Introduction to InvestigationOrganizer

InvestigationOrganizer is a Web-based collaborative information management and modeling tool designed to support mishap investigation teams. InvestigationOrganizer provides a centralized information repository that can be used by a distributed team of investigators to store digital products relevant to an ongoing mishap investigation. The repository is intuitively organized so that related pieces of information are cross-indexed and easily accessible. Users can also create and view overarching analysis models that identify causal factors or hypothesized event sequences leading up to a mishap. The causal models are cross-linked with repository items that provide evidence to support or refute the hypothesized causes. These models can be viewed with linear, hierarchical and network diagrams displayed by the user interface.

  • Why did we develop InvestigationOrganizer?
  • What kinds of information can be represented and related?
  • Can graphical information be managed?
  • Can InvestigationOrganizer manage a diverse myriad of details and relationships?
  • Funding
  • Point of Contact
  • Login to InvestigationOrganizer (authorized users only)


    Why did we develop InvestigationOrganizer?

    Conducting an effective investigation is difficult to accomplish in a distributed envrionment where the investigators and the evidence are not centrally located. A large investigation team typically has different members who are organizationally and geographically distributed, and each investigation team member uses different information technologies such as different e-mail tools, document repositories, and databases. Many types of information products -- including hundreds of megabytes of digital images, documents, schematics, and other information -- need to be referenced and archved for use during the investigation. Providing fast, convenient access to such information is critical to the success of an investigation.  In addition, reporting of mishap causes and lessons learned stemming from an investigation has been inconsistent and difficult to document.  An objective of developing a solution to this problem is to enable distributed teams to better organize and analyze investigation information that they collect in distributed ways using the world-wide web, and by integrating diagrammatic aspects of investigation support with database aspects of investigation support.

    Another motivating concern for the development of InvestigationOrganizer is the investigator's need to correlate evidence with analysis models constructed during a mishap investigation.  Typically, the investigation team will develop a set of analytical models to explain the mishap and drive the investigation process.  For example, the team will develop models of mishap causality, such as a fault tree or a MORT (Mishap Oversight Risk Tree) diagram, which break down the possible causes of the mishap into fine levels of detail.  The team may also try to reconstruct the sequence of events leading up to the mishap.  Another model-based approach is to determine the differences between similar situations that did not exhibit a mishap and the mishap situation.  In all these cases, the model describes an hypothesis about some aspects of the mishap, and the hypothesis must be verified or refuted based on the fit of actual evidence gathered by the investigation team.  Although there are failure analysis tools available in the commercial marketplace, these tools do not provide an integral link back to the primary supporting or refuting evidence gathered by the team.

    In summary, current investigators of mishaps and anomalies struggle to deal with the following challenges:

  • Members and/or investigation sites that are geographically distributed
  • Diverse and extensive information that must be organized, analyzed, and shared
  • Lack of standard methods or tools for information storage, management, dissemination or analysis
  • Current methods that are disparate and cumbersome
  • Lack of team members that are formally trained in mishap investigation
  • Lack of learning from past anomalies and mishaps by broader constituents


    What kinds of information can be represented and related?

    Figure 1 shows an example of the flexibility of InvestigationOrganizer to structure and relate a wide range of data.

    Example showing representation capabilities

    Figure 1: Representation Flexibility

    Can graphical information be managed?

    Storing and relating graphical data is usually a critical requirement in the record keeping of an investigation.

    Example showing graphics capabilities

    Figure 2: Relating Data Products

    Can InvestigationOrganizer manage a diverse myriad of details and relationships?

    The right hand side of the InvestigationOrganizer interface in Figure 3 shows a project information record describing an investigation incident. It includes various data fields describing the investigator, the system under investigation, the incident classification, etc. The left hand side of the interface displays links from the incident to various related records, including observations related to this incident, causal models, and the investigation board. The user simply clicks on a link to navigate to a related record. In additional to textual data, a wide range of images, documents, and other original electronic media can also be accessed using the InvestigationOrganizer interface with an example shown in Figure 4.

    Image of main interface screen

    Figure 3: Interface Overview

    Viewing graphics within InvestigationOrganizer
    Figure 4: Viewing Embedded Graphics

    Funding

    NASA funding for the development of InvestigationOrganizer is being provided by the Engineering for Complex Systems Program. InvestigationOrganizer leverages base research and technology developed for ScienceOrganizer with funding from NASA's Computing, Information, and Communications Technology Program under the Intelligent Systems/Human-Centered Computing program element.


    Points of Contact

    For further information on mishap investigation aspects of this research contact Dr. Tina Panontin.

    For further information on information systems aspects of this research contact Dr. Richard Keller.

    NASA Official:  Richard M. Keller, Ph.D.
    Last Updated: February, 2003