Knowledge-intensive subgroup mining

More about the book

Subgroup mining is a powerful data mining approach aimed at discovering novel and useful knowledge through subgroup patterns. However, real-world applications often face challenges such as scalability issues with large datasets, overwhelming results, and the prevalence of already known patterns. This thesis introduces a combination of techniques to address these challenges. It presents the SD-Map algorithm, which is both fast and effective for automatic methods. Additionally, it describes interactive techniques for subgroup introspection and analysis, along with advanced visualization methods for subgroup optimization, comparison, and exploration. The approach also incorporates various classes and types of background knowledge into the mining process, creating a knowledge-intensive framework that supports both automatic and interactive methods. The evaluation comprises two parts: an objective assessment of efficiency and effectiveness through thorough experimental evaluation using synthetic data, and a subjective assessment focusing on user acceptance, benefits, and the interestingness of results. The proposed methods have been successfully implemented in medical and technical applications, with five case studies utilizing real-world data demonstrating their effectiveness.