Interactive Data Mining

Yan Zhao's Ph.D. dissertation, entitled "Interactive Data Mining" provides a conceptual framework and a systematic study of human-computer interactions and collaborations for effective data mining. The thesis is based on an assumption that the effectiveness of data mining systems depends crucially on semantics information about data and different user requirements. In contrast to many data mining models that concentrate on automation and efficiency, interactive data mining systems focus on adaptive and effective communication between human users and computer systems. Interactive systems fully explore the power of human intuition, creativity, heuristics and strategies with supports from computers.The thesis is well-balanced between theoretical investigation and experimental evaluations. A user-oriented three-layered conceptual model is proposed. Within the framework, user perceptions and requirements are studied formally at the philosophical, technique and the application layers. The separation of the three layers leads to many new insights into data mining. Based on the conceptual framework, a prototype of Interactive Classification System (ICS) has been implemented.