Clinic Projects

Please click on a link below to view the Harvey Mudd College Computer Science Clinic projects for the corresponding time period.

Clinic Projects for 2005-2006

Video Game Interfaces for Power Grid Management

Client
Areva T&D, Inc.

Faculty Advisor
Professor Elizabeth Sweedyk

Student Team
Jason Arold (Project Manager), Michael Beyer, Jeremy Lennert, Robin Schriebman, Matthew Walsh
The Areva T&D Clinic project is to design and prototype a new user interface for Areva T&D power grid management software using ideas from computer game interfaces. In particular we use techniques from real-time strategy games to improve navigation, draw attention to important events, and adaptively display relevant information.

Techniques for Visualizing Large Sets of Business Rules

Client
Fair Isaac Corporation

Faculty Advisor
Professor Christine Alvarado, Professor Belinda Thom (fall)

Student Team
Kristen Kurimoto (Project Manager, spring), John McCullough, Matt Reynolds (Project Manager, fall), Will Shipley
This project investigated various visual structures for sets of business rules. The team developed software for transforming arbitrarily constructed decision trees into Oblivious read-Once Decision Graphs (OODGs) and Exception Directed Acyclic Graphs (EDAGs). The team then analyzed the visual complexity of the different structures to determine whether the OODGs and EDAGs are more human-readable than the original trees. This work will form the foundation for a potential EDAG visualization component for Fair Isaac's Blaze Advisor system.

Computer-Assisted Document Filing

Client
Laserfiche

Faculty Advisor
Professor Melissa O'Neill (fall), Professor Christopher Stone (spring)

Student Team
Rebecca Carson (Project Manager, spring), Russ Osborn, Erik Shimshock (Project Manager, fall), Brian Young
Filing is a tedious chore for organizations storing electronic documents in a large hierarchy of folders. To make this task easier, the team built an easily extended framework that determines a set of likely locations for documents needing to be filed. We generate these recommendations by determining the similarity between a new document and the documents already present in the directory using a combination of text, image, and metadata classifiers.

Faster Disaster News Aggregation and Analysis

Client
NC4

Faculty Advisor
Professor Christopher Stone

Student Team
Ryan Ausanka-Crues, Forrest Briggs (Project Manager, fall), Andrew Campbell, Bill Hewitt (Project Manager, spring)
NC4 provides their customers with breaking news alerts as incidents occur by monitoring and analyzing over 2,000 news sources. The goal of this project is to further automate NC4's online information filtering process, allowing analysts to maintain a sharper focus on news events and their potential impact on clients. The NC4 Clinic team has worked to develop new systems for collecting, analyzing, and distributing information to NC4's analysts.

Design and Implementation of a Object-Based Filesystem Simulator

Client
Sandia National Laboratories

Faculty Advisor
Professor Geoff Kuenning

Student Team
David Coyne, Selene Tan, Esteban Molina-Estolano, Mark Kegel (Project Manager)
As the need for greater computing power becomes more apparent in the simulation of complex systems, the computers themselves have reached the limits of their fundamental designs. Sandia National Laboratories has initiated an ambitious project with the goal of simulating supercomputers in order to find and eliminate the bottlenecks inherent in current architectures. The goal of this Clinic project is to create an I/O node simulator that will be eventually integrated with Sandia's larger project.

Computer Simulation of the GPS Ground Network

Client
The Boeing Company

Faculty Advisor
Professor Michael Erlinger

Student Team
Christopher Dahlberg, Christopher Erickson, Andrew Kim, Marshall Pierce (Project Manager)
To maintain accuracy, GPS satellites require regular corrections to their broadcast orbital parameters. An extensive network of ground antennas and control stations throughout the world periodically updates the satellites' orbital parameters, and transmits the updated information to each satellite. Currently, this process occurs at least once a day. Raising the update frequency would increase the accuracy of GPS, but would increase network traffic within the ground network with unknown consequences. To test the impact of such changes on network load, the Clinic team has developed a simulation framework using inexpensive PCs to represent the various nodes in the ground network.