Information Gathering about Decision Processes Using Geospatial Electronic Records.

Peter Bajcsy and David Clutter

Technical report NCSA-ISDA06-001 January 12, 2006

We address the problems of gathering, archiving and analyzing information about decision making processes using geospatial electronic records (e-records). Our ultimate goals are to gather all observable variables about decision making processes using input geospatial e-records so that (a) a decision making process can be documented in the future, (b) audits and authentications of records can be executed, and (c) accountability of decisions can be achieved.

Our objectives are (1) to evaluate the tradeoffs between the amount of information about decision making processes, e-record authentication and audits, and the associated storage and computer performance cost, (2) to develop prototype software for gathering information about the use of computers in high assurance decision making and high confidence application scenarios, such as in military, medicine, or law-enforcement, and (3) to provide guidelines about the future uses of computers in government decision making. To meet the above objectives, one has to perform studies addressing several high performance computing problems related to information gathering, data storage, management, access and retrieval, and novel computer architectures.

We report our preliminary results obtained for a class of decision processes using geospatial electronic records that are related to emergency response decision processes (e.g., foreign plant disease, hurricane Katrina scenarios), land use decisions (e.g., urban development in the third-world countries) or regulatory decision processes (e.g., nutrient levels allowed by EPA). For this purpose, we have prototyped an information gathering system using multiple geospatial analytical tools that support this class of decision processes. Our tradeoff analyses are performed at several levels of information granularity and the results are reported for three distinct decision process types, such as data intensive, computationally intensive, and high complexity decision processes. The preliminary experimental results indicate that unique signatures of decisions processes based on gathered information could be established for understanding of gathering, archiving and retrieval requirements to provide guidelines about the future uses of computers in government decision making.