MiddGuard: A Framework for Collaborative and Extensible Visual Analytics, Poster 12

We have developed MiddGuard, a web-based visual analytics framework designed for customizable, collaborative investigations. Designed for developers and investigators, MiddGuard provides a framework to ingest, process, and... [ view full abstract ]

We have developed MiddGuard, a web-based visual analytics framework designed for customizable, collaborative investigations. Designed for developers and investigators, MiddGuard provides a framework to ingest, process, and visualize data in a way that complements the analytic process. The core of MiddGuard is a graph composed of nodes, each of which performs an analytic operation or creates a visualization of the data. Investigators alter and extend the graph throughout the investigation based on previous analytic operations and views of the data. Developers contribute the graph’s nodes, which perform analysis that is agnostic of the data source and destination. MiddGuard orchestrates the nodes’ inputs and outputs and provides easy to use APIs to access the data within the node. It facilitates synchronous and asynchronous collaboration by keeping graphs, nodes, data, and visualizations in sync between connected investigators and stored in a central database. The framework includes a chat-like system for annotations. MiddGuard’s abstractions for dealing with data and extensibility make it a powerful framework for developers and investigators in visual analytics.