OK2Say

Data Flow Diagram for OK2Say
Scanner Page
About Page

One-sentence description of the problem you're trying to solve

Some expressions that many feel are "normal" can be controversial and one controversial sentence can ruin someone's life.

One-sentence description of your solution

OK2Say helps users keep their sentences uncontroversial by using NLP to evaluate user text for controversial sentences.

How have you grown through this hackathon (what skills have you learned, what challenges did you face)?

Through this hackathon, I have learned how to create a NLP model in order to classify text. I also learned how to process, tokenize, and vectorize strings. I ran into many one-of-a-kind errors and learned how to more efficiently problem-solve which I will be able to take with me going forward.

What technical aspects of your project would you like to highlight (if any)?

Using TensorFlow, Keras, and Bert, I trained a deep NLP neural network on a dataset containing non-toxic and toxic tweets and achieved an accuracy of 95%. I processed a string input into sentences and preprocessed, tokenized, and vectorized them in order to be passed into the model.