Bibliomining the Digital Library in Determining User Pattern Behavior: A Case Study at UTM Library

Abstract:

The paper described (a) the investigation undertaken to examine the characteristics of data from data reservoirs regarding user/patron information and circulation information. (b) The information seeking to explore the patterns and trends among these data reservoirs using data mining analysis with about 957,224 borrowing history and overall 31,052 registered readers and  139,195 title author of books from the Universiti Teknologi Malaysia library since 2008 to 2010. (c) To study how constructed patterns and trends generate informed decisions on market basket analysis algorithm via the education student. Today it is undeniable for a library relying on the outcome of the digital system (bibliographic data) to plan for future library development and activities, however not many realized the potentials of the data and used it as an advantage.
 
Design/methodology/approach - The data set consists of patrons’ data and loan registration records between 2008 and 2010 which are contained in the digital library of UTM Library main campus. Patron’s ID number, identification type, department and gender are included in each patron’s data. The loan registration record has a huge data set of about a million records and therefore, efficient technique must be applied in the process of analyzing and pre-processing. Simple probabilistic statement about the co-occurrence of certain events in a database and is particularly applicable to sparse transaction data sets. The data mining analysis was conducted by using cluster analysis, frequency statistics, averages and aggregates and market basket analysis (MBA) algorithm. Association rules are used to show the relationships between data items.