Marzilli

Profile: Matt Marzilli

Matt Marzilli (left) points out some of STORM's latest refinements to NMB's Dan Rainey.
Computer Science undergrad Matt Marzilli has spent much of his spare time for the last few months helping the National Mediation Board (NMB) decide how to use computers to facilitate labor-management dispute resolution. His efforts focus on developing STORM, a prototype tool developed as part of a National Science Foundation-funded project, aimed at studying how process definition and analysis technologies can help build trust between parties to labor-management disputes.

The NMB, established in 1934, is responsible for resolving all labor-management disputes in the transportation industry (airlines and railroads). While many disputes deal with collective bargaining agreements, NMB is also charged with minimizing work stoppages. As the industry and volume of its disputes continue to grow, the number of NMB mediators is not keeping pace, so NMB constantly seeks new ways to facilitate dispute resolution. The UMass Amherst team suggested that NMB start by studying their processes. "NMB is very interesting to those of us who study processes, because theirs are very rigorous and formalized, but incorporate quite a bit of flexibility," said Professor Lee Osterweil, the NSF project PI. "Many might think such processes are difficult to define formally and automate, but we believe our process definition language does indeed provide flexibility with rigor, and is thus a strong basis for defining and automating NMB's online dispute resolution systems."

In addition to Osterweil, the team includes Dan Rainey (NMB Director of the Office of Alternative Dispute Resolution Services), Lori Clarke (CS Professor), Norm Sondheimer (co-director of the UMass Amherst Electronic Enterprise Institute), and Ethan Katsh (Legal Studies Professor). This diverse team is attempting to gain insight into such fundamental issues as where computers can facilitate the dispute resolution process and where face-to-face interactions are better.

When the team realized the need to stimulate NMB's thinking by creating something "real", Marzilli stepped in to create a prototype that provides a visceral sense of the look and feel of a computer negotiation intermediary. Working with Clarke and Osterweil, Marzilli, a dual Computer Science and Physics major in his junior year, designed a prototype. "Our LASER lab always has undergraduates working both as programmers and on research. Matt is a great example of one undergrad whose research ability stands out. He provides a strategically important piece of the research project," noted Clarke.

With guidance from LASER software engineer Sandy Wise on how the prototype's user interface should look, Marzilli worked primarily on his own to build the new system. Sondheimer helped fine tune the prototype to get it ready to show to Rainey and the NMB. Marzilli not only provided the behind-the-scenes programming, but also continues to refine the prototype based on input from Rainey and the evaluations of prototype testers, including faculty and students from Legal Studies and the Isenberg School of Management -- even Marzilli family members that Matt recruited to test the prototype.

The STORM prototype (named for NMB's "brainstorming" process) enabled Rainey to see important advantages over current commercial products. For example, Rainey can now visualize how different process variations can use automation to support human negotiators in different ways, suggesting different variants for future research and refinements to STORM. Rainey has shown STORM at national conferences and has used it in teaching classes to both UMass Amherst students and to professional mediators. "It was one of the most exciting experiences for me to see students in front of forty computer screens all with the STORM interface during an actual class," said Marzilli.

Marzilli and CS graduate student Dan Gyllstrom went to Washington D.C. to train NMB mediators to use STORM, and came back with pages of comments and ideas for enhancements. This will help point the way to using the LASER process definition language and STORM to support the various negotiation process definitions needed by NMB. "I've never written a program before that really mattered. But, real people are depending on STORM," said Marzilli. "It is a great experience using what I know and applying it to the project."

While working on STORM this past summer, Marzilli also held a full-time physics research position. As shown by his 3.9 grade point average, Marzilli has no problems juggling his dual majors. "I chose physics, because, like computer science, I find it challenging," said Marzilli. "Physics is another disciplined and mathematical way of thinking, though computer science is my first love." For his educational accomplishments, Marzilli received a 2006 Barry M. Goldwater Scholarship awarded by the Goldwater Foundation.

"One can identify only a handful of young people who can add value to your research, and can also go off and do great things with the knowledge they gained. Occasionally, they can really make an important difference to the world. Matt is in a good position to be one of this small group," said Osterweil. Clarke concurred, "Matt has a nice future ahead of him. He has already shown an aptitude for research, and this is just a start for him."