THE BASIS FOR A THEORY OF INFORMATION RETRIEVAL J. Farradane School of Library & Information Science, University of Western Ontario, London, Ontario, Canada N6A 5B9. Any theory which attempts to explain and predict the behaviour of a system, or to form the basis for construction of a system, must be founded on evidence of the characteristics and behaviour of that system, or of its components. Information retrieval systems have been so much of an empirical nature that there has been little evidence on which to base reliable theory. Some systems have been constructed on principles of linguistics, such as De 5aussure's theory in Gardin's "Synto]" system, or on the assumption of the interrelation of words by their frequency of occurrence or co-occurrence, but both such approaches assume that linguistic structure always has a clear relation to meaning. Meaning originates in thought, but language is only a surrogate, and often a poorly used surrogate, for such thought. Many theoretical studies have been based on probability theory applied to retrieval possibilities, but such theories have had little basis of experimental facts, and have not been developed to make predictions for which experimental verification has been sought. There are in fact
/lp/association-for-computing-machinery/the-basis-for-a-theory-of-information-retrieval-It6gJ8LMb0