I'm currently working on the Concept Map Tool with Dr. Carlyle Brewster. My main area of research in this project is the development of a concept map evaluation tool. See the Concept Map Tool web site for details. I have recently replaced the "down and dirty" expert system grading system with a version based on the Jess expert system shell from Sadnia National Laboratories. Jess is an embeddable Rete-based rule system written in Java. Q: What's cooler than Jess? A. Nothing.

Previous research projects in ISIS

I'm doing some limited research in ISIS related to applying simulated annealing to two processes in the watershed tool project. (See Projects). Simulated annealing is a probabilistic hill climbing search technique analogous to the physical annealing process.

We have some incomplete information about farms and fields in the watershed we are studying. We don't have information about which farms are associated with which fields. The annealer is being used to assign fields to farms based on proximity and the viability of a particular operation (e.g., large beef farm) based on the farm resources.

The annealer will also be used to assign cropping rotations to fields in the watershed.

The idea for using simulated annealing for the tasks comes from Nick Stone. Jim Pease of Ag. Econ. developed the farm characteristics.

Independent Research

Outside of my life at the University, I have developed several simulators in the area of artificial life and learning agents.

• The Flask is an artificial chemistry simulator. A good review article on the subject is "Artificial Chemistries -- A Review" by Peter Dittrich et al in the Summer 2001 issue (Volume 7, Number 3) of Artificial Life. My simulator uses the stochastic molecular collisions approach.
  
• The Pond is a simulator I'm using to study agent learning in difficult environments. This simulator will also be used to study cooperating learning agents. (This project was originally called "JavaScape".)
  RepeaterBot is simulator for studying cooperating agents that carry radio repeaters and attempt to explore an unknown area while providing full radio coverage.

I'm also working on using HTML renderers to extract web content automatically. I use HTML Parser, JRex, and Cobra from the Lobo Project to build Document Object Models (DOMs). Using either JRex or Cobra, I extract HTML content after JavaScript is applied. See my technical notes page for more details.

An old research project I hope to return to is in the area of natural language generation. See the notes on my dissertation research.