Hi everyone, welcome to my first blog post! I’m Ethan, a senior at BASIS Independent McLean.
For my senior research project, I will be researching AI at Clarabridge, a leading software company in advanced text analytics. The purpose of my project is to build a humor detection model. Identifying humor adds another skill to Clarabridge’s wide range of offerings, which already includes sentiment analysis, emotion recognition, intent detection, and predictive analytics. Adding a model that can detect humor would enhance Clarabridge’s understanding of customers’ feedback in surveys, customer service calls, emails, and social media platforms, helping create a better customer experience for users. Moreover, key concepts and findings from building this model could be extraordinarily useful in related natural language processing classification models, including irony and sarcasm detection.
To complete this project, I will first implement a Flask application service that can be deployed using Docker to train machine learning models, make predictions, and test trained models using k-folds cross validation. Afterwards, I will approach humor detection from three angles. The first is the simplest and more naive approach: using rules to classify humor. Then I will explore using a Random Forest Classifier with my Flask app service, and lastly I will use FastText, an open-source text classification library built by Facebook’s AI Research lab, to perform humor classification. Through experimentation as well as individual research in this rapidly advancing field of machine learning, I aim to explore the strengths and weaknesses of each approach. By the end of the project, I hope to have gained a thorough understanding of humor detection using AI. I also hope my humor classification model achieves a high performance and helps Clarabridge better analyze the feelings and emotions of customers.
I can’t wait to begin my internship at Clarabridge.
Stay tuned for next week’s post!