Purpose: The purpose of this research is to determine the pattern of access to information resources and analysis of
loan transactions among users of public libraries in Ilam province using the approach of association rules, clustering and
classification in data mining.
Method: In this research, by using data mining techniques, and algorithms of convergence, clustering, and classification rules on data related to the circulation desk of public libraries in Ilam province, the demographic information affecting
the study behavior of members and the patterns of using information resources by this heterogeneous community were
extracted. The data mining process used in the research (CRISP-DM) consists of six pahses, including business understanding, data understanding, data preparation, modeling (rules of convergence, artificial neural network, and decision
tree), evaluation, and deployment, respectively.
Findings: The findings related to different groups using information sources showed that in addition to studying their
textbooks, students also studied subjects such as Iranian history and Islamic religion. In terms of gender, women with
bachelorʼs degrees were interested in studying Persian language and literature, and fiction. The analysis of resource lending in terms of gender showed that 58 percent of active members of public libraries in Ilam province were women and 42
percent of them were men. This difference shows that women have a greater desire to use the information resources of
public libraries in Ilam province. Furthermore, among those who were late in returning their borrowed books, 54 percent
were women and 46 percent were men; In terms of returning books on time, women showed their responsibility with
52 percent and men with 48 percent. The results show that men with diplomas were interested in issues related to Islam
and this group felt more responsibility for returning their books. In terms of gender and its effect on the studied subjects,
undergraduate women, who had free jobs, showed interest in the classes of Persian literature and fiction and did not delay
in returning their books. The analysis of the behavior model of the patrons in terms of book return revealed that the men
with diploma who had free jobs did not delay in returning loaned books, and the women with diploma and bachelorʼs degrees who had free jobs did not delay in returning loaned books. Moreover, students with undergraduate degrees delayed
in returning loaned books. A neural network model was developed to predict the delay in the return of books based on
characteristics such as gender and educational level, and the accuracy rate of the model during the learning process with
the best prediction accuracy was equal to 69.18%. This shows that it is possible to learn and predict the output by receiving
new entries in the database of public libraries of Ilam province.
Originality/value: The deployment of the proposed neural network model on the database of public libraries of Ilam
province for predicting the delay in the return of books based on different characteristics, shows that it is possible to learn
and predict the output by receiving new entries in the database of these libraries. Moreover, using the association rules
technique, it is possible to design a book recommendation system in public libraries of Ilam province.