rather than waiting until all have been found; 5) clustering the clusters. Also, various other printouts could be made during the waiting time, (such as time-of-day, number of documents expected to be retrieved by the search) and options could be presented to the user (e.g. nun[OCRerr]ber of output documents wanted) to use the waiting time more efficiently.