Technical Paper Sessions

T1: Computational Math I - Thursday, 10:30am to 12:00pm
T2: Collaborative Environments - Thursday, 10:30am to 12:00pm
T3: Programming Languages - Thursday, 1:30pm to 2:15pm
T4: Computational Math II - Friday, 10:30am to 12:00pm
T5: Systems - Friday, 10:30am to 12:00pm
T6: Diversity - Friday, 1:30pm to 3:00pm
T7: Applications - Saturday, 10:30am to 12:00pm

T1: Computational Math I
Session Chair: Illya V. Hicks, Texas A&M University

On a Global Optimization Technique for Solving a Nonlinear Hyperboloid Least Squares Problem
Miguel Argaez, Brenda Bueno, Boguslaw Stec and Leticia Velazquez, The University of Texas at El Paso, leti@math.utep.edu

We present a numerical experimentation of the global optimization algorithm presented by Velazquez, Phillips, Tapia and Zhang applied to a nonlinear hyperboloid least squares problem. This problem arises when beta sheet residues from an allosteric enzyme are fitted onto a hyperboloid by using Newton type methods. The results show that the algorithm performs well on three test cases.

Variational Optimization for Call Center Staffing
Robert C. Hampshire and William A. Massey, Princeton University, wmassey@princeton.edu

We develop an algorithm to design a profit-optimal agent staffing and telephone line provisioning schedule for a call center. We model a call center as a multi-server queue with a time-varying arrival rate, additional waiting spaces and abandonment. We assume that there is a reward for every successful service completion, a penalty for every abandoned call, and a cost for the number of agents and telephone lines used. Using the limiting fluid model for this queuing system, we use the theory of optimal control to construct this efficient schedule.

A Trust-Region Interior-Point Method for Solving Nonlinear Programs
Cristina Villalobos, University of Texas-Pan American and Yin Zhang, Rice University, mcvilla@utpa.edu

Under mild conditions, the Jacobian associated with the Karush-Kuhn-Tucker (KKT) system of a (non-convex) nonlinear program is nonsingular near an isolated solution. However, this property may not hold away from such a solution. To enhance the robustness and efficiency of the primal-dual interior-point approach, we propose a method that at each iteration solves a trust-region, least-squares problem associated with the linearized perturbed KKT conditions. We also consider a strategy of enlarging the nonnegativity feasibility set. As a merit function, we use the Euclidean norm-square of the KKT conditions and provide a theoretical justification. We present some preliminary numerical results.

T2: Collaborative Environments
Session Chair: Jeffrey R. N. Forbes, Duke University

The Challenges of Ambient Collaboration
Hans-Peter Dommel, Santa Clara University, hpdommel@scu.edu

Collaborative capabilities are a hallmark of a new generation of networked applications. While traditional collaboration puts the computer in the foreground to help users interface through personal computing portholes, ambient collaboration reverses this paradigm by placing the machine in the background and enabling users to synergistically share a workspace with focus on mutual presence and tasks rather than tools. Although various ambient collaborative systems have been deployed in recent years, the field itself lacks a conceptual framework, in particular in contrast with legacy collaborative technologies. We introduce a simple systematics and roadmap for ambient collaboration to identify opportunities and challenges unique to this class of computing.

The Design, Implementation and Application of the GrewpEdit Tool
Kenroy Granville and Timothy J. Hickey, Brandeis University, kgg@cs.brandeis.edu

This paper examines the GrewpEdit tool which was developed as part of the Groupware Research in Education and the Workforce Project. GrewpEdit is a same-time, different-place groupware tool that was designed to support close collaboration in programming classes. Surpassing predecessors, it has an improved rendezvous mechanism that affords more users working on more documents and an intuitive interface that minimizes the users' necessary understanding. We describe the design and a general class session use of GrewpEdit and discuss our current and planned uses of this tool.

NusEye: Designing for Social Navigation in Syndicated Content
Azzari C. Jarrett and Brian M. Dennis, Northwestern University, a-caillier@cs.northwestern.edu

The design of NusEye, an online system for social navigation in syndicated content, is presented. NusEye is an information system that provides meta services for syndicated content and exposes the activity of users to each other, supporting social navigation. Social navigation is defined as provoking a user to move through an information space guided by the activity of others in that space.

NusEye augments syndicated content with social navigation by allowing users to apply ad-hoc keywords, or tags, to sources of Web syndicated information (webfeeds). Applied tags are visible to all users in the system and easily monitored. NusEye can identify the most popular tags and webfeeds. The entire community benefits from the aggregation and collective tagging efforts of each individual by being able to search tags and/or users in order to find new webfeed sources. Furthermore, NusEye offers special content analysis based on keyword tags. Users can apply various dynamically updated analyses on the content generated by groups of sources.

The major points of design include the selection of an interface in which social communities and networks can form from the use of tags, content analysis that is beneficial for syndicated content, and the presentation of analysis results.

MetaLab: Supporting Social Grounding and Group Task Management in CSCL Environments through Social Translucence
Deidra Morrison and Brian Dennis, Northwestern University, d-morrison2@northwestern.edu

Computer-supported collaborative learning (CSCL) environments have been developed to promote and enhance the ability of learners to participate in knowledge-building communities through synchronous and asynchronous communication and interaction. Contemporary and traditional cognitive and learning theory explains the role of social interaction between learners in collaborative environments to communicate their actions, share knowledge, and understand the collaborative task at hand. Although there has been work done to support social grounding in CSCL environments by incorporating functionality similar to that which could be employed in face-to-face discourse, it has not yielded conclusive results on collaborative learning. Also, there have been few systems that seek to incorporate features to support the mutual regulation of task within a collaborative group. The MetaLab project seeks to improve on current research on sociable CSCL systems. In this paper I describe the MetaLab project and how it seeks to address the problems of support for the NetLogo collaborative learning environment. I will also describe the current prototype of the MetaLab tool and the Wiki driven MetaLab Notebook environment, and how these tools leverage social awareness to address the problems of social interaction within these environments.

T3: Programming Languages
Session Chair: Andrea Lawrence, Spelman College

Media Centered Languages for New Computing Experiences
John Peterson, Yale University, peterson-john@cs.yale.edu

The goal of this effort is to make computer science accessible and attractive to a wider audience. Instead of focusing on the general purpose languages which are used to build large-scale applications, we propose to use a variety of simple, expressive languages with a strong mathematical foundation as a way of introducing core concepts of computer science while avoiding the complexity of fully featured programming languages. These languages hide an underlying computational process and focus exclusively on describing objects within specific domains.

Visual Programming for Novice Programmer Teachers
Cheryl D. Seals, Auburn University, sealscd@auburn.edu

We developed a construction toolkit for teachers to create visual educational simulation. This work is an effort to support science and inquiry-based education. With teachers as the subject-matter experts in the classroom, they are excellent candidates to develop educational software that meets their own pedagogical goals. We report an evaluation of 19 teachers creating educational simulation microworlds for physical science using this environment.

Teachers were trained in visual simulation programming, working from a minimalist instructional tutorial and an interaction guide. Additional programming support was provided in the form of reusable template objects which contain basic object functionality derived from analysis of our prior work with teachers. The current study helps us further refine this set of reusable components, facilitating ease of use for our novice educational simulation developers during simulation creation. The significance of this work is to provide greater support for teachers by providing content-specific, interactive, modifiable educational software, and thus to facilitate visual programming by this group of novice programmers.

T4: Computational Math II
Session Chair: Monica Martinez-Canales, Sandia National Laboratories

Computation of Nonclassical Shocks using a Spacetime Discontinuous Galerkin Method
Katarina Jegdic, University of Houston, kjegdic@math.uh.edu

We present numerical examples for two systems of one-dimensional conservation laws using a space-time discontinuous Galerkin (SDG) method with causal space-time triangulations and the piecewise constant Galerkin basis. The method is consistent with the weak formulation of conservation laws, and in the case of strictly hyperbolic systems, also with the Lax entropy condition. Convergence of the SDG method has been shown for strictly hyperbolic genuinely nonlinear Temple class systems using their special geometric structure. The initial value problems we consider here lead to nonclassical shocks. The first part of our study is for the Keyfitz-Kranzer system, which is strictly hyperbolic and genuinely nonlinear. We compute the SDG solutions approximating overcompressive and singular shocks, and note that our results are consistent with those obtained earlier using a finite difference technique. The second system we study is an approximation of a three-phase flow in the petroleum reservoirs. This system is of mixed type and has been numerically studied earlier using a continuation technique for ordinary differential equations. We compute the SDG approximation to its solution containing a transitional shock. Our conclusion is that even though the properties of the SDG method are so far proved only for the special class of hyperbolic systems, the numerical study herewith shows that the method can be used successfully in approximating solutions to more general conservation laws.

Empirical Sensitivity Analysis for Computational Procedures
Lorie M. Liebrock, New Mexico Institute of Mining and Technology, liebrock@cs.nmt.edu

Sensitivity analysis in computer science aims to improve stability in computer applications by considering uncertainty due to small perturbations in parameters. Mathematical and computational methods of sensitivity analysis are discussed. Advantages and disadvantages of both methods are addressed. A tool is developed to compute computational sensitivities. The tool was validated using three simple and well-understood problems. The tool was then applied to a dynamic power grid system and an agent-based criminal computation.

The power grid code, developed by Dr. Steve Schaffer at NMT, encapsulates a model for dynamic behavior of electric power systems. The inputs to the model are the dynamic states of the power system including resistances and torques. The outputs are states of the system, such as voltages and angles. The computational sensitivity analysis agrees with the computed (in the power grid code) mathematical sensitivity analysis results.

The criminal code population consists of criminals, gang members, victims and police. The model supports exploration of the movement of members of the population in response to interaction between agents. This code was found to be unstable - due essentially to having a fixed time step that was fairly large relative to the other constants in the program.

T5: Systems
Session Chair: Demetrios Kazakos, University of Idaho

Towards a Cross-Platform Microbenchmark Suite for Evaluating Hardware Performance Counter Data
Roberto Araiza, Maria Gabriela Aguilera, Thientam Pham and Patricia J. Teller, University of Texas-El Paso, raraiza@cs.utep.edu

As useful as performance counters are, the meaning of reported aggregate event counts is sometimes questionable. Questions arise due to unanticipated processor behavior, overhead associated with the interface, the granularity of the monitored code, hardware errors, and lack of standards w.r.t. event definitions. To explore these issues, we are conducting a sequence of studies using carefully crafted microbenchmarks that permit the accurate prediction of event counts and investigation of the differences between hardware-reported and predicted event counts. This paper presents the methodology employed, some of the microbenchmarks developed and some of the information uncovered to date. The information provided by this work allows application developers to better understand the data provided by hardware performance counters and better utilize it to tune application performance. A goal of this research is to develop a cross-platform microbenchmark suite that can be used by application developers for these purposes. Some of the microbenchmarks in this suite are discussed in the paper.

Distributed Calibration and Tracking with Low-Power Image Sensors
Teresa H. Ko and Nina M. Berry, Sandia National Laboratories, thko@sandia.gov

A system with a single camera is restricted to the field of view of the single camera. If we attempt to scale this approach to a large surveillance problem, the resulting multiple fields of view will not provide a concise fused view. A scalable solution to surveillance can be achieved through employing wireless sensor network technology where distributed sensors embedded with processors communicate wirelessly.

To address these constraints of unreliable communications, power-intensive processing and communications, and limited memory, DISCERN (DIStributed Camera Event Recognition Network) distributes reasoning about its organization and detected target information. This enables sensor nodes to collaborate intelligently with one another, forewarning neighboring nodes of possible targets, resolving location ambiguities of the sensor and target, and provide greater intelligence as additional target sample data are collected.

We addressed the limited power and processing speed by incorporating low-power image processing techniques to quickly reduce the large data acquired through images. Robustness was maintained through decision-based fusion for target detection and data-based fusion for target extraction and tracking across the sensor field. Distributed control is capable through our information-based neighbor lists facilitating the transformation of the target's information across sensor nodes as it traverses to the end user.

Integrated Radio Frequency Identification and Wireless Sensor Network Architecture for Automated Inventory Management and Tracking Applications
Mark L. McKelvin, Jr., University of California at Berkeley; Mitch Williams and Nina Berry, Sandia National Laboratory, mckelvin@eecs.berkeley.edu

Equipment management and tracking in an office environment typically involves a time-consuming process of identifying, managing, and tracking assets manually using a traditional bar code system. A number of problems are often encountered with this method of tracking and managing an array of devices, including an inaccurate account for equipment and human error that may result in an imprecise inventory database or relocation of equipment without notification. We propose a method that utilizes radio frequency identification technology (RFID) to automate the process of managing and tracking equipment throughout an office environment. In this approach, equipment is tagged with passive radio frequency devices that send identification information to a RFID reader for processing or uploading to a remote inventory database.

We provide a system architecture that applies RFID and Wireless Sensor Network technologies to automate inventory management and tracking of commercial assets. As a result of environment parameters and application requirements, a system architecture is proposed that automates inventory management and tracking in a large office environment. A prototype implementation under construction applies this system architecture, and provides insight as to areas for further work in data access, network communications, distributed real-time control and monitoring, and advancements in radio frequency technology.

Locating Where Faults Will Be
Thomas J. Ostrand, Elaine J. Weyuker and Robert M. Bell, AT&T Labs - Research, weyuker@research.att.com

We have developed a negative binomial regression model to help predict characteristics of new releases of a software system, based on information collected about prior releases and the new release under development. This model can be used to predict which files in the next release are likely to contain the largest numbers of faults or the highest fault densities in the next release, and allows a tester to focus their effort on the most fault-prone files. We also show how the prediction model can be used to allow a tester to select the files of a new release that will collectively contain any desired percentage of the faults. We include two case studies using large industrial software systems. The first study used 17 consecutive releases of a large inventory system, representing more than four years of field exposure. The second study used nine releases of a service provisioning system with two years of field experience. We show results for predictions made for 60%, 70%, 80%, 90%, 95%, and 99% of the faults and indicate the percentage of files required to satisfy each of these predictions. The case studies show that the model is able to make very accurate predictions.

T6: Diversity
Session Chair: Nina Berry, Sandia National Laboratories

Untapped Resources: Can Intellectual Diversity Promote Cultural Diversity in Technology?
Stephen Fancsali, University of Pittsburgh and Laura McGinnis, Pittsburgh Supercomputing Center, lfm@psc.edu

Many technical professionals and researchers who are successful in their careers did not start their academic experiences in the discipline they now practice. This implies that women and minorities who are studying in other fields may have skills that would make them successful in science and technology, but are studying in other majors, possibly due to the challenges they face in the culturally homogeneous environments common in technology. This presentation investigates the hypothesis that recruiting for intellectual diversity, i.e. from non-traditional majors, may promote cultural diversity by seeking women and minorities in the alternate fields of study they are pursuing. If science and technology are losing women and minorities to non-technical majors, is it feasible to pursue them in those other disciplines and provide opportunities for them to return to an environment that acknowledges their unique interests and talents? The cultural and intellectual diversity of staff and students working at the Pittsburgh Supercomputing Center provide a case study to examine these issues. As a highly visible research group at Carnegie Mellon University, PSC embraces the diversity initiatives promoted by CMU, while actively balancing the conflict of depth (in scientific specialties) versus breadth (of experience and perspective) that defines diversity.

IT Workforce Levels & Trends
Tyneka L. Harris, DePaul University, tharris@cs.depaul.edu

There is a need for minority participation and presence in America's Information Technology (IT) workforce. This research explores developing a Geographical Information System (GIS) application to examine the levels (i.e. employment, gender, race) and trends in the IT workforce by examining regions in America. Metropolitan Statistical Areas and Primary Metropolitan Statistical Areas (MSAs & PMSAs) are used to define regions. Locations for national chapters of this IT-interest organization, Black Data Processing Associates (BDPA), are used initially. BDPA has 54 chapters across the nation in major cities and universities. Data was collected for Black and Hispanic populations and compared to the total employment for a given occupation. These data are now embedded into an Internet-accessible Public Participation GIS site (IPPGIS), which includes a map of the U.S. with region indicators. When an indicator is selected, the respective region's IT workforce information is displayed. Numbers show that, across the nation, most IT occupations are male-dominated though some, such as data entry keyers, are dominated by females. Numerous inferences like this can be made. The GIS application will hopefully be used, not only for workforce levels, but for other areas of concern and interest, such as wage levels, for various racial groups.

Preliminary Analysis of Factors Affecting Women and African Americans in the Computing Sciences
Tiki Suarez, Florida A&M University, tiki.suarez@famu.edu, and Jamika Burge, Virginia Tech, jaburge@cs.vt.edu

This paper reports early results of research that addresses the social, cultural and financial factors that help to recruit and retain women and African Americans in the computing sciences. Computing sciences (computer science, computer information systems, computer engineering, and information technology) is a discipline with a history of disparate numbers of women and African American students. This research takes a systematic approach to understanding why these numbers are disproportionate, while seeking to provide sound solutions for increasing these numbers. By exploring relevant statistical reports, we have developed an initiative that details recommendations for improvement. These recommendations might be used to develop new policies and programs that increase the number of women and African Americans in the computing sciences. The advantages (and importance) of such an initiative are two-fold. Competition in global technological markets requires development of highly skilled personnel, preferably from a diverse society, regardless of gender or race. Secondly, there is opportunity to identify and establish pedagogical practices that can support gender and race learning opportunities. By preparing students in the classroom, they are positioned to develop their own strategies for success, and to even continue their studies in the field.

T7: Applications
Session Chair: Patricia Teller, University of Texas-El Paso

Knowledge Base Design for Environmental Research
Sevki Erdogan, Ted Shaneyfelt, Wade de Smith, Yani Ivanov, Andrew Honma and Cam Muir, University of Hawaii, Hilo, sevki@hawaii.edu

The project described in this paper is funded by the Experimental Program to Stimulate Competitive Research (EPSCoR) managed by the National Science Foundation. It has been implemented as a working prototype for investigating an integrated model for environmental research, which concentrates on the application of spatial and temporal analysis techniques to large-scale ecological datasets and includes satellite imagery and ground-based sensor data. Many of the sites that are of interest to environmental researchers in Hawaii are usually difficult the reach and the topology interferes with cellular or radio coverage. The backbone of this project is therefore a cost-effective telecommunication system that can transport data recorded by environmental sensors to a knowledge base. Experimentation with tools for modeling and visualization is also in the scope of the research and brings in researchers from many fields including computer science, hydrology, mathematics, computer engineering, biology, marine sciences and electronics. Thus, models that test scientific hypotheses relating to evolution, conservation, ecosystem function, and even the relationships between ecosystem and human health could be tested and verified.

Automatic Language Translation for User Interfaces
Manuel A. Perez-Quinones, Olga I. Padilla-Falto and Kathleen McDevitt, Virginia Tech, perez@cs.vt.edu

In user interface (UI) design, when designing interfaces for international audiences, special consideration is given to language. In this work, we discuss three different approaches used in translating website UI: a) automated translation, b) human translation done by the developer of the site, and c) appeal to a site's user base by allowing volunteers to translate the interface.

We informally evaluated the quality of translation using all three methods. A secondary motivation was to explore the naive position that a good translation software should have a transitive property, meaning that after translating a document from A to B, the quality of that translation can be assessed by translating it back to A.

The results of the evaluation were as expected: human translation and community translation produce higher quality translations than machine translations. However, the naive view that the quality of a translation can be assessed by translating it back to the original language is misleading. In our evaluation, the average on the original translations was 1, and the average on the re-translation was 6. This shows that pages translated back to their original language will score higher than the original translation.

Agent-Based Learning Approach for Power Trading
Gopalakrishna Reddy Tadiparthi and Toshiyuki Sueyoshi, New Mexico Tech, gtadiparthi@ieee.org

The U.S. wholesale power market comprises a large commodity market whose recent growth can be correlated to ongoing deregulation within the electric power industry. Most deregulation scenarios indicate a further separation of power production from transmission and retailing in the future due to increasing competition. Unfortunately, the mechanism of power trading is not clearly investigated at a level from which we can predict price changes in the U.S. wholesale power market. This study explores the business issue via a simulation system representing the wholesale market and different traders substituted by agents with learning capabilities. The intelligence in each agent is epitomized by the different bidding strategies and learning method followed. Using the new intelligence system, we investigate the bidding strategies of traders in the wholesale power market and examine how the price changes occur under different economic and engineering environments. We observe that entities with learning capabilities make more profit than those with no intelligence.

The John-e-Box: Fostering Innovation, Inclusion and Collaboration through Accessible Advanced Visualization
Eric Wernert, Mike Boyles, John N. Huffman, Jeff Rogers, John C. Huffman and Craig Stewart, Indiana University, ewernert@indiana.edu

Recent advances in commodity graphics and projection hardware have motivated many notable research projects and community discussions about the potential of these technologies to make advanced visualization more broadly accessible. However, the actual realization of this promise on a significant scale is challenging, requiring strong institutional commitment, expert technical support, and a broader visualization context. This paper describes an ongoing effort at Indiana University (IU) to develop a commodity-based, large-format, 3D stereo display system and to deploy a collection of such systems to a range of classrooms, laboratories, galleries, and learning environments throughout the IU system and the state of Indiana. To date, these systems have been used in over 30 projects by investigators in 15 departments across four different IU campuses. In addition, this technology has been used to reach well over 3,000 individuals through a series of coordinated outreach efforts. This initiative is also notable for fostering new interpersonal collaborations and inter-departmental cooperation, for enabling non-traditional applications in education and artistic expression, and for providing an interface to other advanced information technology efforts.