Snigdha Chaturvedi Wins 2015 IBM Ph.D. Fellowship
The Department welcomes comments, suggestions and corrections. Send email to editor [at] cs [dot] umd [dot] edu.
On March 12th, Snigdha Chaturvedi won a 2015 IBM Ph.D. Fellowship for her research in Cognitive Computing. Snigdha is a fourth year Computer Science graduate student, working with Dr. Hal Daume III, and this is the second year she has won this Fellowship.
Snigdha has a long history doing research for IBM. Before coming to the University of Maryland, Snigdha was a Blue Scholar at the Information Management team at IBM-India Research Labs, New Delhi. There she worked on problems related to text segmentation systems and data-cleansing. Data-cleansing means segmenting given input into various components. And these components are then given to a human expert who writes rules to better understand the data.
The past two summers, she has interned at the T.J. Watson Research Center in Yorktown Heights, New York. She built relevance prediction models for non-factoid question answering in the Statistical Content-Analysis group. Snigdha worked on a team to develop a better means of answering open domain questions for internet search queries.
Her Ph.D.research focuses on machine learning, data mining, and information extraction. Snigdha, Dr. Hal Daume III, and Dan Goldwasser produced a paper in 2014 that analyzed student discussion forums. They created models that would look at different linguistic features in the discussion threads. These models would then be used to select the most important threads for the instructor to respond to. This research was what helped earn her a 2015 IBM Ph.D. Fellowship.
“The fellowship is great,” Snigdha explained. “It gives me the freedom to do the research I want to work on. It lets me set my own timeline.” Her research focuses on making computers read and understand text using latent variable modelling. A latent variable model relates sets of observable variables to inferred variables. She uses this to abstract out the content from sentences and paragraphs so that machines can better analyze them.