A computer-based platform for visualization and interactive analysis of complex data

点击翻译按钮获取中文摘要

Scientific/engineering data are becoming increasingly complex and massive. Typically, such complex data are represented discretely on large grids over space and time. For decision-making, investigations based on these data need to critically analyze data attributes and their interrelations. Routinely, human experts in different disciplines undertake such analyses collaboratively in a co-located setting. The characteristics of the data imposes, however, difficulties for experts’ analyses.  Current practice for analyzing complex data relies mainly on visualization, which results in visual clutter and occlusion overloading the human cognition.  Common methods for interacting with visualized data are inadequate to meet the socio-human needs, causing ineffective collaboration. These difficulties are a bottle-neck issue, which deters experts from making use of complex data in decision-making. Hence, this research program aims to create enabling technologies to target this issue.  Its five-year objective is to catalyze a computer-based virtual environment (VE) platform, which facilitates the experts to visualize and interact with key patterns of the data for collaborative analyses. As outcomes of the program, the vital enablers of the platform are new algorithms for recognizing key patterns reliably and rendering the patterns effectively to ensure intuitive visualization and interaction. Using complex flow data as a heuristic case, the creation of the algorithms is an interdisciplinary endeavor in collaboration with engineering scientists of fluid dynamics at both national and international levels. It is generally challenging to recognize key patterns of complex data; to depict the patterns by complying with the human cognitive constraints; and to analyze interrelations among the patterns. The research program meets the challenge through two interwoven approaches: (1) integrating pattern recognition with visualization and interaction; and (2) considering the human cognitive constraints and socio-human needs. From this program, two novelties arise: one is to compute dynamics of a data domain independently from other domains, permitting versatile applications of the recognition algorithms to identify key patterns. Another is to represent energy associated to the key patterns by using haptic (pertinent to the sense of touch) feedback, prompting a unique way of stimulating the human senses to achieve cognitive responses. Although based on flow data, the creation of both recognition and rendering algorithms signifies the solution to the bottle-neck issue for experts’ analyses of complex data. This solution enables the VE platform to be tailored for experts’ actions including: flood mitigation for planning public safety; improving design processes for boosting industrial competitiveness; and effective surgical training/planning for innovative medical treatments.