Monthly Archives: April 2016

Kathryn Coulter: Tracking the Untrackable: A Social Network Analysis Approach to Investigate ISIS’s Recruitment Techniques Using Twitter Data

Proposal abstract:

The rise of non-state actors and globalization in the international system presents clear challenges to the historical methodology on power, legitimacy, and decision making. Terrorist groups like ISIS have since found new means to establish legitimacy in their control and power. Among their most successful techniques is the use of social media, particularly Twitter, to recruit and spread propaganda. Social network analysis and text analysis can provide frameworks to uncover why ISIS uses Twitter and can identify key trends in how they reach prosperity so an attempt at countering their momentum can be made. In this research, tweets pertaining to ISIS are analyzed in the hope to investigate their recruitment strategies and techniques.

K. Coulter Proposal (PDF)

K. Coulter Final (PDF)

Rahkeem George: The Software Development Process and Git

In theory, software engineering can be described by clear and systematic processes akin to those defined by the hardware engineering community. In practice, however, software development is a messy and error-­prone process that may differ greatly from those described in texts. How code is actually developed in practice is not very well understood, varying a great deal from one development team and project to the next. Understanding how code development evolves in practice may aid in understanding how processes can be improved not just in theory but in the field. Using a revision control system to track the evolution of code for projects considered to be very well developed may shed light on this problem. In this work, we use the distributed version system, git, to track the evolution through time of well­-developed projects on Github to see how the actual development process compares to software development theories. By checking out successive commitments, parsing the source code, and tracking various metrics for each commitment, quantitative measures will be obtained and analyzed.

R. George Final (PDF)

Jacob Koko and Sebastian Florez: Cost Effective IR-Free Eye Tracking on Mobile Devices

The purpose of our project is to improve upon modern eye tracking technologies that are used for psychological experiments. As it stands, the technology used for such experiments is expensive, nonintuitive, and utilizes infrared cameras. We plan to offer an alternative to existing technologies by introducing an effective mobile solution that does not make use of any infrared technology. More specifically, our technology can act as a low cost substitution for eye tracking and data analysis because it only requires a standard mobile device. We believe that our research is applicable to more than just psychological experiments. For example, brand marketing could find substantial use in this technology. This research is largely theoretical in regards to the mathematical algorithms used for eye tracking, but the implementation is quite extensive as well. Therefore, we intend to delegate the work into two separate areas including mathematics and software implementation.

J. Koko and S. Florez Final (PDF)

Connor Mashburn: Fitbit iOS Fitness Application

Smartphones have become an easy and affordable commodity, has opened the door to easily and constantly recording and computing sensor data. Bluetooth enabled heath sensors, such as fitbit and Shimmer modules, have allowed the extrapolation and storage of pertinent medical and health data. Using social media along with fitbit would allow for people to compare and be ranked in their performance during exercise and running, with friends who also use the application validated through social media. The application would allow for social integration, and competition, without having to expose their scores to all of their friends or other users who aren’t Facebook friends.

C. Mashburn Final (PDF)

Melissa Abramson: Signer-Independent Recognition of Static ASL Signs

As the Deaf community continues to grow, so does the market for tools that aid communication between the Deaf and hearing communities. One such tool is software for automatically recognizing ASL signs that could assist in learning and communicating in ASL. We have investigated the development of such a system that recognizes 28 static ASL signs. We have also developed several tools for data collection, modification, and classification. We found several common machine learning techniques that are not ideal to use for ASL recognition. We also explored the use of Caffe, a deep learning framework using convolutional neural networks. Due to hardware limitations and time constraints, we were not able to complete the training using Caffe and thus are unable to confirm if this is the ideal approach.

M. Abramson Final (PDF)