Sentiment Analysis Negometer aimed to evaluate and display in a 5.0 meter scale the fact of how negative a parent was being in front of his/her children in everyday life. I worked in formulating and putting together the pieces of this nascent project at Duke University.
Analysing the sentiments of the speech of a parent was done in the following way - separating the audio from the speech, analysing the noise signals along with detecting sentiments in the text converted from speech. The features were then aggregated for a classification task. I focused on sentiments of text via scoring each sentence as a negative, positive or neutral comments via Support Vector Machines.
As a beginner in the field of machine learning, this project proved to be an immense source of inspiration for me. Looking back, I realize the various other methods that could have been used to tackle this problem as compared to our rudimentary approach.