Forming a CREW of Motivated Women in Computer Science and Engineering

lingering perception about computing is that of the loner scientist, toiling away in a cubicle with little human contact. The image has been popularized in books, television, and movies, where scientists and researchers are characterized as "nerds" with few interpersonal skills. However, in reality, most computer science research is done by groups working together to solve a problem. To emphasize the interaction in research activities, the Collaborative Research Environment for Women (CREW) program supports teams of undergraduate women in pursuing joint research projects. In CREW's first year, teams have tackled projects such as robot navigation and vision, parallel processor communication, Web navigation, and integrated circuit design.

"The stereotype of the lone scientist can be a deterrent to women," said Jan Cuny, co-chair of the Computing Research Association Committee on the Status of Women in Computing Research (CRA-W). "Women tend to be more motivated by interaction, and so may be rebuffed by the isolation assumed to be associated with research." The CREW initiative is sponsored by the CRA-W in cooperation with the Education, Outreach and Training Partnership for Advanced Computational Infrastructure (EOT-PACI), a joint program of NPACI and the National Computational Science Alliance.

BUILDING RESEARCH TEAMS

A SUCCESSFUL FIRST YEAR

CREW's goal in selecting projects is to identify where it might make a difference for the students and their institutions. In many cases, this was the only opportunity for this research to get done. It's a win-win situation for the advisors and their students.

BUILDING RESEARCH TEAMS

The CREW program is designed to provide collaborative research experiences for groups of two to three undergraduate women working with a faculty member. By increasing research opportunities and decreasing the isolation that may arise in independent research, CREW encourages women scientists and engineers to pursue similar work in graduate school. The first year of the CREW program, which wrapped up June 1, supported eight different team projects across the country. Project summaries, required of each project team, will be posted to the CREW Web site. Students are also encouraged to submit papers and present their work to other appropriate journals and conferences.

"The projects are student driven," said Sheila Castañeda, associate professor in the Computer Science Department at Clarke College in Dubuque, Iowa, and CREW program director. "We want to make sure that the women have a good research experience and learn to work together as a team. We're going to track the students over the next couple of years and see how this program affects their choices on their career paths."

The second year's awards will be announced June 30 for the 1999-2000 academic year. Eligible projects must be jointly submitted by the proposing students and a sponsoring faculty member (who need not be female). The research projects are conducted at the students' home institutions and must be directly related to computer science or computer engineering and be suitable for undergraduate research. The entire student research team must consist of two to three undergraduate women who are entering their junior or senior year and majoring in computing. Most teams have three members, although the first year had one team with four members.

"We had a substantial response for the first year of the program," Castañeda said. "More than 20 project proposals were submitted. I'm looking forward to some strong proposals this year from students and their advisors."

A SUCCESSFUL FIRST YEAR

The eight first-year projects of the CREW program covered "smart objects," robot motion planning, and expert systems.

Yuliya Dushkina and Shalva S. Landy, with advisor Lori L Scarlatos of the Department of Computer and Information Science at Brooklyn College, City University of New York, studied "Mathematical Puzzles with Smart Objects Interfaces." Smart objects are physical objects in the real world that serve as input devices to a computer system. The project used pieces of a physical puzzle--the Tangram--as smart objects that will help middle school students understand and appreciate mathematical and scientific concepts. The goal is to reach those students who are turned off by math and science and provide a way for them to learn collaboratively by manipulating real objects in the world around them.

Maralee LaBarge, Emily F. Greenfest, and Scott Klaum, with advisor Deepak Kumar of the Department of Mathematics and Computer Science at Bryn Mawr College, investigated "Visual Simulation Environments and Robot Drivers." In this project, they developed, on Apple Macintosh computers, a visual environment for robots that will allow empirical analyses of robot behavior. In addition, the environment provides software drivers to control the physical robots through a serial computer interface.

Marina Rabinovich, Julie Szymd, and Kimberly Robarge with advisor Joan Carletta of the Department of Electrical, Systems, and Computer Engineering and Science at Case Western Reserve University performed "An Experimental Evaluation of Methods for VLSI Partitioning." The team investigated cutting-edge algorithms for partitioning, implementing the algorithms in software and running them on various circuits to gauge the relative merits of the algorithms. They also investigated the mapping of other VLSI design problems onto the partitioning problem. Partitioning is an area of high interest in VLSI design automation research.

Rachel Heck, Sarah Luebke, Weichao Ma, and Hilary Mason, with advisor Samuel Rebelsky of the Department of Mathematics and Computer Science at Grinnell College, developed and tested "Trailblazing Tools for the World-Wide Web." The Web lacks many key navigation features, in particular, convenient mechanisms for constructing useful trails through Web pages. Web pages that contain links to a series of nodes do not provide appropriate mechanisms for visiting the nodes in sequence. Through this project, the team planned to remedy that situation and gain a better understanding of how people build and use trails.

Michele R. Cofield and Kamilah K. Walker, with advisor Albert Esterline of the Department of Computer Science at North Carolina A&T State University, studied "Motion Planning in a Dynamic Multi-agent Environment." The team looked at both low-level and high-level aspects of multi-agent motion planning. An accessibility-graph method was used for low-level motion planning in a dynamic environment, while state charts enhanced with annotations were used for high-level aspects. They also wrote software to interpret certain features of the enhanced state charts in the presence of the low-level motion planning.

Leha Blaney and Gina Castano, with advisor Lynn Stauffer of the Department of Computer Science at Sonoma State University, examined "Lazy Approximation in Expert Systems." Some relational programming languages use so-called "lazy" evaluation to delay evaluation of parameters and structured data elements until the values are needed in computation. Some of these same languages are used in expert systems, programs that emulate the reasoning ability and wisdom of a human expert. The project's goal was to explore lazy evaluation as a tool for faster and more economical implementations of expert systems.

Tze-Yun Lin and Erin Greening, with advisors Bonnie Melhart and Craig Morgenstern of the Department of Computer Science at Texas Christian University, studied "Processor Allocation to Independent Tasks in Low-Dimensional Mesh Systems." The team investigated the trade-offs between contiguous and noncontiguous processor allocation to tasks in 2-D mesh systems using the Procsimity simulator from the University of Oregon. The goal is to achieve a high level of processor utilization by relaxing locality constraints, while minimizing the inter-task contention for limited communication bandwidth.

Jennifer Davison and Kelly Hasler with advisor Karen Sutherland of the Department of Computer Science at the University of Wisconsin, La Crosse, looked at "Exploiting Composite Features in Robot Navigation." Robots, unlike humans, cannot easily pick out landmarks in outdoor environments and use those landmarks to navigate. And whereas humans identify natural features such as ridgelines and valleys to navigate by, robot navigation has tended to use single point features rather than these composite features as landmarks. This project focused on deciding what features in a typical outdoor environment can be used as landmarks and how those features can be combined in an automated navigation system.

"CREW's goal in selecting projects is to identify where it might make a difference for the students and their institutions," Castañeda said. "In many cases, this was the only opportunity for this research to get done. It's a win-win situation for the advisors and their students."--DH