The team designed and prototyped a system with web-based interfaces for managing data from high-performance automated semiconductor test equipment based on distributed persistent data storage. The system enables users of a test service to query and process results from a wide range of geographic sites using interfaces that decouple test equipment specifics. The system is entirely based on the emergent Java Two Enterprise Edition (J2EE) framework and includes capabilities for fault tolerance and recovery.

The team created a tool for simultaneously visualizing several different reductions of multi-dimensional data sets. It also analyzed these sets using geometric techniques and clustering. To assess clustering fitness, the team implemented validity metrics that quantitatively compare solutions and algorithms in various ways. The tool was used to analyze data sets ranging from journal citations to popular music.

Teradyne's DFT (Design for Testability) Software Group is developing a software system that enables semiconductor manufacturers to exploit DFTtechnologies in order to reduce test costs, accelerate time to market, and enhance yield. However, to make the best use of DFT, semiconductor manufacturers must also be able to easily manage and interpret the automatically generated test results along with existing semiconductor design information and new information produced by DFT diagnosis tools. This Clinic team created a Java and XML/XSLT based system to meet this need.

Current space-based digital electronics lag their commercial counterparts by several generations. The physical hardening process required to protect these components from exoatmospheric radiation events is both costly and time-consuming. This project examined the feasibility of implementing a software-based alternative to this process. Specifically, the team examined the question of whether the increased processing power and memory capacity of commercially available components provide sufficient resources for software-implemented detection of radiation effects.

The purpose of the Magulandia Studio Clinic project was to create a 30-second public service announcement addressing road rage. The team used Maya (a 3D computer graphics animation tool used in movies such as Shrek) to build the necessary models and animate a short sequence for the PSA. The animation style utilizes exaggerated characters and scenery. The sequence uses personifications of cars and animals to make a comic point about the need to control one's temper on the road.

The LaserFiche project involved creating a software tool that automatically matches the image of a filled-in paper form to a corresponding blank template from a database. This tool permits a document-processing system to handle a large set of mixed forms without requiring a human to identify each one. Once an input form is identified, an anchoring step aligns and extracts the user's input in order to improve the performance of additional form processing.

In order to effectively deal with a high reporter turnover rate, the Green Media Toolshed team's goal was to develop a community-supported media-contact database that is economically viable for non-profit organizations by minimizing the need for a large research staff. The team implemented and assessed the effectiveness of user and data quality metrics, as well as automated checking tools, to coordinate the efforts of a network of volunteer data checkers.

The goal of the Kofax Clinic Project was to design and construct a system that can learn by example to provide candidate responses to customer service questions. The system learns to respond to incoming E-mail by first observing human responses. Then as new questions are asked to the customer service staff the system will suggest possible answers by drawing on responses to similar questions that the system has previously seen.

The team investigated content-based similarity relationships between musical performances. Similarity is a complex concept involving many judgments. The team focused their attention on a combination of rhythm, timbre, and apparent loudness. The team developed software that extracts these features from a recording and uses them to categorize new music, to make recommendations, and to generate playlists that arrange music in a sequence with smooth transitions between songs.

The team designed and implemented a proof-of-concept design for delivering weather data to aircraft pilots in-flight. Using a Pocket PC PDA as a hardware architecture and a custom client and server, they were able to deliver METAR (Meteorological Reports) and NEXRAD (NEXt-generation RADar) to pilots. The current implementation uses 802.11b wireless technology for the communication, but is ideally suited for satellite-based broadcast as a final product.

In the world of Intrusion Detection there is a need for a common message format and transport protocol so that different organizations can collaborate. This allows for the easy correlation, display, and long term storage of intrusion information. This year's project builds on the work of previous intrusion detection projects sponsored by The Aeorspace Corporation. It provides for messages to securely pass through firewalls using a newly specified BEEP (Blocks Extensible Exchange Protocol) profile called Tunnel.