19.01.2013 Views

ABI-ONE ghc program v14.indd - Grace Hopper Celebration of ...

ABI-ONE ghc program v14.indd - Grace Hopper Celebration of ...

ABI-ONE ghc program v14.indd - Grace Hopper Celebration of ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

GENERAL POSTER SESSION<br />

W e d n e s d ay | s e p T 3 0<br />

Integrating and Querying a Web <strong>of</strong><br />

Hybrid Data Using Wikitology<br />

Presenter: Zareen S. Syed (University <strong>of</strong><br />

Maryland, Baltimore County)<br />

The Web is mixture <strong>of</strong> interwoven<br />

networks <strong>of</strong> hybrid data such as free<br />

text, links, media, tables, triples etc. There<br />

is a need for new approaches for integrating<br />

information from heterogeneous sources<br />

and in heterogeneous representations.<br />

Through the development <strong>of</strong> Wikitology<br />

knowledge base derived from Wikipedia,<br />

we demonstrate a novel architecture for<br />

integrating structured and un-structured data<br />

and provide an integrated query interface to<br />

applications.<br />

Intrusion Detection System Using<br />

Support Vector Machine<br />

Presenter: Meenakshi Amirtharaj (Amrita<br />

Vishwa Vidyapeetham)<br />

Nowadays the need to classify legitimate<br />

users and intruders is <strong>of</strong> high priority.<br />

Support Vector Machines (SVM) is used for<br />

classification. Proximal SVM gives a wider<br />

margin by pushing the data points as far<br />

as possible, yielding better classification.<br />

Reduced SVM reduces the data set size and<br />

uses it for training, leading to less training<br />

time. Therefore the data set is reduced and<br />

feed into ProximalSVM. Feature selection is<br />

preformed. Efficient and faster classification<br />

is obtained.<br />

It Does What and It’s Shipping When!?<br />

Testing in an Experimental Environment<br />

Presenter: Lilia Paradis (Micros<strong>of</strong>t Live<br />

Labs)<br />

Testing for Micros<strong>of</strong>t Live Labs presents<br />

unique challenges. The Live Labs ethos<br />

involves experimentation on a small scale,<br />

taking place during quarterly innovation<br />

weeks. Whether and when to ship the<br />

projects that arise from experimentation is<br />

decided democratically.<br />

This poster describes the test problems<br />

and questions existing in Live Labs and the<br />

heuristics that our testers use to turn around<br />

quickly on the next cool thing.<br />

JCinema: S<strong>of</strong>tware for Producing<br />

Photorealistic Renderings <strong>of</strong><br />

Biomechanical Simulations<br />

Presenter: Spiridoula Lagaditis (University<br />

<strong>of</strong> British Columbia)<br />

My research focused on developing 3D<br />

visualization tools for s<strong>of</strong>tware that<br />

works as a bridge between a biomechanical<br />

simulator and an open source rendering<br />

engine to produce high-quality output. I<br />

developed a UI that produces the system in<br />

3D given a large data-set generated by the<br />

simulator which describes how the model<br />

interacts in a physical environment. The UI<br />

allows for user control and ease for further<br />

investigation.<br />

Jesus and the Jimson Weed - How Do<br />

We Derive Meaning from Tags, Text,<br />

and Queries to Support Improved Image<br />

Access?<br />

Presenter: Irene Eleta (University <strong>of</strong><br />

Maryland, College Park)<br />

Accessing images from on-line digital art<br />

collections remains challenging because<br />

traditional text-based search techniques are<br />

inadequate. Social tagging is a promising<br />

strategy for improving image retrieval in<br />

such collections. This poster examines<br />

how four areas <strong>of</strong> research - multi-lingual<br />

tagging, subject categorization <strong>of</strong> tags<br />

and queries, multi-word tags and terms,<br />

and behavioral approach to understanding<br />

search - shed light on this problem.<br />

Just a Spoonful <strong>of</strong> Sugar: Mitigating the<br />

Effects <strong>of</strong> Technology Change<br />

Presenter: Beverly Magda (Georgetown<br />

University)<br />

Because <strong>of</strong> a White House mandate<br />

for electronic health records by 2014,<br />

many hospitals are experiencing technology<br />

change at a fast pace. Technology projects<br />

notoriously have high failure rates and<br />

research has shown the top reasons why<br />

is not because <strong>of</strong> the technology itself, but<br />

because <strong>of</strong> human elements. This research<br />

examined benefits and application <strong>of</strong> leadership<br />

support, communication, training, and<br />

end-user involvement in hospital emergency<br />

departments implementing electronic<br />

healthcare records.<br />

Land Cover Change Detection Using<br />

Data Mining<br />

Presenter: Vasudha Mithal (Indian<br />

Institute <strong>of</strong> Technology, Kanpur)<br />

The land-cover-change-detection problem<br />

is one <strong>of</strong> detecting when the land cover<br />

at a given location has been converted from<br />

one type to another.<br />

The data used in this problem is the<br />

timeseries <strong>of</strong> FPAR(spectral measure <strong>of</strong><br />

the amount <strong>of</strong> vegetation) for different<br />

geographical locations. The idea is to<br />

develop algorithms which take advantage<br />

<strong>of</strong> the inherent characteristics <strong>of</strong> spatiotemporal<br />

data and are scalable i.e. suitable<br />

for high-resolution earth science datasets.<br />

Language Detection Using Hidden<br />

Markov Models - Optimized Algorithm<br />

for Wikipedia Pages<br />

Presenter: Rajitha Rani Satharla<br />

(Autozone)<br />

We intend to solve the problem <strong>of</strong><br />

language detection -determining<br />

if given text is in a particular language or<br />

not. We designed and implemented an<br />

algorithm that can extract Wikipedia pages<br />

<strong>of</strong> a given language, parse the HTML to<br />

retrieve content <strong>of</strong> that particular language.<br />

While automatically filtering out unnecessary<br />

and redundant content, and Model that<br />

language as a hidden Markov model for easy<br />

detection using an optimized algorithm that<br />

works with bigrams.<br />

56 Anita Borg Institute for Women and Technology | <strong>Grace</strong> <strong>Hopper</strong> <strong>Celebration</strong> <strong>of</strong> Women in Computing

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!