IRS13
Scientific Report No. IRS-13 Information Storage and Retrieval
Search Matching Functions
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.
111-6
search time frequently becomes a problem, requiring the batching
of search requests for reasons of economy. One specific part of
the search time problem reflects on the use of the logical relations
in the search formulations, since complex logic may tend to increase
search time. This problem is circumvented to some extent in the
NASA search system by the use of term weights to get the same results
as logical formulations, with reduced machine effort. [1]
3. A third problem relates to the cut-off, which is rarely critical
in manual systems where a searcher/system interaction may control
the search as it proceeds, but in mechanized systems the total
search formulation IIrLiSt be made in advance of the search and no
changes are normally possible during the search. This problem
is again linked to the logical relations in the formulation, since
too few documents will be retrieved if the search logic is made too
restrictive and tt[OCRerr]g[OCRerr][OCRerr]I; contrariwise, if it is made too `looset
more documents than the user is willing to examine may be retrieved.
An example of this latter point is given using an actual search
made in the Medlars system. The request statement and search formulation
are shown in Figure 2, and the search of the total document file of nearly
half a million items eventually produced eleven references. The user was
asked, before the search was carried out to indicate the number of journal
articles relevant to his request that he considered likely to have been
published since January l96[OCRerr]; 6 - 20 was chosen as a likely range for the
number of relevant items. It is surprising therefore that after receiving
the 11 references retrieved in the search, the requestor did not seem to be
satisfied, and the librarian who handled the request reports the requestor
as saying, "In view of the small return ... Dr. X would be interested in
any articles dealing with oxygen concentration in the cerebrospinal fluid."
Clearly the system and the search formulation are not at fault in this case:
it is just a case of a change in requestor need, presumably arising from the
requestor's examination of some retrieved document. This changed need,
in the Medlars system, requires a completely new search of the system.