SBIR Phase I: Scale Discovery in Data-Driven Ecological Modeling

This Small Business Innovation Research Phase I project will develop a relational Bayesian modeling framework for automated discovery of tamporal and spatial scale in ecological modeling. Scale resolution and cross-scale articulation are key problems in scientific model development, especially for complex ecosystem-scale and mixed human-natural systems modeling. Modeler's facilities for automatic generation of aggregators and for clustering based on hidden variables provide a natural and tractable basis for scale resolution and cross-scale articulation for ecological modeling. The objective of this Phase I project is to develop both fully automated and mixed-initiative scale discovery methods. A mixed-initiative approach, in which intelligent computing is used to assist humans to harness the vast amount of data at their disposal, will permit investigator-driven development and exploration of the multiple, simultaneous hypotheses necessary to describe complex ecosystem behaviors. The immediate commercial application of this project is in the area of ecological modeling. However, the core technology may find additional uses in biomedical research, epidemiology and commerce.