| The ability to perform effective situation assessment in complex, dynamic work environments is often challenged by the unsophisticated nature of the data displays which predominate in many of these settings. These displays can create a great deal of cognitive overhead associated with the need to locate, collect and integrate the relevant raw data readings into higher-order, semantic properties on which situation assessment and control decisions can be based. Integrated display techniques have sought to address these problems by identifying and explicitly representing these semantic properties. While promising, such techniques remain an active area of research and many issues remain open, not the least of which concerns the need to deal with meaningful properties that are extended in time (what we refer to as "events").; Observations of front-line operators performing supervisory monitoring suggest that events are a prevalent and important class of properties for system operators. Despite this, current display tools provide at best weak support for the detection and interpretation of events. In general, events have proven to be difficult to deal with, both for display designers and for designers of automated reasoning tools.; The work reported here represents an attempt to develop a better appreciation of how events participate in situation assessment activities. The issue at the center of the research is the question of what is informative to skilled system operators. What is the nature of the information which practitioners actually extract from a continuous, low-level telemetry stream? A study was conducted in which medical professionals monitored telemetry from a simulated surgical scenario, using a standard medical monitoring display. The scenario included a rich set of event patterns varying in terms of their complexity, timescale, and salience. A technique adapted from studies of social perception was used to identify points where observers noticed "meaningful" patterns in the telemetry data. Cued retrospective verbal reports were collected to uncover the nature of the patterns that the observers responded to. The results offer converging evidence for the fundamental importance of events in situation assessment and emphasize the intelligence required to extract meaningful events from low level telemetry displays. |
