Clint P. George

I am a doctorate student from the Department of Computer and Information Science and Engineering, University of Florida. My doctoral advisors are Dr. Joseph N. Wilson and Dr. Hani Doss.

My research interests include:

• Bayesian nonparametric models
• Markov chain Monte Carlo methods
• Large scale statistical learning

I am also collaborating with Dr. Daisy Zhe Wang in the domain of large scale statistical learning problems at the Data Science Research lab and Database Systems Research Center, University of Florida.

Before joining to the doctoral program, I was a master's student at the Department of Computer and Information Science and Engineering, University of Florida. Dr. Joseph N. Wilson was my thesis advisor. I completed my undergraduation in Computer Science and Engineering from the College of Engineering, Trivandrum, University of Kerala, India.

We are looking at different methods for the hyperparameter selection of hierarchical Bayesian models such as topic models. This project deals with modleing large document collections using different topic models such latent semantic indexing, latent Dirichlet allocation, and hierarchial Dirichlet process. We are also trying to improve on the state of the art of statistical learning methods such as supervised classification, unsupervised clustering, and learning to rank have been employed to reduce manual labor and increase investigative speed and efficiencies for EDiscovery.

This was a joint research project we did with SurveyMonkey (SM), a large scale online survey management system. We looked at a problem of automatic topic extraction, categorization, and relevance ranking model for surveys and their questions from different languages such as English, Spanish, Portuguese, German, and French. Automatically generated question and survey categories are used to build question banks and category-specific survey templates. For this work, I collaborated with Dr. Joseph Wilson, Dr. Daisy Zhe Wang, Dr. Liana M. Epstein (SM), and Dr. Philip Garland (SM). We looked at improving the traditional collapsed Gibbs sampling algorithm of the well known topic model latent Dirichlet allocation and came up with a 2-stage Gibbs sampler based on product partition models. Also, a new topic inference algoirthm for the newly encountered data points without retraining the learned model based on a Metropolis Hasting algorithm was developed. For this work, I collaborated with Dr. Paul Gader , Dr. Joseph Wilson, Dr. George Casella (late), Taylor Glenn, Dr. Claudio Fuentes, and Dr. Vikneshwaran Gopal. I worked with Dr. Joseph N. Wilson and the Morpheus research team, to build a semantic question answering system using deep web sources by exploiting information from the DBpedia and Wikipedia along with sample query answering strategies provided by users.

1. A Machine Learning Based Topic Exploration and Categorization on Surveys  [URL]. Clint P. George, Daisy Zhe Wang, Joseph N. Wilson, Liana M. Epstein, Philip Garland, Annabell Suh. ICMLA 2012, Boca Raton, Florida, USA. December 2012.
2. Online Topic Modeling for Real-time Twitter Search [URL]. Christan Grant, Clint P. George, Chris Jenneisch, and Joseph N. Wilson. TREC 2011 Notebook. October 2011.
3. Topic Learning and Inference Using Dirichlet Allocation Product Partition Models and Hybrid Metropolis Search [URL]. Clint P. George, Taylor C. Glenn, Joseph N. Wilson, Paul D. Gader, Claudio Fuentes, Vikneshwaran Gopal, and George Casella Technical Report. The department of Computer and Information Science and Engg., University of Florida. September 2011.
4. Dirichlet Allocation Using Product Partition Models [URL]. Claudio Fuentes, Vikneshwaran Gopal, George Casella, Clint P. George, Taylor C. Glenn, Joseph N. Wilson, and Paul D. Gader Technical Report. The department of Computer and Information Science and Engg., University of Florida. September 2011.
5. Morpheus: A Deep Web Question Answering System [URL]. Christan Grant, Clint P. George, Joir-dan Gumbs, Joseph N. Wilson, and Peter J. Dobbins, In Proceedings Of The 12th International Conference on Information Integration and Web-based Applications and Services (iiWAS2010). Paris, France. November 2010.

1. Clint P. George, A Realm based Question Answering System using Probabilistic Modeling [URL]. MS Thesis, University of Florida, Advisor: Prof. Joseph N. Wilson