Skip to content

Latest commit

 

History

History
21 lines (16 loc) · 962 Bytes

README.md

File metadata and controls

21 lines (16 loc) · 962 Bytes

Soft Spoken

By Adam Snaider, Mathew Li, Shivan Vipani, and Michael Wono

About

This project was created at SB Hacks IV

It is a machine learning algorithm to detect the levels of toxicity in online comments. The idea and the dataset came from this Kaggle competition

Usage

In order to run the training step, one must first clean the data and create the word embeddings. To do this, run the python script vocab.py. It will generate all the data needed.

After all the data has been processed, it's time to train the model by running the train.py module. This model was created using TensorFlow so make sure that TensorFlow is installed.

After train.py finishes its run, you can run the server.py module and navigate to 'http://127.0.0.1:9876/'. From here, you just write a comment in the text box and click submit to show the results in the chart