At UW-Madison my research contributes towards developing NLP and ML algorithms for applications on user generated text data and social media content. A quick and brief overview of these projects is as follows,
Center for Health Enhancement System Studies (CHESS):
The goal at CHESS is to help people with Substance Use Disorders (SUDs) by providing machine and human moderated intervention. At CHESS my research has contributed towards building an online flagging system that analyzes the day-to-day message content of study participants and predicts relapse risk. Novel techniques combining psycholinguistic analysis and classic machine learning techniques were used to build the first iteration of this flagging system. Currently the flagging system flags messages indicative of relapse risk for approval/intervention by a human moderator. In the second iteration of the system, human assisted feedback will be used to improve classification accuracy.
Social Media and Democracy (SMAD):
The goal of this research project is to explore the social and psychological influences of social media and user generated content such as political blogs through novel NLP algorithms. On this project, my focus is towards extracting efficient features from short texts such as Tweets and/or social media commentaries for use in down stream tasks such as Text Classification, Sentiment Analysis, Machine Translation etc using multiple sources of text, audio and video signals.
Center for Health Enhancement System Studies (CHESS):
The goal at CHESS is to help people with Substance Use Disorders (SUDs) by providing machine and human moderated intervention. At CHESS my research has contributed towards building an online flagging system that analyzes the day-to-day message content of study participants and predicts relapse risk. Novel techniques combining psycholinguistic analysis and classic machine learning techniques were used to build the first iteration of this flagging system. Currently the flagging system flags messages indicative of relapse risk for approval/intervention by a human moderator. In the second iteration of the system, human assisted feedback will be used to improve classification accuracy.
Social Media and Democracy (SMAD):
The goal of this research project is to explore the social and psychological influences of social media and user generated content such as political blogs through novel NLP algorithms. On this project, my focus is towards extracting efficient features from short texts such as Tweets and/or social media commentaries for use in down stream tasks such as Text Classification, Sentiment Analysis, Machine Translation etc using multiple sources of text, audio and video signals.
