Poster Presenter Bios | Photo Gallery | ACM SRC Winners
“My Mom is a Cell-Phone Illiterate” – The Influence of Students’ Perceptions of Parents on Pre-Adolescent Attitudes Towards ICT (SRC Competition)
Manpreet Kaur, Dept of Information Systems, UMBC
Abstract: Despite novel intervention programs, the proportion of women in the field of Information Technology (IT) has decreased. In order to reverse this trend it is important to understand the factors that lead young women into or away from IT careers. Exploring gender differences in technology use and attitudes is important towards understanding how career choices are influenced. Based on this motivation we explored gender differences in the perception and use of Information and Communication Technologies (ICTs) among US middle-school students. The study highlighted the importance of parental influence in forming ICT use habits and attitudes. In this poster we describe gender differences in parents’ use of and attitudes toward ICT as reflected in their children’s perceptions of them. We discuss parents’ differing patterns of encouragement for children’s ICT usage and examine the impact of these patterns on children’s actual ICT usage and perception of gender roles associated with ICT.
A Bit Error Rate Tester for Variable Encoding of High-speed Local Networks
Rosemary M. Francis, Cambridge University, Michael W. Dales, Intel Research, Madeleine Glick (Intel Research)
Abstract: Recent reductions in cost and increased capabilities of photonic components, along with higher demand, have opened up the field of optical networking to many more applications. Recent work has suggested that some 8b10b codes have disproportionately high bit error rates over low power optical connections. As part of on-going research into adaptive line encoding schemes we would like to investigate these effects further. We have been building an FPGA based hardware Bit Error Rate Tester, RoBERT. The principle function of RoBERT is to send diagnostic test packets over the network between a pair of nodes and collate error rate data on a per data symbol basis. We will be able to choose better encodings for our data and investigate the possibility of running systems at low power as well as develop systems for dynamic adaptive line encoding.
A Non-intrusive Data-driven Approach to Debugging Schema Mappings for Data Exchange (SRC Competition)
Laura Chiticariu, UC Santa Cruz
Abstract: Schema mappings are logical assertions that describe the correspondences between two database schemas. They are fundamental components in both data exchange and data integration. To date, a language for specifying (or programming) schema mappings exists. However, a tool for debugging schema mappings has not yet been developed. We present an approach for debugging schema mappings in which (test) data is used to drive the process of exploring, understanding and refining schema mappings. At the heart of our approach lies the notion of a “route”. Routes describe the relationship between source and target data at the level of schema mappings and have declarative semantics. We developed two polynomial time algorithms for computing one, and respectively, all routes for selected source and target data. The latter produces a complete, polynomial-size representation of the possibly exponential set of all routes. We present experimental results that demonstrate the feasibility of our route algorithms.
A Remote Server-Based Network Simulation System
Yan Gu, College of Computing, Georgia Institute of Technology
Abstract: With recent advances in parallel simulation, it is becoming feasible to emulate large-scale networks at a high level of fidelity as parallel simulation techniques can greatly expand the size of the network modeled. However, for many users, parallel computing facilities may not be readily available, and co-locating application code with a remote high performance computing facility may be cumbersome and inconvenient. A solution to this problem is to utilize a remote network emulation facility where the parallel network simulator (emulator) executes on a distant high performance computing facility. In this project we propose a new approach to remote emulation by augmenting the remote emulator with a local low fidelity model to provide QoS predictions rapidly on demand to the applications. Analogous to dead reckoning in distributed virtual environments, the server and clients inform each other of their status to keep their simulations synchronized while reducing the communication overhead incurred.
A Simple Parallel Approximation Algorithm for the Weighted Matching Problem (SRC Competition)
Alicia A Permell, Michigan Technological University, Fredrik Manne, University of Bergen, Phillip Merkey, Michigan Technological University
Abstract: Given a weighted graph, the weighted matching problem is to find a subset of edges of maximum weight, such that no two edges are incident to the same vertex. The fastest known exact algorithm for this problem runs in O(nm + n^2 log n), however for many real world applications this is too costly, and an approximated matching is sufficient. A c-approximation algorithm is one which always finds a weight of at least c times the optimal weight. Drake and Hougardy presented a linear time 2/3-epsilon approximation algorithm which is the best known serial algorithm. They also designed a parallel 1-epsilon approximation algorithm for the theoretical Parallel Random-Access Machine (PRAM) model, however it requires a large number of processors which is not as useful in practice. We present a shared memory parallel version of the Drake and Hougardy 2/3-epsilon algorithm, which is easy to implement and performs well in practice.
A Universal Approach for Error Characterization for Monte Carlo and Quasi Monte Carlo Sampling (SRC Competition)
Shiwani P Sambarey, Department of Computer Science, University of Illinois at Chicago, Urmila M Diwekar Vishwamitra Research Institute, Center for Uncertain Systems - Tools for Optimization and Management
Abstract: We propose a novel approach based on fractal geometry for characterizing error in Monte Carlo and Quasi Monte Carlo Sampling. The plots of relative error in prediction of mean, variance, fractiles for a probabilistic function against the number of samples are observed to be of fractal nature. A scaling relationship is established between the error and the sample size ( N ) as error ~ Nd ( ~ interpreted as “scales as”). By systematic experiments, we verified that “d” is indeed the dimension of the fractal represented by the curve. The Fractal Dimension model is universally applicable – independent of functional form, number of dimensions, probability distribution used and applies to Pseudo and Quasi Random number sequences alike. For Quasi Random Number sequences in higher dimensions, as the uniformity breaks down, the scaling relationship no longer holds true. In essence this model can also capture the breakdown of uniformity.
An Approach to Providing Remote Debugging Feedback without Revealing Sensitive Information
Emily Gibson, University of Delaware, Lori Pollock, University of Delaware
Abstract: Software users inevitably have to send an error report to the software developer so the developer can locate and fix bugs. Privacy-conscious users have been forced to either send sensitive data to the developer or send no error report at all—and potentially never have the bug fixed. Current work suggests either performing the debugging process at the user site or collecting enough random data such that no individual user can be identified. Until now, users have not had a systematic way of evaluating the amount of sensitive information revealed by an error report. In this poster we present a method of viewing the information revealed by path-based debugging feedback. Depending on the severity of the bug, the user may then decide whether the amount of information revealed is acceptable. Our approach enables the user to feel confident the developer will gain no sensitive information, while still providing valuable debugging feedback.
Auction-based Dynamic Task Allocation to Robots
Maitreyi Nanjanath, Dept. of CSE, Univ. of Minnesota, Maria Gini, Dept. of CSE, Univ. of Minnesota
Abstract: We present a method for dynamic allocation of tasks to robots based on single-item reverse auctions. While robots bid on one task at a time, all the tasks are auctioned before execution begins. On task completion, all the tasks not yet completed are re-bid in an attempt to improve the overall task completion time. We use a distributed algorithm for the auction mechanism, where each robot auctions its own tasks. This is useful in dynamic environments when robots may break-down or the environment may change unexpectedly. We present experimental results obtained in simulation using Player/Stage. We study scalability in the number of robots, number of tasks, and robustness to changing environments. The robots and tasks are randomly placed in an environment with multiple rooms. The robots use a Rapidly-Exploring Random Tree to estimate path costs for bidding and to plan the path to the task locations.
Building a Reference Resolution System Using Human Language Processing for Inspiration
Shana K. Watters, University of Minnesota
Abstract: For over 30 years, reference resolution, the process of determining what a noun phrase refers to, has been an important area of research. Accurate reference resolution is essential for tasks such as information extraction, text summarization, and language translation. Current systems for pronominal reference resolution have plateaued with accuracy rates around 80%. To produce accuracy rates comparable to humans, we propose to use human-like language processing. Our system combines subsystems that specialize in an aspect of reference resolution for the pronoun ‘it’. First, we create a set of prospective antecedents using guidelines established by our empirical study investigating the cognitive status of being ‘in focus’. Syntactic information, grammatical information, and surface features are then used to disqualify antecedents from this set. When two or more antecedents remain in the set, semantic information is used by the system to determine the best antecedent for the pronoun.
Cheating Shared Nothing: Using Adaptive Queries and an Object Oriented Approach to Improve the Performance of Shared-nothing Clusters
Marlena Compton, Equifax, Inc.
Abstract: The disadvantages of shared-nothing computer clusters are few, but can have an impact on the performance of queries distributed on these clusters. This project explores how a dynamic query which has the capacity to make decisions about the data being processed in real time can be optimized to lessen the effects of these disadvantages. The creation of a light, component based subsystem with algorithms for dynamic adaptive query optimization of communicating nodes in a parallel database that will, in effect, “cheat” the disadvantages of a shared-nothing architecture is proposed. This subsystem will allow participating nodes to communicate with each other as a consensus. This will allow the query to exit early in some cases and open the door to having nodes learn from each other creating further opportunity for query optimization.
Computer Security Awareness
Melissa Karolewski, Indiana University of PA, Alicia Coon, Indiana University of PA, Ometere Tute Ehinlaiye, Indiana University of PA, Sarah Raffensperger, Indiana University of PA, Rose Shumba, Indiana University of PA
Abstract: In an ever-changing technological landscape, running rampant with viruses, spy-ware, identify theft and the ping-of-death it is imperative to educate less knowledgeable computer users on the necessity of computer security. 2004 - 2005 CREU (Collaborative Research Experience for Undergraduates) research focused on the shocking statistics in computer security and educating computer users. A questionnaire acted as the medium for collection of the statistics. A mere 38% of users surveyed knew what encryption is. Furthermore just 26 % of those surveyed backed up their computer monthly. Firewall usage is even more dismal, only 55 % of those surveyed used a firewall on their computer. However 81% used an anti-virus program. Based on the enormous array of understanding of computer security knowledge among users, it was necessary to facilitate a tutorial on computer security awareness. This tutorial was presented to high school students at the 2005 Natural Sciences Science Festival and well received.
Conversation Lens: A System to Bridge Conversation and Item-Oriented Web Sites
Sara Drenner, University of Minnesota
Abstract: Item-oriented Web sites maintain repositories of information about things such as books, games, or products. Many of these Web sites offer discussion forums. However, these forums are often disconnected from the rich data available in the item repositories. We describe a system, movie linking, that bridges a movie recommendation Web site and a movie oriented discussion forum. Through automatic detection and an interactive component, the system recognizes references to movies in the forum and adds recommendation data to the forums and conversation threads to movie pages. An eight week observational study shows that the system was able to identify movie references with high precision and recall. Surveyed users found the interface fun and useful.
Coverage Metrics for Requirements-Based Testing (SRC Competition)
Ajitha Rajan, University of Minnesota
Abstract: In black-box testing, one is interested in creating a suite of tests from requirements without regard to the internal structure of the implementation. In current practice, the adequacy of black-box test suites is inferred by examining coverage on an executable artifact, either source code or a software model. There are several problems with using the executable artifacts to measure the adequacy of black-box tests. Foremost among them being it is an indirect measure, and adequacy cannot be determined until late in the development process. We propose the notion of defining structural coverage metrics directly on high-level formal software requirements and discuss how they can be adapted to measure finite test cases. These metrics provide objective, implementation-independent measures of how well a black-box test suite exercises a set of requirements. We focus on structural coverage criteria on requirements formalized as Linear Temporal Logic properties. These criteria can also be used to automatically generate a requirements-based test suite.
Decision-Theoretic Intelligent Assistance (SRC Competition)
Bowen Hui, University of Toronto, Craig Boutilier, University of Toronto
Abstract: Automated software customization is drawing increasing attention as a means to help users deal with the scope, complexity, potential intrusiveness, and ever-changing nature of modern software. Following recent probabilistic techniques in user modeling, we model our user with a dynamic Bayesian network (DBN) and propose to explicitly infer the “user’s type'’ in real time. We design the system to reason about the impact of its actions given the user’s current attitudes. To illustrate the benefits of this approach, we describe a DBN model for a text-editing help task. We show, through simulations, that user types can be inferred quickly, and that a myopic policy offers considerable benefit by adapting to both different types and changing attitudes. We then learn a more realistic user model with 45 users and create a topology of our proposed user types. Finally, we describe a pilot usability experiment and discuss encouraging results of our approach.
Detecting Emulated Environments
Tauhida Parveen, Florida Institute of Technology
Abstract: In today’s networked world, threats such as illegal reverse engineering to intellectual property and copyrighted materials is increasing. One way of circumventing protection mechanisms is to reverse engineer an application and monitor its control flow while it executes in an emulated environment. The resultant information can be used to help understand the application’s protection mechanisms. This research presents an approach to countering this threat by providing a means of automatically detecting the attempted reverse engineering. The approach can be used to determine whether or not a program is executing in a real or an emulated environment, thereby enabling the author to take the necessary steps to protect their valuable intellectual property. To illustrate the approach, proof-of-concept of the methodology is implemented. An evaluation of the results from experiments conducted to test the methodology suggests that the methodology is sound, and that the approach can successfully detect execution in an emulated environment.
Distributed Application Management Using Plush (SRC Competition)
Jeannie R Albrecht, UC San Diego, Christopher Tuttle, Google, Alex Snoeren, UC San Diego, Amin Vahdat, UC San Diego
Abstract: Although a number of solutions exist for subtasks of application deployment and monitoring in large-scale, distributed environments, few tools provide a unified framework for distributed application management. Many of the existing tools address the needs of a specific application or service that run in a specific environment, and are not extensible enough to be used for other applications. In this poster, we outline the design and implementation of Plush, a fully configurable application control infrastructure designed to meet the general requirements of several different classes of distributed applications. We also take a closer look at a few specific distributed applications to gain an understanding of how Plush provides support for each.
Distribution of Stream Queries to Utilize Location in a Cluster of Virtual Machines (SRC Competition)
Gauri S. Mawalankar, Indiana University Bloomington
Abstract: Distribution of stream queries to utilize location in a cluster of Virtual Machines, exploits the fundamental advantage of Virtual Machines of being able to deploy an executable dynamically. Virtual Machines appear to the user as physical machines and provide the user with all the functionalities of the physical machines. We propose a few hypotheses based on the literature survey and experimental evaluations about the manner in which stream queries can be distributed in a virtual machine setting. This will optimize performance after taking into account the overhead involved in deploying a VM on a physical machine. We will be using our results to enhance the Calder project which aims at integration of stream processing applications with Grid Computing through dynamic deployment of query execution engines and stream query distribution.
Driving Focused Innovation via Six Sigma
Rodi Tountcheva, Ohio State University
Dynamic Particle Systems for Efficient and Accurate Finite Element Visualization
Miriah Meyer, Scientific Computing and Imaging Institute
Abstract: Visualization has become an important component of the simulation pipeline, providing scientists and engineers a visual intuition of their models. However, simulations that make use of the finite element method for spatial subdivision present a challenge to conventional isosurface visualization techniques. Finite element isosurfaces are defined by basis functions in reference space, which give rise to a world space solution through a coordinate transformation, which does not necessarily have a closed-form inverse. We propose visualizing these isosurfaces with a dynamic particle system that samples an isosurface in reference space, avoiding the costly inverse mapping of positions from world space when evaluating the basis functions. The distribution of particles across the reference space isosurface is controlled by geometric information from the world space isosurface, such as the surface gradient and curvature. This provides compact, efficient, and accurate isosurface representations of these challenging data sets.
Electronic Encapsulating Notebook: Separating Control and Data Planes for a Secure, Usable System
Marianne L. Shaw, Univ. of Washington
Abstract: Widespread network attacks have highlighted the inadequacies of modern computer security. We revisit fundamental system security by developing software that emulates a real-world system — the laboratory notebook. Confidentiality, integrity, availability, accountability, and usability are basic security requirements. How are these properties provided in a laboratory notebook’s security environment? Are they provided in computer systems; if not, can they be? Can we realistically emulate a real-world security environment in a computer system? We are constructing a prototype laboratory notebook that meets our security requirements by separating the system’s control and data planes. We incorporate virtual machines for isolation, log-structured file systems for an immutable record of events, and a simple GUI environment for manipulating security. Physical access is a key aspect of overall system security.
Encouraging Girls in Technology: Effective K-12 Outreach by Corporations and Role Models
Linda Kekelis, Techbridge, Jeri Countryman Techbridge, Jennifer Wei, Techbridge, Shannon Madison, Google, Inc.
Abstract: Decisions that affect career choice are made long before young adults enter the workforce. As early as kindergarten, children’s career choices are influenced by sex-role stereotypes, often with girls perceiving roles such as computer scientists or engineers as boring, nerdy and too difficult. For many girls a career in technology is never considered an option. A number of factors can limit girls’ participation: little or no encouragement, limited enrichment opportunities, uninviting science and math courses, lack of confidence, pervasive stereotypes and little knowledge about these careers. The involvement of corporations and professionals in reaching out to girls can counteract these barriers and inspire them to consider careers in technology. Techbridge, a program designed to encourage girls in technology, and Google have partnered to present this poster session and share a resource guide for role models, companies and education programs to learn from the successes of our collaborative outreach efforts.
Extensible and Dynamic Data Structure Viewers in Java (SRC Competition)
Jhilmil Jain, Auburn University, James Cross II, Auburn University, Dean Hendrix, Auburn University
Abstract: Many visualization techniques though shown to be effective pedagogically are still not widely adopted. The reasons include: lack of suitable automatic generation of the visualizations; lack of integration among visualizations; and lack of integration with basic IDE support. To obtain the full benefit of visualizations when developing code, it is useful to automatically generate multiple synchronized views without leaving the IDE. The jGRASP IDE provides object viewers that automatically generate dynamic, state-based visualizations of objects and primitive variables in Java. Such a seamless integration of a lightweight IDE with a set of pedagogically effective software visualizations is unique and is currently unavailable in any other environment. Multiple instructors have used these viewers and have reported positive anecdotal evidence of their usefulness. The authors will be conducting formal, repeatable experiments to investigate the effect, if any, these viewers have on student learning, performance, and retention.
Fast Classification with Online Support Vector Machines (SRC Competition)
Seyda Ertekin, The Pennsylvania State University
Abstract: Technological advances in the past several decades have facilitated the generation and storage of digital information. As the amount of digital information increases, there arises the need for more effective tools to better find, filter and manage these resources. Therefore, developing fast and highly accurate algorithms for automatic classification of digital data has become an important part of the machine learning and knowledge discovery research. This poster presents a fast online Support Vector Machine (SVM) classifier algorithm that has an outstanding speed improvement over the classical (batch) SVMs and the other online SVM algorithms, while preserving the classification accuracy rates of the state-of-the-art SVM solvers. The speed improvement and the demand for less memory with the online learning setting enable the SVMs to be applicable to very large data sets.
Fault Seeding vs. Mutation Operators: An Empirical Comparison of Techniques for Web Applications (SRC Competition)
Stacey Ecott, Tufts University, Sara Sprenkle, University of Delaware, Lori Pollock, University of Delaware
Abstract: Organizations depend on web applications to perform daily tasks. Web applications are large, complex applications and must be thoroughly tested with the best available techniques. To thoroughly evaluate testing techniques for web applications, we must be able to compare their relative abilities to detect faults. Fault detection is important in creating effective, sustainable testing techniques and is usually performed by seeding known faults into a application and measuring the number of faults that a technique exposes. Currently, there are three main approaches to seeding faults: using known faults (e.g., from previous versions of the application), manually seeding faults, or using mutation operators, which automatically transform code to contain a fault. The best approach for seeding faults is an open research problem. In this poster, we propose mutation operators for web applications and perform a comparison study of using mutation operators against using manually seeded faults for a realistic web application.
Filecules: A New Granularity for Data Management in Science Grids (SRC Competition)
Shyamala Doraimani, University of South Florida, Adriana Iamnitchi, University of South Florida
Abstract: Grid computing has reached the stage where deployments are mature and many collaborations run in production mode. Such deployments offer the opportunity for revisiting and updating traditional beliefs related to workload models. This leads to the re-evaluation of traditional resource management techniques. This work analyzes usage patterns in a typical Grid community - a large-scale data-intensive scientific collaboration in high-energy physics. We focus mainly on data usage since data is the major resource for this class of applications. Our observations led us to propose a new abstraction for resource management in scientific data analysis applications – a filecule. A filecule is defined as a group of files that is always used together. We show that filecules exist and present their characteristics. The existence of filecules suggests a new granularity for data management. Incorporating this in design can significantly outperform the traditional solutions for data management based on single-file granularity.
Finding Grace Hopper on the Web
Rosie Jones, Yahoo! Research, Moira Regelson, Yahoo! Research, Wei Vivian Zhang, Yahoo! Research
Abstract: Grace Hopper has a presence on the world-wide web: as a biography about a prominent woman in computer science, as well as the web-site for the Grace Hopper Celebration of Women in Computing. In this poster we examine how web searchers find out about Grace Hopper. First we examine a sample of queries containing the term “grace hopper'’ and provide a break-down of the types of information sought. Then we show the types of queries which lead web searchers to gracehopper.org, as well as to other web-sites for women in computer science, and show the bipartite graph which arises when joining web-sites and query terms. Finally we show how query volume for “grace hopper'’ varies over time, and conjecture causes including school semesters, and the Grace Hopper conference itself.
Framework for a Tutorial Response Generator in an Intelligent Tutoring System
Rashida Z. Davis, University of Delaware
Abstract: ICICLE (Interactive Computer Identification and Correction of Language Errors) is an intelligent tutoring system to tutor deaf users on grammatical principles of written English. The system acts like an intelligent grammar checker. The focus of this research is to develop a tutorial response generation component that will plan expository text that is individualized to a user’s learning style and current knowledge level. The response generator is a multi-level text planner where each level reasons about a different aspect of the response: content (which errors to tutor in which order), method (which tutoring strategy for a particular error), form (instantiating the tutoring strategy according to user’s knowledge), history (relating to previous text), and manner (actually producing English phrases). The result is a cohesive, contextually-aware response to the user’s errors that provides intelligent and informative guidance. Presented is how ICICLE will accomplish generating tutorial explanations during a single response session.
Gender in Free/Libre/Open Source Software: A Social Science Perspective
Dawn Nafus, Intel Corp.
Abstract: This presentation will report findings from FLOSSPOLS, a recent European Commission-sponsored study of the role gender plays in Free/Libre/Open Source software (FLOSS) development. An earlier EC study showed that 1.5% of FLOSS community members were female at that time, compared with approximately 28% in proprietary software. Through an ethnographic study consisting of empirical surveys, participant observation and qualitative interviews, we set out to understand what was significant about FLOSS organizational and cultural forms that made it so acutely skewed towards men, and, together with members of FLOSS communities, make recommendations for actions that might encourage broader participation in FLOSS communities.
Holistic Usability Measure: Evaluating Speech Systems (SRC Competition)
Priyanka Gupta, Auburn University
ID-Based Clustering in Mobile Ad-hoc Networks (SRC Competition)
Foroohar Foroozan, Computer Science & Eng. ,York University, Uyen Trang Nguyen, Computer Science & Eng. ,York University
Abstract: An important quality measure of a clustering algorithm is the stability of the clustering structure, given the dynamic behavior of mobile ad-hoc networks (MANETs). The algorithm should minimize the number of node transitions from one cluster to another, allowing the resulting cluster structure to stay as stable as possible while the topology changes. In this paper, we propose an ID-based clustering algorithm. A stability factor is associated with the nodes to improve the stability of clusters produced. In our algorithm, nodes use periodic beacons as the only means of communications with their neighbors. The provided experimental results show that our algorithm outperforms the other mentioned clustering methods in terms of the stability of the created clusters.
Identifying Non Crystallographic Symmetry in Electron Density Maps: Artificial Intelligence Techniques in X-ray Crystallography (SRC Competition)
Reetal Pai, Texas A&M University, Reetal Pai, Texas A&M University, Thomas Ioerger, Texas A&M University, James Sacchettini, Texas A&M University
Abstract: X-ray crystallography is commonly used (80% of all known structures) to determine the 3D structure of a protein. An electron density map, created based on the diffraction of electrons in the protein, can sometimes be of low-medium quality. If a single crystal contains multiple copies of the same protein related by Non Crystallographic Symmetry (NCS), the redundant information contained in these multiple copies can be averaged to improve the quality of the electron density map. This process of NCS averaging requires the knowledge of NCS operators. In our preliminary approach, the similarities between NCS related regions of the electron density map were measured using a local density correlation metric. This approach, while accurate, was time-consuming. In order to improve the performance of the algorithm, rotation-invariant features were used to filter out dissimilar regions. The feature-based algorithm proved to be an accurate, robust and time-efficient alternative to NCS operator determination.
Improving the User Interface: Eye-Tracking in the Microsoft Office Usability Lab
Maria Fernandez Trevino, Microsoft, Amy Alberts, Microsoft
Abstract: Microsoft Usability piloted, acquired, and integrated Tobii Eye Tracking technology into our research labs with the goal to achieve a deeper understanding of customers’ interaction with our products. The pilot resulted in an internal training course and usability heuristics for eye tracking research at Microsoft. The Microsoft Office Usability team utilized this technology to iterate on interaction models for Beta 1 and Beta 2 versions of Office 2007. Data were surprising and informative to team members. Office program management considered eye tracking results, other usability data, and direct customer feedback to make strategic decisions about the direction of the Office 2007 user interface. Eye tracking research, in conjunction with other customer technology and business needs had direct impact on the final design of Office 2007. Eye tracking has proven to be a valuable source of customer data for improving the user interface at Microsoft, and for Office 2007 in particular.
Innovation and R&D in IT Services
Martha Lyons, Hewlett Packard
Abstract: Traditional views of product oriented innovation and R&D don’t neatly align within the IT services domain. The unique dimensions of services, the unique services innovation environment, and the unique role of R&D in services all come together to create new challenges and great opportunities for future research activities. In IT services, “meaningful” innovation occurs when we unlock the true potential of technology through customer solutions, when we solve customer problems, and when we optimize the way we conduct business and deliver our services.
Integrating Biological Data: Dealing with Incompleteness (SRC Competition)
Amanda Zeigler, Arizona State University
Abstract: CbioC is a collaborative system that lists interactions given in a biomedical article (like gene-disease or protein-protein). CBioC works in conjunction with the biomedical repository PubMed. Once a user selects an abstract, available facts are displayed. The facts are extracted from the abstract automatically, and are integrated with facts retrieved from other biological databases or entered by different users. My work on the project deals with extracting and integrating information from other databases that might be related to the extracted facts. The simplest integration involves getting facts for the same article found in public databases. A more complicated aspect is finding possible matches for the extracted biological entities to try to identify the correct match. For example, if “p53″ is extracted from an abstract, it might refer to TP53 (a human gene) or TRP53 (a mouse gene). This identification is important to allow integration of extracted data with other existing databases.
Interference-aware Concurrent Video Streaming Over Wireless Mesh Networks Using Multi-Source Multi-Path Diversity (SRC Competition)
Danjue Li, University of California at Davis, Chen-Nee Chuah, University of California at Davis, S. J. Ben Yoo, University of California at Davis
Abstract: With recent advances in wireless technologies, wireless mesh networks (WMN) have introduced a paradigm shift to the next generation wireless networks by enabling quick-and-easy extension of local area networks into a wide area. In this poster, we propose a new streaming architecture for supporting concurrent Video-on-Demand (VoD) services to various clients over WMNs. For the proposed architecture, we design an efficient path selection scheme to select multiple streaming paths simultaneously. If we allow intermediate WMN nodes along streaming paths to cache information locally, streaming path selection will also determine where the system is going to cache video contents, and as a result affect how the system evolves, i.e. disseminating information across the network. Nodes in different types of WMNs have different capacities, power-consumption constraints, and mobility. In this poster, we consider WMNs deployed in business or residential areas, where nodes (e.g., APs) have little mobility and are not power limited.
It Takes a Village (and Vision): The Role of Communities and Interoperability in Next Generation Networks
Ingrid Van den Hoogen, Sun Microsystems
Abstract: A growing concern around the explosion of digital data (conversion projects as well as new content) is how to make things systematically and efficiently work together. The business drive is generally for this to not happen. Non-interoperability and non-collaboration are still the norm. What can we do as women pioneers in building communities do to persuade and promote the opposite to happen? By providing examples from our own community building efforts, as original contributors to OpenSolaris, OpenOffice, OpenSPARC, NetBeans, JXTA, and Jini and as code contributors to Gnome, Apache, and Mozilla our objective is to challenge the audience to recognize how the core values behind Open Source projects and their communities – open access and inclusion, individual recognition, and freedoms to innovate – can and should be elevated to a more compelling vision – a “Global Village” where technical interoperability and free access to data spans across next generation networks.
Iterative Design in Practice
Erin Dean, Microsoft, Marina Dukhon, Microsoft
Abstract: The User Interface of Microsoft Office is taking on a complete overhaul in its next release. In order to reach the final design, the process of Iterative Design was applied throughout the release cycle. From the initial sketches of the design on paper, to several prototypes using PowerPoint, to various iterations on the design while in the development process, we have uncovered that in the business world of software development, the iterative design process is truly a crucial aspect of creating a User Experience that fulfills user needs. In practice, the process of iterative design often deviates from the seemingly straightforward steps outlined in textbooks. Iterations of design happen at every stage of the process and with varying levels of impact. This poster describes the stages involved while creating different elements of the User Interface and how these impacted designs.
Making Snapshots Perform Better (SRC Competition)
Bhavana Shah, University of British Columbia, Norman C. Hutchinson, University of British Columbia
Abstract: Data Snapshot technology is a standard feature of modern storage systems. Most such systems use copy-on-write techniques to manage snapshot data in order to optimize storage space requirements. Copy-on-write methods tend to write data out-of-place at a location which may be far away from the original location of the data on the disk. This phenomenon gradually leads to fragmentation of the on-disk snapshot data and degradation in the disk I/O performance. This work analyzes Logical Volume Manager’s snapshot technology and studies the effect of copy-on-write on the disk placement of the snapshot data. Based on these findings, we propose new disk space allocation and data placement techniques for snapshot volumes in order to reduce physical distance between related blocks and improve disk access performance. A prototype is implemented and its performance is compared with the original LVM2 implementation in order to measure the effectiveness of the proposed schemes.
Making the Invisible Visible: Creating Interactive Visualizations of Home Networking (SRC Competition)
Erika Shehan, Georgia Institute of Technology, W. Keith Edwards, Georgia Institute of Technology
Abstract: Building and maintaining a home network can be extremely vexing for the average householder. One of the key reasons for this difficulty is the invisibility of many salient features of a typical network, including its structure, state, and configuration of its devices. We believe that bringing visibility to the seemingly intangible, invisible aspects of home networking infrastructure can empower householders to better understand and manage their networks. To test this hypothesis, we have created a software tool that provides interactive visual representations of home networks, key aspects of which are strongly rooted in a prior ethnographic study of early adopters of complex home networks. This poster reports on our design process, our current interface design, evaluations with our target audience, and a discussion of the issues associated with visually presenting the invisible in home networks.
Making Waves — Computing Educators Oral History Project
Barbara Boucher Owens, Southwestern University, Vicki L. Almstrum, The University of Texas at Austin, Lecia J. Barker, University of Colorado
Abstract: This poster session describes an oral history project to establish and maintain a web-based collection of artifacts illustrating pathways through careers in computing education. The project’s first emphasis is women computing educators late in their careers. In addition to preserving the history of pioneers in computing education, this body of narratives will provide role models that may help attract and retain students, in particular women, to computing. The project vision is a searchable repository of archived information that includes digital audio interviews, transcripts of those interviews, photographs, vitas, and other relevant artifacts. Current work includes negotiations for a permanent home for the collection, fine-tuning the questions and protocol, formulating criteria for selection and inclusion of individuals in the collection, and developing strategies for funding the project. As of June 2006, we have held one workshop to train future interviewers and interviewed twelve educators from North America, Europe, Africa, and Australasia.
Meshing Genus-1 Point Clouds using Discrete One-Forms
Geetika Tewari, Harvard University
Abstract: We present an algorithm to mesh point clouds sampled from a closed manifold surface of genus 1. The method relies on a doubly-periodic global parameterization of the point cloud to the plane, so no segmentation of the point cloud is required. Based on some recent techniques for parameterizing higher genus meshes, when some mild conditions on the sampling density are satisfied, the algorithm generates a closed toroidal manifold which interpolates the input and is geometrically similar to the sampled surface.
Mobile Phone Keypad Design for Fast Chinese Text Entry by Phonetic Spelling (SRC Competition)
Yuan-Ting E. Huang, Department of Computer Science, University of British Columbia
Abstract: The trend of using text messaging on mobile phones has grown rapidly in the last decade. However, mapping the alphabet to twelve phone keys introduces challenging ambiguities for text entry. This challenge is exacerbated in Chinese by large phonetic alphabets and homophonic Chinese characters. In response, we developed an algorithm to generate keypad layouts that reduce ambiguity for Chinese text entry. The layouts may greatly facilitate text messaging for Chinese users. Furthermore, our algorithm should improve layout design for other ideographic languages (such as Korean and Japanese) and English T-9 style entry. Our solution begins with a metric based on Hick’s Law to model a layout’s ambiguity. We then use best-improvement, random-restart local search to generate high quality layouts. We have outlined a user study plan to compare our layouts against successful commercial layouts.
Model-based Safety Analysis (SRC Competition)
Anjali Joshi, University of Minnesota
Abstract: Safety engineers traditionally perform analysis, such as fault tree analysis, based on informal design models and various other documents such as requirements documents. Fault trees are one of the most common techniques used by safety engineers; yet different safety engineers often produce fault trees for the same system that differ in substantive ways. The final fault tree is often produced only through a rigorous process of review. We hypothesize that most of the review effort is focused on uncovering and resolving misunderstandings and missing information in the system design or the informal fault model. By redirecting some of this manual analysis and review effort to build formal models of the system and its fault model and performing analysis based on these models, an approach we call model-based safety analysis, we believe we can both reduce the effort involved (by introducing automated tools) and increase the quality of the safety analysis.
Close Multi-Angle View of a Mobile Target
Esra Kadioglu, University of Minnesota
Abstract: The problem of using vision for tracking and surveillance is a very important one. Mobility of visual sensors has further extended the variations of this problem. The Art Gallery theorem addresses the question of finding the minimum number of cameras sufficient to cover a room. This refers to the fixed-target/fixed-camera case. We extend this in multiple dimensions such that both the target and the pursuer are mobile; and address the problem of obtaining views of this target from different angles. Unlike classical tracking research where the pursuer follows the target from behind for localization purposes, in our work the pursuer follows the target and captures the images of it from various angles. This improves the identification of people in areas currently being monitored by fixed cameras which may not allow specific images (ex. face of the person). Multiple views also allow the construction of 3D images of the target.
Pathview – An Information Visualization Tool
Pooja Gupta, Advanced Visualization Laboratory, Indiana University, Michael J. Boyles, Advanced Visualization Laboratory, Indiana University, J. Scott Long, Department of Sociology, Indiana University, Shelley L. Nelson, Department of Sociology, Indiana University, James Robert Nelson, Indiana University
Abstract: This poster describes a software called PathView (PV) being developed by the Indiana University Advanced Visualization Laboratory in collaboration with Department of Sociology. PV is an information visualization tool that allows for interactive data mining to extract hidden patterns from complex data structures. A unique aspect of PV is that it is designed to analyze a sequence of outcomes for multiple cases. This provides a method of interactive visualization of the ways people move through a sequence of outcomes related to some aspect of their life. For example, analysis may be required of patients as they progress through sequences of health states during their lives. PV, hence, is best interpreted as a method of classification analysis where the extracted information is used to classify individuals into categories useful for later analysis.
Power Management in External Memory using PA-CDRAM (SRC Competition)
Nevine AbouGhazaleh, University of Pittsburgh
Abstract: Main memory has become one of the largest contributors to overall energy consumption and offers many opportunities for power/energy reduction. In this work, we explore the benefit of having on-memory cache with respect to its energy consumption. We optimize the cached-DRAM organization to reduce its energy consumption yet maintain its low memory average access time. We integrate a moderately sized cache within the chip boundary of a power-aware multi-banked memory. We call this organization power-aware cached DRAM (PA-CDRAM). We use the on-memory cache to reduce memory accesses to the DRAM-core; thus, increasing its idleness. In addition to improving performance, we use this caching to allow the DRAM-core to be quickly transitioned to a low power state for longer time periods. We show that PA-CDRAM is more energy efficient than using an extra level of off-chip cache of equal size. For the SPEC benchmarks, saving in energy-delay product is 28% on-average.
QoS Scheduling Algorithms for Grids
Claris Castillo, NC State University
Abstract: We consider the problem of providing QoS guarantees to Grid users through advance reservation of resources. Advance reservation mechanisms provide the ability to allocate resources to users based on agreed-upon QoS requirements and increase the predictability of a Grid system, yet incorporating such mechanisms into current Grid environments has proven to be a challenging task due to the resulting resource fragmentation. We use concepts from computational geometry to present a framework for tackling the resource fragmentation, and for formulating a suite of scheduling strategies. We also develop efficient implementations of the scheduling algorithms that scale to large grids. We conduct a comprehensive performance evaluation study using simulation, and we present numerical results to demonstrate that our strategies perform well across several metrics that reflects both user- and system specific goals. Our main contribution is a timely, practical, and efficient solution to the problem of scheduling resources in emerging on-demand computing environments.
Retaining Women in First Year CS Courses
Aurelia T. Williams, Norfolk State University, Sandra J. DeLoatch, Norfolk State University
Abstract: This poster describes interventions employed at Norfolk State University to retain women and minority students in computer science beyond their first year. Interventions seldom emphasize change in both faculty and students and most focus on changing specific behaviors, such as programming practices in IT. We present a research-based model that integrates practices that have been shown individually to be effective to guide efforts that we believe will foster successful change. We will briefly discuss each of these practices, theoretical and empirical support for them, how they may be tested and used by others, and results from the implementation of these practices at NSU.
Route Planner and Scheduler for Wheelchair Users (SRC Competition)
Suling Yang, University of British Columbia, Alan K. Mackworth, University of British Columbia
Abstract: The population of wheelchair users is very significant and increasing dramatically. Thus, finding accessible wheelchair routes is an important problem. However, the routes to the closest elevator in a new building, temporary road conditions and bus schedules may be unknown to wheelchair users. Therefore, we were motivated to create route planning software which can be installed on a small device to give wheelchair users route accessibility information while they are traveling. Our software also contains a simple scheduler which can synchronize with a route planner to provide more accurate commuting information for clients. After obtaining the destination from the scheduler, the route planner establishes some possible paths to the destination and displays the best one to the client. The poster will present an overview of the route planner and scheduler system through simulated real life scenarios.
Scalable Key Predistribution Scheme for Wireless Sensor Networks with Enhanced Resilience (SRC Competition)
Katerina Simonova, the University of Vermont, Alan Ling, the University of Vermont, Sean Wang, University of Vermont
Abstract: Limited resources and unavailability of pre-deployment information make authentication and maintaining confidentiality in Wireless Sensor Network (WSN) a challenging problem. WSN could be deployed in adverse environment, where message protection from eavesdropping is essential. However, well-developed asymmetric cryptography is unsuitable due to substantial resources requirements. Presently, Key Predistribution schemes that provide private keys to sensors prior to deployment is the prevalent technique used to secure communication in WSN. Early key predistribution schemes try to guarantee high connectivity between every pair of sensors in the network. These schemes perform well in small networks, but do not scale well when the size of the network increases. In order to solve the scalability problem, we note that, in wide-area WSN providing high connectivity between every pair of sensors is unnecessary as each sensor could reach only small fraction of sensors. We propose the framework that achieves substantial performance enhancement in wide-area deployment.
Strategy and Research to Inform the Design of Interactive Whiteboard Interfaces for Teachers
Jacqueline Bauer, Indiana University, Dennis Groth, Indiana University
Sundials in the Shade: A Study of Women’s persistence in the First Year of a Computer Science Program at a Selective University
Rita Powell, University of Pennsylvania
Abstract: Most attrition in computer science occurs early in the major and women drop out in disproportionately greater numbers than males. Taking an ethnographic approach to investigating women’s experiences and progress in the first year courses in the computer science major at the University of Pennsylvania, this study examined pre-college influences that led these women to the major and the nature of their experiences in and outside of class with faculty, peers, and academic support services. This study sought an understanding of the challenges these women faced with the goal of informing institutional practice about how to best support their persistence. The findings of this study suggest that women could benefit from multiple pathways into the major designed for students with varying degrees of prior experience with computer science. A computer science community characterized by more frequent interaction and collaboration with faculty and peers could positively impact women’s persistence. The response of Penn’s computer science dept. to these findings and subsequent change also will be discussed.
The Impact of Pair Programming on Women Students
Linda Werner, UCSC
Tracing Internet Packet Sources
Vijaya Gudimetla, Indiana University
Abstract: In Trace Back scheme some fields of the IP header are taken and are used to provide more information about the path the packets are traversing. We are using 16bit header, 16bit checksum, 13 bit offset and TTL fields for this scheme. In this scheme IP addresses are being stored along with other information that will give approximate position of the router that has marked the packet. When a packet starts there won’t be any changes made to it. But as the packets goes on one of the router will mark this packet with some probability and puts part of the IP address along with the changes in TTL are being marked. And if another router picks up the packet its IP address again with TTL is marked. So by the time the packet reaches the destination. The destination will have additional information about how the packet traversed.
TWiCE Undergraduate Experience in Research and Community Service
Bettina Bair, Ohio State University
Abstract: Studies suggest various approaches that could attract and retain more women in computing; such as social relevance, role models, research experience and networking. These principles, and applicable best practices, were incorporated into the design of TWiCE (The Women In Computer Engineering). TWiCE is an innovative and holistic program designed to reverse declining enrollment and improve the gender imbalance in computing. Undergraduate women are given opportunities to explore a variety of career possibilities in information technology. They use their developing technical skills to support community non-profits, local businesses and academic research. TWiCE also provides a supportive peer environment, as well as opportunities for students to act as role models and mentors to girls in their community. TWiCE has already shown a small, measurable impact on reducing gendered attrition in computer science at Ohio State. More significantly, 100% of participants in TWiCE have been retained in their CS-related major until graduation.
Ubiquitous Presenter: Student Access and Control in a Digital Lecturing Environment
Tamara Denning, UCSD, Erik Buchanan, UCSD, Michael Kelly, UCSD, David Lindquist, UCSD
Abstract: Ubiquitous Presenter (UP) empowers students in the classroom through • web-capture of instructor inked-notes for review outside of class • anonymous classroom interaction and support for active learning Instructors use a Tablet PC in lecture to support free-form inking on top of digitally prepared materials. This ink is captured and can be reviewed by students in a time sensitive manner. Additionally, students with web-enabled devices (laptops, cell phones, Tablets, etc.) can respond to instructor posed questions or complete active learning activities and immediate share their work with the instructor. The instructor can, within the lecturing framework, review and immediately project student-generated work for discussion and analysis. Additionally, student-generated work can be shared on the class notes web site for student review after class.
Using Perceptual Grouping for Object Selection
Hoda Dehmeshki, York University, Wolfgang Stuerzlinger, York University
Abstract: Modern graphical user interfaces support the direct manipulation of objects. The efficient selection of objects is an integral part of this user interface paradigm. For the selection of object groups, most systems implement only rectangle selection and shift-clicking or offer lasso selection (typically on pen-based systems). This poster presents a new approach to group selection that is based on the way human perception naturally groups objects, also known as the “Gestalt” phenomenon. Based on known results from perception research, we present an approach to group objects by the Gestalt principles of proximity, curve-linearity, and closure which is based on known results from perception research. We demonstrate the results with several examples. Then we introduce a new selection technique that exploits Gestalt grouping. Finally, we discuss the results of a first evaluation of the new technique, which shows that it can outperform traditional group selection methods.
