Akademik Veri Yönetim Sistemi
14th International Workshop on Computer Science and Engineering, WCSE 2024, Phuket Island, Tayland, 19 - 21 Haziran 2024, ss.86-93
Digital banking provides customers access to a wide range of banking services with rich graphical user interfaces to conduct banking activities efficiently and effectively. Many digital banking sites analyze end-users’ navigational patterns in order to extract information that can be used to increase customer loyalty, and to predict their next action. This can be accomplished by utilizing machine learning techniques. However, data with high dimensions pose computational challenges to machine learning, as well as an increased risk of overfitting, making it difficult for meaningful patterns to be extracted. Embedding techniques can help mitigate these issues by transforming complex, high-dimensional data into a manageable, low-dimensional space, which makes it possible for machine learning algorithms to perform effectively. Embedding methods that utilize graph structure-based embedding approaches are required to capture and model user behavior to provide better predictions. In this study, graph structure-based embedding approaches are proposed as a means of representing user navigational patterns during browsing. A prototype implementation of the proposed embedding approach is provided in order to facilitate the validation of the approach. Experimental results suggest that the proposed approach has the potential to capture the navigational behavior of the user.