Clinic Projects

Please click on a link below to view the Harvey Mudd College Computer Science Clinic projects for the corresponding time period.

Clinic Projects for 2010-2011

A Virtual Telescope for Global Monitoring of Earth Surface Changes

Client
Caltech Keck Institute of Space Sciences

Faculty Advisor
Professor Michael Erlinger

Student Team
Sonja Bohr (Project Manager), Emma Carlson, Singer Ma, Julia Matsieva
COSI-Corr (Co-registration of Optically Sensed Images and Correlation) is a software package developed at Caltech that estimates ground displacements using multi-temporal satellite images. Among other uses, correlating such images enables geologists to locate earthquake fault ruptures soon after an event. Unfortunately, analysis of a few pairs of images may take on the order of several hours. To improve this response time, our project is to propose a framework that allows the migration of COSI-Corr to a multitude of distributed computing environments.

ODEToolkit

Client
Community of Ordinary Differential Equations Educators (CODEE)

Faculty Advisor
Professor Talithia Williams, Professor Christopher Stone

Student Team
Beky Kotcon, Samantha Mesuro (Project Manager), Daniel Rozenfeld, Anak Yodpinyanee
The Consortium of Ordinary Differential Equations Educators (CODEE), an organization dedicated to improving the teaching and learning of ordinary differential equations, created an educational, differential equation-solving software called ODEToolkit to pursue its goals. During this year, we have implemented an additional numerical solver capable of handling stiff differential equations, made necessary changes to the software's architecture, fixed bugs in the software, and conducted user testing regarding bugs and the usability of the software.

Indexing Improvements for the Hive: Database System

Client
Facebook

Faculty Advisor
Professor Robert Keller

Student Team
Skye Berghel, Jeffrey Lym, Russell Melick (Project Manager-Spring), Marquis Wang (Project Manager-Fall)
Apache Hive is an open source distributed database/data warehouse system built on the Map/Reduce framework of Hadoop. Designed to handle massive datasets, it is currently heavily used by Facebook. Facebook desires improvements to the indexing framework within Hive, both in handling different types of data and in ease of index use. The team has developed a new index type based on bitmaps, which offers improvements for self-similar datasets. They have also augmented the existing framework to enable automatic creation and use indexes.

Common Data Services to Support Continental-Scale Ecology

Client
National Ecological Observatory Network, Inc. (NEON)

Faculty Advisor
Professor Zachary Dodds

Student Team
Jason Garrett-Glaser, Keith Ingram (Project Manager), Alejandro Lopez-Lago, HamsterBob Stewart
The HMC NEON Clinic team has designed, prototyped and delivered a Common Data Services (CDS) subsystem for NEON. Through a single language and a single file format, CDS offers programmatic access both to NEON's central database and its many peripheral datasources. CDS is a software layer with which NEON engineers can create a wide variety of end-user applications for visualizing and analyzing ecological data. CDS's design provides NEON the flexibility to incorporate new data sources and functionality in the future.

Quantifying the Elements of Film Piracy

Client
Paramount Pictures

Faculty Advisor
Professor Christine Alvarado

Student Team
Jessica Blevins (Project Manager-Spring), Dan Ciliske, Rebecca Green (Project Manager-Fall), Moira Tagle
Online movie piracy is a major concern for Paramount Pictures. Since there are thousands of sites with illegal content, Paramount's antipiracy team requested a means of determining which websites are the most dangerous and which advertisers support those sites. Our first project deliverable is an algorithm for scoring the threat represented by a given website. The second deliverable is a web browser extension that retrieves information about the advertisements that appear on a given piracy site.

3D Social Augmented Reality

Client
QUALCOMM

Faculty Advisor
Professor Elizabeth Sweedyk

Student Team
Karen Gragg (Project Manager), Emily Fujimoto, Dan Fielder, Eric Mullen
Using QUALCOMM's Augmented Reality Software Development Kit, we have created two social augmented reality applications that support on-device content creation for the Android phone. Our initial application, Life Markup, allows users to place virtual content in the real world. Our second application, Virtual Room Designer, focuses the idea of Life Markup to allow users to arrange virtual furniture in a real room. In order to develop intuitive interfaces for the applications, we conducted a series of user studies.

Exploring the Feasibility of 2D Matrix Partitioning

Client
Sandia National Laboratories

Faculty Advisor
Professor Christopher Stone

Student Team
Audrey Lawrence (Project Manager-Spring), Michael Leece (Project Manager-Fall), Joe DeBlasio, Katie Ewing
Sandia National Labs has developed the Trilinos software framework for large-scale scientific and engineering problems. Large, sparse, matrix-vector multiplications arise frequently in their problems, and distributing the matrix and vector among many processors can produce significant speedup. We extended the Trilinos support for distributing large matrices, including additional partitioning algorithms and code to visualize and evaluate these partitions, and performed an empirical study of the results.

Yelp Crawler

Client
Yelp!, Inc.

Faculty Advisor
Professor Melissa O’Neill

Student Team
David Huie, Sabreen Lakhani, Anatole Paine, Yael Mayer (Project Manager)
Yelp is a review-based online business directory. To construct the most accurate and extensive local guide on the Internet, Yelp would like to use the multitude of business websites available on the Internet. To this purpose, the Yelp Clinic team built a web crawler that lever- ages machine learning algorithms to identify, examine and index the content of business websites. The resulting system will be integrated as a feature of Yelp's search engine.