Abstracts for the Third Annual NEUCS Symposium

1. Flash Drive Terrorism
by Mark Grinberg, a senior at Brandeis University

For the past several decades, the world has become increasingly interconnected through computer networks and the internet. While there are obvious benefits to these linkages, it is also true that terrorists and criminals have begun to use our interconnectivity for their nefarious purposes. Conversely, computer interconnectivity has been used to more positive ends recently; of particular note is the Stuxnet virus to delay the Iranian nuclear capability. My project discusses the increasing threats of our inter-connectivity and the challenge of securing our networks and lives against these threats. I have been studying the concept of cyber terrorism through my blog: http:// “Flash Drive Terrorism” refers to the idea that terrorism is now just as simple as having access to a flash drive or USB thumb drive. In our technological world, an individual with no combat training or equipment can, with proper knowledge of computers, inflict a substantial amount of harm on a population.

2. Visualization for Online Process Guidance
by Mario Barrenechea, a senior at UMass Amherst

Human-intensive processes, such as many medical procedures, are often prone to error because of interruptions, synchronization of multiple agents (including humans) and resources, and complex scheduling of tasks. We are investigating whether providing Process Guidance can reduce process errors. The Process Guidance we are considering involves displaying, in real-time and on an intuitive graphical user-interface, process context information such as the recently completed and pending tasks, the agents involved, and the resources required. To evaluate this approach, we are using a blood transfusion process modeled in the Little-JIL process definition language as a case study. However, our progress so far shows that designing such an interface is not straightforward; visualizing the blood transfusion specified in Little-JIL requires careful consideration of process semantics and abstraction. We show several UI mockups that meet these requirements and present issues that are associated with them.

3. The Esoma Project: At home physical therapy
by Ryan Orendorff, a senior at Tufts University/MIT

The Esoma project aims to increase the effectiveness of physical rehabilitation done by a patient at home after invasive cardiac surgeries. To achieve this goal, the Microsoft Kinect has been harnessed to find the positions of a patient's joints in space during their rehabilitation exercises. Using this data the Esoma system is able to determine how well the patient is performing certain exercises, and coupled with a pulse oximeter is able to
measure patient strength during recovery. This data can then be transmitted through electronic medical records to medical personnel (such as physical therapists) to tweak the exercises and to provide information about the patient's progress back to their supervising clinician. An added benefit of this system is alerts for abnormalities that can prompt a visit to the clinician or hospital, which might not otherwise occur due to the imperceptibility of the medical cues by the patient. The system currently features a modular, driver inspired architecture for obtaining and processing the data, with an additional layer added for visualization and patient feedback. Many of the important physiological cues are based on cotemporal features found in parameters such as arm extension and movement direction, which is segregated into temporal "packets". These are then analyzed both inside and between packets to determine time to perform exercise, velocity of movement, exersion and, when coupled with a pulse oximeter, cardiac ability. Esoma aims at providing a patient centric view of medicine by operating via telemedicine to increase exercise compliance and assist patient recovery.

4. Enhanced Campus Resources using the iPhone
by Consuelo Valdes, a senior at Wellesley College

Mobile technology has become a greater part of the average college studentʼs life with the popularity of the iPhone and iPod touch. We noticed that students still struggle, however, with the same short attention span problems from years ago, only this problem if extremely prevalent in museums among other campus resources. To remedy this, we have developed an iPhone application that invites students to make the most of campus resources. One that allows them to gain a deeper understanding of a work of art simply by entering a code. Another that allows students to learn more about specimens in a green house and view it through the seasons to get the whole picture of its life cycle. The application was created through a user-centered iterative design process that involved participatory design sessions with experts and weekly user studies over the course of a year. The applications show how a simple iPhone application template can serve multiple places with the appropriate information. We are currently further developing the botanical gardens application to adapt it to the iPad in communication with a large multi-touch tabletop interface for use in a horticulture course.

5. Ever wished you had superpowers?
by Consuelo Valdes, a senior at Wellesley College

Ever wished you had super powers? That's what TEI 2011 asked of its student in their super hero design competition. We took that concept and decided to explore physical output in embedded systems, i.e., movement. We explored simple day-to-day transformations in an automated system and found them less than functional. Which lead us to an outrageous costume and a great deal of engineering. We decided to explore transforming from one outfit to another as a superhero needs to in their day-to- day. The initial outfit would be simple while the transformation would lead to an amazing
outfit, perhaps creating a theatrical display during transformation; and "Morphess" was born. Morphess is a super villain that deceives you with her negligible appearance until you least expect it and she freezes you in awe with her transformations into a force to be reckoned with. The design won honorable mention in TEI 2011.

6. Incorporating ORCA functionality into Elasticfox
by Justin Duperre, a senior at Western New England College

There are several open source options available for private cloud computing, including Apache Virtual Computing Laboratory (VCL), Ubuntu Enterprise Cloud (UEC) and Global Environment for Network Innovations (GENI). However, each option has its own unique design and set of tools for cloud manipulation. Suppose a user wishes to utilize multiple different back-end clouds. In this case, it is desirable to have a “pluggable” and easy-to-use web interface that is compatible with these different back-end architectures. Open Resource Control Architecture (ORCA), designed by Duke University, provides a service-oriented resource control architecture that is designed to manage guests in a pool of networked hardware resources. ORCA can be used to manage the resources of a GENI cluster – hence, it is necessary that the “pluggable” web interface be compatible with ORCA in order to manipulate a GENI cluster.
VCL and UEC were investigated with respect to toolset and user interfaces. UEC was selected over VCL based on its well established back-end architecture and large support community. However, UEC provides only command line tools for many essential cloud manipulation functions. Elasticfox, an existing Firefox extension, provides an ideal user interface for managing UEC clouds. Elasticfox was then investigated, with the goal of modifying it to be compatible with ORCA while conserving its existing functionality with UEC. This project is currently in the early stages of incorporating ORCA functionality into Elasticfox. A successful implementation will support the essential manipulation functions for both types of clouds/frameworks using Elasticfox.

7. The feasibility and potential effectiveness of a hierarchical swarm topolog
by Nicholas Alunni , a senior at Worcester Polytechnic Institute

Distributed computing is becoming more and more prevalent in engineering today. Swarm robotics is simply an extension of that, not only dividing the computing power, but also the physical capabilities. This project served as a proof of concept investigation into the feasibility and potential effectiveness of a hierarchical swarm topology (HST), which better mimics the organization of many societal structures. This goal was approached by designing a three-tier robotic swarm as well as a specialized abstract coverage algorithm designed to map an unknown area as well as detect and extinguish simulated fires. Experiments were conducted by modifying various
parameters of an HST including the number of tiers and robots per tier. Results supported the original hypothesis that by adding robots, overall runtime and individual workload is reduced.

8. Hiatt Internship Database
by Murtaza Jafferji, a junior at Brandeis University

The Hiatt Career Center and the Office of the Dean of Arts and Sciences needed a centralized database to track undergraduate student internships. I created this database.

9. Leveraging online social network data and external data sources to predict personality
by Daniel Chapsky, at Hampshire College

People express their personalities through online social networks in a variety of ways, such as their relationships with their friends and their listed interests. In this work I present a method for automatically predicting an individualʼs personality by combining his Facebook profile information with external data sources using a machine learning method known as a Bayesian Network. The developed models use representations of peopleʼs connections to other people, places, cultures, and ideas, as expressed through Facebook. Due to the nature of Bayesian Networks, the semantics underlying the models are clear enough to not only predict personality, but also use knowledge of oneʼs personality to predict his behavioral attributes and actions. I will present some of the more interesting models of personality that my systems have produced thus far. These models demonstrate the potential of my methodology in two ways: First, they are able to explain up to 65% of all variation in a personality trait from a sample of 615 individuals. Second, they are able to clearly describe underlying relationships in the model through findings such as how to predict a manʼs agreeableness based on his age, hometown, number of Facebook wall posts, and his willingness to disclose his preference for music made by Lady Gaga.

10. Automatic localization of “Will.E”: An Autonomous Wheelchair
by Bei Li, a sophomore at Mount Holyoke College with Angela Ruohan Wang

Smart wheelchairs are being developed for patients unable to operate traditional power wheelchairs to achieve independence and safety. Their common features include automatic navigation and obstacle avoidance. Although the potential need for them is huge, very few smart wheelchairs have been integrated into the clinical environment.
Different from many previous researches, our approach to improve “Will.E” is based on evaluations from clinical user trials in cooperation with Spaulding Rehabilitation Hospital in Boston, MA. From these evaluations, we found the process of localizing “Will.E” to be one of the biggest concerns hindering its usability. To address this problem, we developed a Perl script which serves as a module to allow automatic localization in response to a button by loading stored coordinates.
The whole process of localization and map changing, which used to take up to 12 minutes and involves the operating of another computer, now can be completed within 3 seconds. This made it possible for the patient to start up the smart wheelchair without external help, vastly improving its usability.
“Will.E” was originally developed by ActivMedia Robotics of Amherst, New Hampshire (now Adept MobileRobots) in 2005 with Small Business Innovation Research (SBIR) grants from the National Institutes of Health.

11. Heapviz: Interactive Heap Visualization for Program Understand and Debugging
by Sean Kelley, a junior at Tufts University

Heapviz is a tool designed for all programmers, whether they are in industry, academia, or simply hobbyists, intended to facilitate program development when used alongside traditional tools. Heapviz has the unique ability to visualize the entirety of a program's memory at once by summarizing it. Summarization converts the data into a schematized form, revealing patterns in data structures without getting bogged down in details.
Heapviz allows users to push, pull, show, hide and generally rearrange the summarized data to visually debug programs or to quickly familiarize themselves with a new code base. It shows the user a conceptual context for the bug they are trying to find and supplies a rich set of interactions to navigate the data -- trivializing bug-finding in data structures by presenting the structures similarly to how the user would draw them manually. Heapviz also assists learning an unfamiliar code base by complementing the code with an image of what running the program does at the conceptual level -- revealing patterns in a program's data structures that were unexpected or too complex to understand from the source code alone.

12. The Hebrew Calendar: Computation through the Ages
by Max Goldstein, a freshman at Tufts University

The Hebrew calendar is lunisolar, meaning it accounts for both the sun and the moon, but has long been standardized to rely not on astronomic observation but rather computation. My C++ program determines the weekday on which the Hebrew year begins, the corresponding Julian and then Gregorian dates, and appropriate text output with calendrical formatting. We highlight the design choices (i.e. using global variables to permit code reuse) that led to approachable and elegant code. The algorithms were
partly taken from a program and documentation my father wrote in college, some 40 years ago. While his FORTRAN punchcards were among the first Hebrew calendar programs, today several similar date converters are available online, although source code is considerably scarcer.

13. Developing a Strategy Game for the iPad
by Foster Lockwood, a sophomore at Tufts University URL:

As an independent study, I've been working closely with a professor on a game development project, more specifically a real-time strategy game on the Apple iPad. Through this experience I've been exposed to the glaring (and the more subtle) differences between developing for a mobile, touch screen device vs. the more traditional PC with keyboard and mouse. Also, working with a more limited CPU and Graphics Processor has made me much more conscious about the different design choices and optimizations for slower, and single core processors. I will talk about my journey through the Objective-C jungle and how multiple languages (including C and C+ +) can be used to drastically improve performance. Also, I can show how the iOS performance tuning tools can be used to expose slow code. In addition, I'd love to talk about how the design of the game has changed and evolved from the beginning.

14. Using Integer Programming to Predict Interactions between mRNA and
by Alexandra Gendreau, a junior at Wellesley College

Small RNA (sRNA) is a type of RNA found in bacteria that does not code for a protein. These sRNAs act as regulators for protein production in the cell usually by preventing translation of a strand of mRNA through base pairing with some of the nucleotides in the mRNA. Regulation of mRNA by sRNA most commonly occurs under conditions involving environmental stressors or changes. Once these conditions are present, the sRNA can respond quickly and can lead to more rapid evolution of the bacteria allowing it to be more durable. Although the most comprehensive way to study sRNA:mRNA interactions is through wet-lab investigations, these investigations are time consuming and costly. Therefore it is useful to have efficient computational approaches that can guide more focused experimentation. RactIP is one such computational approach that uses integer programming to determine if an interaction exists between an sRNA and an mRNA and where that interaction occurs. It claims to be faster and more accurate than the other computational programs that already exist. RactIP has only been tested on a small number of interactions. Our goal is to study RactIPʼs ability to predict a broad range of experimentally confirmed interactions. Using a strain of E.coli, we will test RactIPʼs ability to predict sRNA:mRNA interactions. We also vary the inter versus intra molecular interaction parameter to see if we can improve RactIPʼs
initial results.

15. Automated Protein Classification Using Rigidity Analysis
by Courtney Schirf, a senior at Mount Holyoke

Proteins are one of the most important biological structures found in nature. Consequently, the ability to determine a protein's function quickly and accurately is of considerable importance to the scientific community. Many modern computational techniques seek to solve this problem by using available structural information to classify proteins in order to infer protein function.
This project attempts to improve upon current techniques by augmenting the current available data with rigidity analysis. This approach treats proteins as abstract geometric models in order to gain information about their relative flexibility. In particular, this project focuses on classification using protein architecture based on the CATH hierarchy. Standard machine learning techniques are used to do this classification.

16. A Feature Mapping Approach To CAD System Interoperability
by Felicia Cordeiro, a junior at Mount Holyoke

Mechanical engineers use Computer Aided Design (CAD) software to create sophisticated designs. Lack of interoperability between CAD systems leads to data loss, misinterpretation, and distortion, and permeates collaborative design. Existing techniques lose the design intent (semantics) of the user. To facilitate collaboration, our research group is working toward feature-based interoperability that captures design intent. We are working mainly with Pro/Engineer and SolidWorks, two popular CAD programs. Our method is to convert from each programʼs native language to a common XML representation. We hope this will enable collaborative design by providing accurate crossover of models between the two CAD systems and retaining full design intent. My role was to investigate the internal representations used in SolidWorks and to manipulate them via its API. This contributed to the research groupʼs overall goal of implementing and evaluating mechanisms, based on modern programming language concepts and techniques, for improving interoperability between CAD systems.

17. Distributed Radar Simulator
by Lucas Scotta, a senior at Worcester Polytechnic Institute with James Montgomery and Matthew Lyon

This project involved the design and prototype implementation of a distributed radar simulator for MIT Lincoln Laboratory. The simulator is designed to test radar control software developed at the Laboratory by mimicking radar hardware and simulating radar returns from targets in a virtual environment. The team worked closely with Lincoln Laboratory staff to ensure that the simulator design would be extensible to support different types of radar systems and scalable to thousands of targets. Finally, a distributed simulator was implemented in order to validate the project design.
The simulator design consists of models, which represent domain-specific objects, such as radar transmitters or airplanes, and engines, which coordinate event- driven communication between the models. The architecture consists of a single radar hardware simulator and many target simulators that have a set of targets distributed across them. The simulators can be executed on separate computers, and they send messages over a network to synchronize time and exchange events.
The team developed a prototype simulator implementation to validate the design. The prototype proved that the design could be distributed across multiple computers. It also ensured the correct operation of the model objects and the accuracy of all calculations used to determine radar returns. The team also developed an extensive testing suite and created a set of documentation to ease the adoption of the new simulator framework by the Laboratory. Lincoln Laboratory has expressed substantial interest in continuing the development of a new distributed radar simulator using the design and research conducted through this project.

18. Visualization of Process Guidance
by Mario Barrenechea, a senior at UMass Amherst with Lori Clarke

Human-intensive processes, such as many medical procedures, are often prone to error because of interruptions, synchronization of multiple agents (including humans) and resources, and complex scheduling of tasks. We are investigating whether providing Process Guidance can reduce process errors. The Process Guidance we are considering involves displaying, in real-time and on an intuitive graphical user-interface, process context information such as the recently completed and pending tasks, the agents involved, and the resources required. To evaluate this approach, we are using a blood transfusion process modeled in the Little-JIL process definition language as a case study. However, our progress so far shows that designing such an interface is not straightforward; visualizing the blood transfusion specified in Little-JIL requires careful consideration of process semantics and abstraction. We show several UI mockups that meet these requirements and present issues that are associated with them.

by Diana Esteves, a senior at UMass Boston WordPress plugin.

20. Rush Hour Traffic Optimization
by Jennifer Spinney, a senior at Worcester Polytechnic Institute with Matthew Runkle

How often do you drive during rush hour? And how often does your actual travel time during those peak hours match the time estimate given to you by sites like Google Maps or MapQuest? Do you ever wonder if you're missing some traffic shortcut known only to the locals? For our senior-year capstone project, we sought to create a driving directions website designed to provide accurate time estimates and optimal routes
based on real-world effective road speeds, which vary based on the time of day. We leverage driver frustration of slow traffic conditions in order to crowd-source speed data through our website, where users can report specific road speed observations. Our service is implemented in the Google App Engine cloud and takes advantage of their distributed computing ability to efficiently calculate optimal paths.

21. A Model of Hierarchical Spatial Reasoning
by Surabhi Gupta, a senior at Mount Holyoke College

The brain is incredibly efficient at storing and searching through large environments in our habitat. Edward Tolman was the first to propose the existence of cognitive maps in 1948. Although we have come a long way then, the fundamental questions remain unanswered. How is the environment represented in the brain? How are the locations acquired, composed, stored, recalled and decoded? Consider this: if each location was mapped onto one neuron, the spatial map would be extremely vulnerable to damage. It is necessary to model concepts or entities in a distributed fashion as a pattern of activity over a network of neurons. I propose a model of hierarchical spatial reasoning using Holographic Reduced Representations1, a method for encoding compositional structure using distributed representations. I began this project under the mentorship of Professor Dave Touretzky at Carnegie Mellon University. Path finding is based on hierarchical search across the different levels or scales in the environment. Auto-associative recall using Hopfield memory is used for retrieval of locations during the path-finding process. This model has many salient features such as providing automatic generalization, avoiding sequential search (thus increasing efficiency) and graceful degradation in larger environments.

22. Walt Disney/Pixar Animation: Controlling High-Resolution Fluid Simulations Using a Constraint-Based Method
by Visala Rani Satyam, a senior at Tufts University with Dana Botesteanu, Alexandros Fragkopoulos, and Anaid Guel-Mata

Animators use physics-based models to create simulations of effects with realistic appearances. For modeling fluids, like water and smoke, the Navier-Stokes equations are used. However, to satisfy artistic goals, animators want the simulations to behave in a specific way, so they must experiment with different parameters. They work in low resolution first because high-resolution simulations take longer to run. Artists would then like to preserve the same behavior when increasing the resolution. However, the behavior of fluids changes as we change resolution, making it difficult to predict the result. Our goal is, then, to develop methods that ensure the features of low-resolution simulations are retained in high-resolution. The animator can then predict, given a low- resolution version, what the high-resolution version will look like.
To do this, we insert an additional constraint into the already-existing numerical process for computationally solving the Navier-Stokes equation for incompressible fluids. We treat each simulation to be taking place on a grid, where fine grid means a high resolution and coarse grid a low resolution. The velocities of the fluid at every point are stored on the sides of each grid square. Our method applies a certain constraint as a guiding force on the fine grid: the force is applied inversely proportional to the difference between the fine and average coarse grid velocities. We compare the high- resolution result of our method (pictures and videos of smoke and waves) to the low- resolution original; it matches the original closely but with more detail.

23. Snoctopus
by Elmer Rodriguez, a senior at University of Massachusetts with Lior Ben-kiki, Royce Stubbs, Fabio Elia, Thomas Norden, and Evan Cordeiro

Snoctopus is a web based interface whose goal is to simplify the task of coordinating the many social networking utilities available to users. With SNOctopus, users can enter a content source such as a blog or anything that generates an XML- based feed (RSS, atom, etc.) and a set of target networks. The data from the content source is automatically propagated to other social networks such as Twitter and Facebook. The user can therefore concentrate exclusively on generating quality content, not resubmitting work to multiple web utilities.

24. Cascading Crater Detection with Active Learning
by Henry Z Lo, a senior at UMass Boston with William I. Miller, Tomasz F. Stepinski, Yang Mu, and Wei Ding.

Surveying impact craters is the only way to detect the age of geological formations on a remote planet. High resolution images reveal countless numbers of these craters on planetary bodies; yet the large number of craters in such images requires an effective method for automated detection.
In this study, we adapt a cascading AdaBoost classifier to the problem of crater detection on two high resolution images of regions on Mars. One of these images was the training set for our classifier, while the other was used as a test set. First, we decompose the image into a set of sub-windows of varying sizes. Then, we train our classifier on the test image sub-windows, which are labeled as craters or non-craters. Each sub-window image is represented by thousands of Haar-like features, and it is on these features that the classifier makes its judgment.
In order for our model to be effective across images, however, it must be able to correctly label a variety of sub-windows, some of which may not be represented in the training set image. Our solution is to map sub-windows from the training set and the test set onto a self-organizing map (SOM). This allows us to identify those sub-windows which have different features from the training set, which an analyst will then label and
place into the training set. With this technique, we hope to design a system which can detect craters with high accuracy and minimal human intervention.

25. OpenMRS Draftforms Module
by Hannah Deutsch, a senior at Mount Holyoke College with Yaa Asantewaa, Appiah Korang, and Gabby Sneider

OpenMRS is an open source medical records system platform for developing countries, currently being used by Partners in Health. It allows clinics and hospitals to build their own medical records system with their own forms. Our goal is to develop a module for the OpenMRS framework that will simplify the process of converting paper forms to electronic forms. Currently paper forms must be converted by hand into HTML which requires knowledge of HTML, OpenMRS's own HTML tags, and CSS. The process to convert a form from paper to its digital equivalent could take months of work.
The DraftForms module for OpenMRS is designed to automate part of this process. The module allows somebody to use a graphical user interface to select various components of a form - such as text responses, images, etc - and have the module automatically generate HTML. From there, all one has to do to complete that form would be to write the CSS.
So far we have implemented the database and most of the user interface. We are currently working on finishing the user interface. By the end of the semester we hope to have completed enough of the module to create a draft of a form with it. This would cut down on the time needed to create a form, and on the expertise needed since we are replacing hand coding with a graphical user interface. Ultimately, this module could enable someone with no experience using HTML to create an electronic form.

26. An HCI Approach to Raising Green Computing Awareness
by Gabby Snyder, a sophomore at Mount Holyoke College with Cleo Schneider URL:

The need for renewable energy as well as a concerted effort to reduce our energy consumption is apparent; however, the energy consumption of computing is greatly overlooked. Laptops are often left plugged into the wall even after being fully charged and unused programs or windows are often kept open, all with a disregard for how much energy these resources might be consuming. Green computing is an effort to study ways of improving the energy efficiency of computers, often focused on consumersʼ use of computers. This project uses an HCI design approach to develop an interactive quiz and game that inspire users to think about their computing carbon footprint. Both the quiz and the game were programmed in Flash utilizing the iHart API developed by Cleo Schneider, and were created in collaboration with Maria Kazandjieva, a Mount Holyoke alumna currently pursuing a PhD at Stanford University.

27. E-trawler
by Christopher Griffith, a junior at UMass Boston

with Brendan Davis, John McGourty, Tu Tran, Felix Green, and Ronnie Sekamwa
E-trawler is, at its core, a specialized and automated search engine. Unlike a typical search engine whose goal is to return links to web pages matching your provided key words (along with a brief excerpt/description), E-trawler's goal is to provide statistical analysis based on the keywords it has been provided. This entails determining, among other things, frequency the word occurs on a given page and associated words that regularly show up alongside one of the key words. These associations would then be used to improve future search iterations.

28. Sig Fig CalcPeer and Authority Influence in Social Networks
by Kevin Y. Zhao, a freshman at Brandeis University with Ross Newman

Sig Fig Calc is a project that calculates significant figures in science. Significant figures are important when making calculations in chemistry. When one measurement is more or less accurate than another and we use both figures in a formula to calculate a value, the new value will only be as accurate as the less accurate of the two values. When someone measures 1.5 meters and 1.51 meters and then multiplies the two to get 2.265 meters squared, the .005 meters squared in this case can hardly be trusted to be accurate. A distance measured to be 1.5 meters means that the measurement is more or less .1 meters within 1.5, so the real value may lie about .151 off of the calculated value.
The internal implementation deviates from the classical postfix stack method of parsing mathematical expressions. Instead, it takes advantage of some recursive properties within a mathematical expression to parse the expression. Every parentheses must contain an expression, so we parse and calculate that value first before proceeding in the calculation of the overall expression. On every return call of the expression parser, there is a guarantee that the value in the parentheses will contain a numerical value. By the time the call stack comes back to the original position, the entire expression has been parsed.
In order to make the application useful for students, it has been released for Android operating systems.

29. Auditory Scanning to Evaluate Cognitive Function
by Allison DeJordy, a junior at Mount Holyoke College with Melissa Frechette URL:

Cerebral palsy is a condition that causes varying degrees of motor impairment. The most severely affected of cerebral palsy patients cannot speak, write, or use a standard computer mouse. These patients communicate using a large button, called a switch, connected to a computer. The computer browses through various options
automatically, and the patient presses the switch when the option they want is selected. The switch can also be used to determine how much motor control a patient does have, by seeing how quickly they respond when told to select a specific option.
A number of programs are already available for use with the switch. However, for all of these programs, the patient must be looking at the screen. This can be difficult for patients who cannot support their heads for long periods of time, or who have visual impairment. David Kontak of Crotched Mountain Rehabilitation Center in Greenfield, NH contacted us to develop a switch program that has an auditory component that allows use without any visual prompting at all.
Our program will function similarly to the existing programs for the switch. However, as the computer selects each option, it will also play a sound corresponding to that option. This will be useful to occupational therapists to distinguish between a patient who has poor motor control or cognitive function and a patient who cannot see the screen. In the future this could be adapted to help cerebral palsy patients to communicate their wants and needs.

30. Crater Seeker
by Joseph Paul Cohen, a senior at UMass Boston URL:

As part of a NASA outreach project, a crater seeker game was made to showcase work done by UMass Boston, Ohio State University, and NASA. A Mars Day event was planned in collaboration with the Museum of Science Boston to showcase this work as well as this game. The game is designed to simulate NASA and JPLʼs experience when controlling rovers on Mars. The game allows the user to drive a 3D rover from overhead and point of view in and out of craters. The user can use tools such as slope maps and point of view orthophotos to help navigate in and out of the craters gathering points without tipping more then 25 degrees.    This game uses real MER-B mission images and elevation data provided from the Ohio State University.

31. Developing for the BlackBerry Platform
by Antonio Cancio, a sophomore at Brandeis University URL:

This project follows the development process followed by the presenter to develop an application idea for the BlackBerry platform. The presenter is a long-time user of BlackBerry smartphones and found a need for a fun application that was missing from the limited selection on the BlackBerry App World. This project will showcase how you can take an idea use the BlackBerry developer program and the incentives it offers to make it a reality. This includes the use of BlackBerry development tools, application development workflow, the use of the Research in Motion API, the App World submission and approval process, the process of updating applications to newer versions, and the use of download and use statistics. In addition, the project will touch on how to advertise such a small application. Most importantly, this project will touch
base on how a second-year student in computer science can use what they have learned in their introductory courses to develop an idea.

32. Formatt: Correcting Protein Multiple Structural Alignments by Sequence Peeking
by Shilpa Nadimpalli, a senior at Tufts University with Noah Daniels

We present Formatt, a multiple structure alignment program based on the Matt purely geometric multiple structural alignment program, that also takes into account sequence similarity when constructing alignments. We show that Formatt is superior to Matt in alignment quality based on objective measures (most notably Staccato sequence and structure scores) while preserving the same advantages in core length and RMSD that Matt, as a flexible structure aligner, has as compared to other multiple structure alignment programs on popular benchmark datasets. Applications include producing better training data for threading methods.

33. Open Source Data Visualization for the Masses
by Krithika Manohar, a senior at UMass Lowell with Kyle Monico, John Fallon, Patrick Stickney

Weave is a web-based data analysis and visualization framework. Although many visualization systems exist, these systems typically specialize and provide limited user interaction. Weave addresses these issues by providing unparalleled user interactivity. Some of Weaveʼs advanced features are session history, collaboration, on- the-fly map projections and geographic layering, streamed data and an interface to the R statistics package. Weave was designed to allow access to existing datasets or simple upload of local data. Weave is developed using the Adobe Flex framework and utilizes the Adobe Flash Player to achieve cross-platform support. The source code is scheduled to be released under an open-source license later this month.

34. Repurposing the Nintendo Gameboy for Computational Music
by Eric Fairbanks, a senior at UMass Lowell

In its heyday, the Gameboy was one of the most successful handheld gaming consoles. Its wild popularity and subsequent obsolescence left many models homeless and purposeless. A robust, simple and inexpensive computer, the Gameboy is a prime candidate for mods, hacks and reverse-engineering.
As a personal exploration into computer architecture, electronics and embedded systems, I assembled a home-made software development system for the original Nintendo Gameboy. It consists of a modified Gameboy game cartridge, a custom hardware design based on the Arduino for reading and writing binaries to the cartridge, and existing software development toolchains.
As an old game console, the Gameboy has seen a resurgence in popularity with
the recent rise of chip music and retro-gaming culture. Music artists are interested in capturing the nostalgia of old game music, and have written software for the handheld that allows users to manipulate its dedicated sound hardware. My poster will present the Gameboy computer architecture design, including CPU, memory, video and audio architecture, the design of my cartridge programming tool, and discussion of computer music projects built on this platform.

35. Moleint: Reducing Cognitive Workload through Adaptive Interaction
by Connor Gramazio, a junior at Tufts University with Megan Strait, Jisoo Park, Sara Su, and Lenore Cowen

We introduce moleint, an integrated, adaptive interface for supporting inquiry- based proteomics learning and research. In this application domain, we address the challenges of exploratory protein visualizations across linked heterogeneous data. Biologists studying a protein must keep track of three types of associated information – sequence, structure, and homology. We demonstrate that an implementation of integrated multiple views is an effective method for offloading this data from working memory. Furthermore, we give a proof-of-concept demonstration of two adaptive mechanisms for countering the cognitive demand of the increase in information presented to the user.

36. VAST 2011
by John Fallon, a sophomore at UMass Lowell with Patrick Stickney, Shawn Konecni, Georges Grinstein, Baochen Sun, Yen-Fu Luo, and Loura Costello

Visual Analytics Science and Technology (VAST) is a visual analytics competition. This competition is intended to push the advancement of the field of visual analytics. The competition encompasses many domains such as intelligence analysis, business intelligence, law enforcement, climate change, etc. The challenge requires the creation of realistic synthetic data sets. The data sets in this yearʼs challenge involve mostly text data, along with some map images. The challenge pushes cutting edge research and allows for the testing of different analytics tools in a realistic scenario. There are three mini-challenges and one grand challenge each year for the VAST challenge. This yearʼs focus is on a fictional city, Vastopolis, which is experiencing an epidemic. It is up to teams to answer questions such as the epidemicʼs cause and details as to its containment.

37. Definitious - Find Meaning
by Yale Spector, a senior at Brandeis University with Murtaza Jafferji

Definitious is an online dictionary where you decide how words are defined. On Definitious, you can express your knowledge, humor and creativity by writing your own definitions for words. Or, you can search and browse Definitious to explore worlds of meaning for both education and recreation. Depending on what you're in the mood for, Definitious can provide you with funny, poetic or helpful definitions, so whether you're writing a paper, or you're looking to laugh, Definitious has you covered. Also, Definitious can be used cross-lingually. You can submit a word in any language and define it in any language. Definitious was developed in Ruby on Rails over a 2-month period by Murtaza Jafferji and Yale Spector for the Justice Brandeis Semester, a summer program in web & mobile development at Brandeis University.