A literature search and file organization model

A principle of sequential optimization in search theory distributes the search effort at each stage so as to maximize the probability of target detection with the effort expended thus far. As an application of this principle to the search for pertinent items in a literature file, the file items should be arranged in decreasing order of the probability that an item will yield the information sought. Complete ordering in this manner may not be feasible, and it is proposed that the files be partially ordered in search zones with some loss in search efficiency. A model for assessing the relative efficiency of partial ordering is developed and used to determine optimal zone sizes under an assumed target distribution pattern. In this way, trade-offs between file organization effort and search effort can be meaningfully evaluated. A representative target distribution function for research literature appears to follow the Bradford law of scattering. This function is used to demonstrate the application of the model over a meaningful range of parameters obtained from empirical studies. It is shown that a good two-zone search plan for scientific literature is one in which 15–20% of the most useful documents are examined first. Only about one of three searches should have to go on to a search over the remainder of the file.