Case Study: Tweet classification microservice

Summary: A marketing startup wanted to know what topics specific Twitter users were talking about and how those users felt about those topics. This information would enable the company to target undiscovered influencers in particular domains, decreasing the cost of brand partnerships. I worked with the client to create a useful topic taxonomy, designed the machine learning system, got the training data, built the models and deployed the microservice that classified tweets and scored their sentiment.

Tools: Python, Java, CoreNLP, Spring, Bash, Naive-Bayesian Classification, TF-IDF, Bag-of-words, AWS