A Hybrid Intelligent Recommender System Based on Data Mining for Personalizing Services in Digital Libraries

Introduction
The rapid growth of digital resources has increased the difficulty of identifying relevant information efficiently in digital libraries. Traditional search and filtering approaches are often unable to address users’ diverse and personalized information needs. Intelligent recommender systems have therefore emerged as an important solution for improving information discovery and user experience. However, relying solely on content-based or collaborative filtering may lead to limitations such as the cold-start problem, popularity bias, and insufficient recommendation coverage. Accordingly, this study aims to develop a hybrid intelligent recommender system based on data mining to personalize services in digital libraries. The proposed approach combines users’ behavioral information with content-based features to generate more accurate, diverse, and reliable recommendations.
 
Methodology
This study developed a hybrid recommendation framework consisting of three main components: data preprocessing, two complementary recommendation engines, and an intelligent aggregation layer. The first engine uses content-based information to identify resources with characteristics similar to users’ interests, whereas the second employs collaborative filtering based on users’ behavioral patterns and interactions. The outputs of these two engines are subsequently integrated through an intelligent aggregation mechanism to produce the final recommendations. Behavioral data were obtained from the Book-Crossing dataset, while content-based information was extracted from both the Book-Crossing and CiteULike datasets. The performance of the proposed system was evaluated using precision, recall, diversity, coverage, and a simulated user-satisfaction measure. The evaluation focused on comparing the hybrid approach with single-strategy recommendation methods.
 
Findings
The findings demonstrate that the proposed hybrid recommender system performs better than single-strategy recommendation approaches in terms of recommendation accuracy and coverage. Combining behavioral and content-based information enables the system to provide recommendations that are simultaneously more precise and diverse. The results also indicate that the hybrid framework can alleviate important limitations of conventional recommender systems, particularly the cold-start problem and popularity bias. The integration of data mining with semantic analysis of textual content contributes to more effective modeling of user preferences and resource characteristics. Overall, the experimental results confirm the effectiveness of combining complementary recommendation strategies for improving personalized information services in digital libraries.
 
Discussion and Conclusion
The findings suggest that hybrid recommendation provides an effective framework for addressing the complexity of personalized information retrieval in digital libraries. Integrating collaborative and content-based approaches allows the system to exploit both behavioral patterns and the semantic characteristics of digital resources, thereby improving recommendation quality and coverage. From a practical perspective, the proposed framework can help digital libraries deliver more personalized services, improve users’ information-search experience, and facilitate more effective utilization of available resources. Its scalable and adaptable structure also provides potential for application in next-generation digital library environments. In conclusion, combining data mining, collaborative filtering, and semantic content analysis can provide a practical foundation for developing intelligent and personalized recommendation services in digital libraries.