InvestigationOrganizer (IO) is a Web-based information system that integrates the generic functionality of a database, a document repository, a semantic hyper-media browser, and a rule-based inference system with specialized modeling and visualization functionality to support mishap investigations. The semantic hypermedia component includes a customizable ontology that specifies various types of items (people, places, events, causes, systems, and associated information products) relevant to mishap investigations. The ontology also describes important properties of each item type, and details the potential relationships among items.
The graphical interface mirrors the hierarchical causal structure defined by investigators within the semantic hypermedia system. The causal structure is displayed in the form of a tree-structured graphical representation that provides information on the supporting and refuting evidence for each causal factor. Users can move back and forth between the default Web-based hypermedia display and the specialized tree display. The causal tree is based on causal factors defined in the semantic hypermedia network, and interrelates the causes with evidence and associated investigatory information (documents, digital images, and other electronic products) stored within the network.
Items stored in the repository are classified according to the ontology. Users can set properties of stored items, and can associate a relevant electronic file with an item. Users can also establish links between items based on the relationships defined in the ontology, and the items can be viewed with the hyper-media browser by using the established links to navigate through the space of interrelated items.
In addition to providing repository functionality, users can create and view overarching analysis models that specify causal factors or hypothesized event sequences leading up to the mishap. The causal models are linked to repository items that provide evidence to support or refute the hypothesized causes. The models can be viewed with linear, hierarchical, and network diagrams displaced by the user interface. The InvestigationOrganizer system is built as a customized application of the SemanticOrganizer knowledge management system.
This work was done by Richard Keller, Tina Panontin, Robert Carvalho, James Williams, Shawn Wolfe, and Yuri Gawdiak of Ames Research Center; and Ian Sturken of QSS Group, Inc.