Library Service Optimization with Machine Learning: Concepts, Implementation, and Evaluation

Libraries, such as the ZBW – Leibniz Information Centre for Economics,  serve as vital institutions for the access and dissemination of knowledge, yet they face ongoing challenges in the digital era. This thesis delves into three pivotal components aimed at bolstering library systems: the utilization of document titles for classification and recommendations, the analysis of trends in user-generated content, and the enhancement of recommender systems through deep learning.

The findings of this thesis contribute to the enhancement of the discovery systems of ZBW through the exploration of document titles for classification and recommendations, user-generated content trend analysis, and deep learning in recommender systems. The study reveals that leveraging document titles can be a practical and resource-efficient approach for libraries, with competitive results compared to full-text analysis. The ”attend2trend” model introduces an interpretable method for predicting emerging trends in user-generated content, and the integration of Paragraph Vectors improves recommendation quality, enhancing user experiences within library systems.

Rights

Use and reproduction:

No license. The provisions of the German Copyright Act (UrhG) apply.

Please note that individual components of the publication may be subject to other licensing or copyright conditions.

Cite

Citation style:
Could not load citation form.