基于用户交互加权的最小程度描述原则的数据钻取的优化

Interaction exploration is an effective way to obtain potential information from the treemap that visualized large-scale data. Data exploration techniques that has been proposed (such as translation and scaling on a two-dimensional plane) provide limited context information only or provide excessive distortion. Query was proposed to get interested view, but it need clear destination. The method that get a detailed view according to scores evaluated by user interaction can\'t get a suitable initial view. To this, we propose an interactive exploration method that performs unbalanced weighting according to the user's focus and uses the minimum description length principle (MDL) for node aggregation. This method can freely expand the focus area in a good degree of aggregation, which can provide the necessary data information and reduce the visual confusion caused by the excessive data set. This enables efficient data mining in large-scale datasets. We further verify the availability of method by experiment. In addition, the interaction mode in this paper has no conflict with the existing interaction modes (such as zooming, rotating, etc.), and the better results can be achieved by selectively combining according to different scenarios.