DB Seminar [Spring 2015]: Gisele Pappa

Abstract:

In this seminar I will present three of my ongoing projects. I will start talking about dengue fever modeling, its challenges and opportunities.
Dengue fever is a tropical, mosquito transmitted disease that has been growing significantly in the past decade. The main goal of this project is to exploit real cases data and Twitter data to generate a predictive system that allows government policies to be put in place up to two weeks before the disease outbreaks really happen.
Following, I will present a Bayesian approach that integrates graph-based and textual information to improve location inference on online social networks. This is relevant to the previous project, as the dengue prediction model is highly dependent on spatial information.
Finally, I will present some work on automatically generating data mining algorithms, i.e., given a new dataset, how can you aid a practitioner to generate a good algorithm tailored to that dataset