Solving problems in Library and Information Science using Fuzzy Set Theory

Library Trends, Wntr, 2002 by William W. Hood, Concepcion S. Wilson

   A small committee of experts is formed. For reasons of simplicity, we shall
   consider a team of two experts. Three criteria will be used:

   --number of citations obtained by the journal, as measured by ISI's
   (Institute for Scientific Information) citation files;

   --percentage of missing issues;

   --number of circulations (local use) per issue.

   Each committee member must decide on his/her membership function for each
   of these variables. So, although each of these criteria can be measured in
   an objective way, the interpretation of the measurements with respect to
   the ultimate binding decision is subjective and requires an application of
   concepts borrowed from fuzzy set theory.

   When experts have decided on membership functions, every journal set can be
   judged on all criteria. This can now be done in a straightforward way and
   no longer requires a specific intellectual input.

   Finally, each expert must also have decided, beforehand, on the relative
   importance of each of the three criteria, and the library committee must
   have decided on the relative importance of each expert (before the data
   were collected!). This leads to a ranking of journals according to their
   suitability for binding. (Egghe & Rousseau 1990, pp. 200-201)

A similar approach can be used to making tattletaping decisions of periodicals (Turner, 1981). The methodology outlined above may help the library produce a list of journals that are the most likely candidates for tattletaping.

Information Retrieval Applications of Fuzzy Set Theory

The main application for Fuzzy Sets in LIS to date has been in the area of IR. As mentioned earlier, the concept of "relevance" is essentially a fuzzy concept; for any search, there will be documents that are more relevant or less relevant than others. IR is concerned with retrieving documents that meet some particular user need, or are relevant for some particular situation. The earliest attempts to apply Fuzzy Sets in this area appear to be those of Tahani (1976) and Radecki (1976). An ARIST review of this area is prodded by Bookstein (1985) and a survey of the use of FST in IR and databases is given by Kerre, Zenner, & De Caluwe (1986). A theoretical background to the application of Fuzzy Sets to IR is given in Radecki (1983).

Traditionally, the main mathematical tool in IR has been Boolean algebra. Nearly everyone who has done any searching using bibliographic databases (such as those available through the DIALOG information systems), or searched library catalogs or the World Wide Web has used Boolean operators to construct sophisticated searches. In turn, Boolean algebra is based on Set Theory: Each search or index term results in a set of retrieved documents, which can then be combined using the Boolean operators (AND, OR, NOT). An IR system can be regarded as consisting of a set of "documents" and a set of "index terms." Each index term corresponds to a set of documents, which will be a subset of the universe of all documents in the system. This subset will consist of all those documents related to the index term. Traditional Boolean searches correspond to set operations on these index-term subsets.