EAGER: Collaborative Visualization for Knowledge Computing

University Of California-Santa Barbara, Santa Barbara CA

Investigators

Abstract

This work proposes novel approaches to collaboratively visualize and interactively explore large interconnected high-dimensional data sets represented as large unstructured graphs. The proposed work builds upon scalable visualization methodology that is able to support interactive exploration of over a hundred thousand nodes on standard desktop computers, even without the use of hierarchical clustering. This research advances the state of the art in interactive network visualization in the following three research directions towards a visual tool set for collaborative interactive sense-making of social network data, represented as interconnected graphs: Web-based Scalability, Collaborative Graph Analysis, and New Visualization Paradigms, all together facilitating the visualization and collaborative comparison of large graphs from (and over) the semantic web. A specific focus of these novel visualization techniques is the visualization of uncertainty associated with the data sources or inferences. The intellectual merit of this work is established by novel contributions to the fields of information and scientific visualization, as well as human-computer interaction: Interactive collaborative manipulation and analysis of very large graph structures is not currently possible with existing tools and techniques. Our proposed research provides a new collaborative and user-driven perspective on social network analysis which is interactive, flexible, scalable and extensible enough to keep pace with the rapid expansion of the social web. The intended results, a novel drag-and-drop approach for down-scaling, testing, comparing, and evaluating information networks, will establish and evaluate novel mechanisms to manage the onslaught of networked information in a collaborative manner. Results will be demonstrated through integration and evaluation of the novel visualization and interaction techniques within a semantic web-based analysis and interaction infrastructure. The proposed work will enable significant broader impacts by making innovative scalable graph visualization methodologies available to broad audiences via open-source semantic web platforms. Web audiences will be enabled to access novel state-of-the-art graph analysis and visualization tools directly in their browsers without the need for downloading applets, plugins, or virtual machines of any kind. The PI will be using the proposed research as a case study and platform for projects supporting the teaching of human-computer interaction fundamentals. Interdisciplinarity is a cornerstone of successful user interface technology projects such as this one, and the investigator's research group has a demonstrated commitment to collaborations and partnerships with other departments on campus, as well as representatives from industry and the public community, targeting broad dissemination of the research results.