Sentiment Analysis NLP for QMIND

For QMIND (Canada's largest undergraduate artificial intelligence and machine learning organization), I was part of an innovation design team that developed a machine learning program to analyze sentiment from typed text using large quantities of data and natural language processing techniques.

On the team, I researched use cases of multiple ML models and helped source data to develop an innovative solution for mapping sentiment to text by using a dataset of Reddit comments representing 1 of 27 emotions provided by Google. We used NLP techniques such as tokenization, stemming, lemmatization, and stop word removal to preprocess our dataset, then fine-tune an ALBERT model to classify a sentence based on which emotion is most prevalent.