The ViziMine tool provides a comprehensive visualization tool to support the cooperative data mining process. This tool gives the user the ability to visualize the data and the data covered by the rules in various ways in order to understand it, and to gain new insight into the data mining process. The ability to visualize the results of the data mining effort, both during individual and cooperative learning, helps the user to understand and trust the knowledge embedded in it. The tool thus gives the user the ability to get an intuitive "feel" for the data and the rules created. This ability can be fruitfully used in many business areas, for example, for fraud detection, diagnosis in medical domains, and credit screening, among others.
Recall that the user is modeled as one of the participants in the CILT cooperative data mining environment, as visualized by the ViziMine tool. The visualization process is chronological, following the data mining life cycle, and is thus intuitive, easy to use, and understandable (Multiple Authors, 2000). ViziMine provides a mechanism to enable the user to monitor the data mining process, its inputs, and results, and to interact with the cooperative data mining process and influence the decisions being made. In this way, the powerful human visual system is used to increase the user's understanding of and trust in the data mining effort.
Future development of the ViziMine tool will include the study of a suitable visualization technique for a variety of data set sizes and types, as well as an investigation into the scalability of our approach. This will allow the tool to be compatible with additional real-world data mining problems. The use of our approach for Web mining, with the subsequent application thereof for e-commerce and e-business, should be further investigated. Recall that current implementation of the ViziMine tool incorporates the C4.5, CN2, and ANNSER data mining tools. Other data mining tools may be incorporated through the transformation of their outputs to DNF rule format. This aspect will be further investigated.
Virtual reality and virtual collaborative environments are opening up challenging new avenues for data mining. There is a wealth of multimedia information waiting to be data mined. In the past, due to a lack of proper content-based description, this information was neglected. With the recent advent of a wide variety of content-based descriptors and the MPEG-7 standard to handle them, the fundamental framework is now in place to undertake this task. Virtual reality is perfectly adapted to manipulate and visualize both data and descriptors. VR is also perfectly adapted to analyze alphanumerical data and to map them to a virtually infinite number of representations.
VEs are intuitive and, as such, can help specialists to efficiently transfer their analysis to upper management. Data are distributed worldwide and enterprises operate from various remote locations. These enterprises have a huge amount of data but they lack the right framework to convert them into a valuable asset. Collaborative virtual environments provide a framework for collaborative and distributed data mining by making an immersive and synergic analysis of data and related patterns possible.