Mad Hatter's Whisper

• 

  App Logo

• 

  Mobile Version of Project

• 

  Main UI for inserting piece of dialogue

Our inspiration was the interesting in machine learning and how it can benefit us.

Our project takes the Alice in Wonderland Script and predicts

We used machine learning and NLP to predict dialogue speakers by training SVM, Naïve Bayes, and Logistic Regression models. The text was processed with TF-IDF vectorization, and an ensemble-like decision system ensured accuracy.

Handling characters with very few dialogue lines was a major challenge, requiring us to filter out insufficient data. We also had to design a system to resolve conflicting predictions between models.

We successfully built a model that predicts speakers with decent accuracy by leveraging multiple classification techniques. Our ensemble-based approach improves reliability in uncertain cases.

We deepened our understanding of machine learning and NLP tasks. Additionally, we learned how to handle imbalanced datasets and conflicting model outputs effectively.

We plan to expand our dataset, fine-tune model performance, and explore deep learning approaches for improved accuracy. A user-friendly interface is also in the works for easier accessibility.

Built With

• css
• flask
• html
• javascript
• machine-learning
• pandas
• python
• react
• sklearn