Norma is a Doctor of Professional Studies in Information Management (DPS, iSchool, Syracuse University) with dissertation work on computational linguistics. She modeled negation prediction in social media data using supervised machine learning. As a practitioner, she developed an extensive career as information manager working for the private and public sectors in three countries (Argentina, Canada and the US). Norma’s areas of expertise range from leading the development of innovative solutions for effectively managing and disseminating knowledge-oriented digital assets (such as data and documents), to building the required synergies among partners that will guarantee product success. She based the success of her projects on the design and implementation of applied research methods in order to make decisions that are based on rigorous evidence and scientific reasoning.
Norma was the 2018 Doctoral Prize Recipient for the best dissertation in the Information Science and Technology Program, School of Information Studies.
I am a fitness and nutrition enthusiast. I also enjoy very much listening to music, particularly American southern blues. I also have two adorable pets who keep our little house in Alexandria, Northern Virginia, very lively.
My current research focuses on mining and analyzing social media text (tweets) for studying the phenomenon of negation in natural language and how it links to irony, sarcasm and (more broadly) negative sentiment in public opinion.
The natural language literature has widely reported that modeling the automatic detection of negation in text is a challenge task. This is mostly due to the elusive ways in which humans negate things, both indirectly when we try to be polite (like refusing an invitation) or when taking somebody/something as the target of sharp critique, as when using sarcasm.
I use Python programs (both open source and my own) and classification algorithms (mainly SVM and CRF) to find predictive patterns in textual data that will help modeling negation with the main goal to, eventually, improve F1 scores of algorithms that identify negation in natural language. I am particularly interested in how valence shifters (specially approximate negators) as distinctive types of tokens affect negation in chunks of text. I also develop Python programs for data manipulation such as cleaning and formatting.
Besides my appointment as Adjunct Professor for computational linguistics courses at the iSchool, I designed the curricula and learning approach for “Data for Effective Policy Making”, an edX-IDB MOOC on foundations of statistical procedures and data literacy for evidence-based decision making processes by public policy makers, primarily for Latin American and Caribbean government audiences but readily applicable elsewhere.