| Student Research |
|
Computer Science Papers and Proposals
|
| 2001 | 2002 | 2003 | 2004 | 2005 | 2006 | 2007 | 2008 | 2009 |
| 2008-2009 Computer Science Papers and Proposals |
Document Classification Through Data Mining Social Media Networks
Paul Kleiber
ABSTRACT: Social Media Networks, such as Delicious, Facebook, and Flickr are some of the fastest growing
sites on the Internet. Many of these sites allow users to share content, and to tag this content
with categorical meta-information. The enormous number of documents contained on these
sites, some of which have been tagged by thousands of users, can be used as an invaluable
resource for document classification. In this research, Social Media Networks will be used to
categorize plain-text documents, and the results will be discussed. The method makes use of
a modified term frequency-inverse document frequency phrase weighting to retrieve similar
documents from social media networks, then analyzes the tags that have been applied to
them and heuristically categorizes the original document based on these tags.
|
Applications of Neural Networks and Hidden Markov Models to Gesture Recognition Through Head Tracking
Zeb Long
ABSTRACT: Both Neural Networks and Hidden Markov Models have a history of use in tasks involving pattern recognition and other forms of data classification. On the surface, both systems perform a similar task - that of recognizing and categorizing data sets. In both systems, this is accomplished through iterative training of the system using data sets whose classification is know. The way this training is accomplished differs between the two systems and is the focus of a large part of this paper. Once the system has been trained it is able to accurately categorize a new data set based on which category of training sets it most closely resembles. The systems implemented here will be trained to recognize gestures such as "lean left" and "nod" using data derived from two infrared markers on either side of the head of a user. For each system, the implementation will be described and the performance of the system will be evaluated.
|
RISC- Risk Identification Status and Control
Andrew Packard
ABSTRACT: With rising concerns in project management due to economic conditions and job security, more
companies are looking at proper risk management to safeguard their assets and capitalize on
opportunities. Modern project management in the United States started in the 1950s with two
breakthrough techniques for planning and scheduling, the critical path method or CPM, and
the Program Evaluation and Review Technique or PERT. Both techniques are mathematical
tools for estimating uncertainty in task duration and cost, however, neither technique handles
tracking risks that potentially threaten the overall cost, or critical path of any project or
portfolio of projects. The purpose of this research is to implement a unique project
management software package that assists the user in tracking and controlling risk. This is done
not only for a single project, but for an entire portfolio of projects.
|
Finger Tracking and Gesture Interfacing Using the Nintendo Wiimote
Ross Williams
ABSTRACT: The Nintendo Wii-Mote is a revolution in gaming technology, being able to detect motion
as it is moved around as well as infrared sensors linking it to the Wii console to show what the
user is pointing at. Other gaming enthusiasts have been able to use the Wii-Mote for more than
just a controller for the Wii by connecting it to a computer through its Bluetooth connectivity and
turning it into a receiver for infrared signals to create finger-tracking. This new type of gesture
interface could prove to be a very useful tool, not only for gaming but for anyone who wants to
make a presentation more interactive rather than using the traditional mouse and keyboard.
|
Mobile Advertising Engine
Jeffrey Wray
ABSTRACT: In recent years, our society has become engrossed by mobile devices. Likewise,
we have also seen a significant decline in print media, with organizations now favoring
the web. With this significant change, companies and advertisers are seeking new ways
to deliver marketing materials to consumers. With this emergence of mobile devices and
applications, there is now an opportunity to develop a delivery engine that would serve as
the backbone to deliver advertising content to mobile devices. With the primary focus of
developing a delivery and reporting engine for advertising campaigns, this paper will
discuss the pitfalls of existing technology and propose enhanced ways to deliver pertinent
location based advertisements and sales offers.
|
|
