This is the Furry Detector program as seen on my YouTube!
It uses the Twitter API, TensorFlow, and Tkinter to make an all-in-one furry detecting program.
This machine learning model uses the text given from tweets to determine if a Twitter user is furry or not (it can also be applied more generally to non-twitter applications, but results may vary).
First it takes a Twitter user, and collects their tweets. Then, using word_list.json, a 1000 length word vector is generated that has the counts of all the words used. It is normalized between [0-1], 1 being the most common occuring work, and 0 being not used at all. Finally, this word vector is passed through the model, and a confidence score is given.
For my own sanity, the classes in this repository are NOT packageble and are self contained. main.py takes advantage of this, and is a simply GUI interface for you to take advantage of and sample the project. Here are some of the different files you can find:
This directory contains the furry_detector.py and parser.py files. Here, you can parse text and run it through the FurryDetector class to determine if the text written was done so by a furry.
This directory contains the wrapper.py file. This has a basic class that does API calls to Twitter for tweets of a user, and the user itself. It is barebones, and is really only meant for usage within main.py.
In the file twitter_key.json you must put your bearer token and whether or not you want to use Twitter v1.1. By default, this program will use v2.0
Read Twitter's tutorial here for obtaining an API key.
Run main.py to see a GUI for selecting a Twitter User! Type who you want, and the model will run on their account to see if they are a furry! :D
Please be sure to read requirements.txt for the modules used
Contributions are welcome!
I am always willing to look at PR's and issues if you think there is something that can be fixed or added :)
This repository has gotten a long overdue makeover. I have refactored the project so that the code is much cleaner (and does not look like it was written by someone who only had 6 months experience with Python), and also so that it runs somewhat faster.