Discovering knowledge in information centers data with data mining methods

With the impact of big data caused by information technologies, both individuals and institutions are facing an intense information bombardment. Accessing the needed information amidst this abundance can be both laborious and costly. However, through the application of appropriate techniques and methods on large data sets, valuable information of critical importance can be extracted. The aim of this study is to uncover user behaviors, user patterns, collection patterns, and trends from the data stored in library automation systems using data mining methods; and to provide a roadmap for collection development, collection management, and effective use of information resources in information centers. The study has two main motivational focuses: (i) to recommend information resources relevant to each user's area of interest based on their tendencies to use information resources; (ii) to provide decision support to library management and top management in the development of products and services that meet users' expectations and tendencies. For the execution of the research, loan records of printed books related to users held in the library automation system (Yordam Library Information and Document Automation) used by the Library and Documentation Department of Burdur Mehmet Akif Ersoy University were utilized. The obtained data set was preprocessed through data cleaning, data transformation, and data reduction processes, and analyzed using RapidMiner Studio Educational 10.3 and Spyder IDE v5.5.4 software packages. Descriptive statistics were provided in accordance with the aim of the study, the most borrowed books on an annual basis were explained, and the subject distributions and annual loan numbers of the borrowed books were presented. Moreover, book preferences were examined according to gender and age groups. Subsequently, k-means clustering analysis and SVD collaborative filtering analysis were applied for the personalized book recommendation system. At the end of the analyses, book recommendations were listed, and the system's RMSE performance values were shared.