Predicting Book Use in University Libraries by Synchronous Obsolescence

Collection management in libraries should be conducted on the basis of analyses of circulation data. In addition to the number of times a book is loaned out, the change in the number of loans over time can be a useful measure for managing books. In this study, we investigate the relationship between diachronous obsolescence (that is, the change in use over time) and synchronous obsolescence (that is, the distribution of use over the elapsed time from accession, for a relatively short period). We evaluate three algorithms based on the two types of obsolescence, in order to predict future book use from past use information. We apply these algorithms to actual circulation data from a university library and investigate prediction accuracy. The results show that the synchronous obsolescence-based algorithm can predict future book use more accurately than the diachronous obsolescence-based algorithms, especially for information recorded over a relatively short period.