Skip to content
ML to predict the likelihood of a tweet coming from a bot
Python JavaScript Jupyter Notebook C C++ HTML Other
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.


My second-ever machine learning project. This assigns a "spamminess" score to any input text using an embedding encoder + NN architecture.

I tried using one-hot encoding at first. 0/10 would not recommend. Bricked my computer. Twice.

WARNING: Data not included!

Download and place the .gz file into the root directory to get the needed embedding vector files. will do the following:

If gnews.embedding is not present, the vector file will be used to generate one. This will take a few minutes.

If classifier.joblib is not present, the network will be trained with the CSV data, which is also not included in this project. You should not normally need to do this.

If leaderboard.txt is not present, a new spam leaderboard will be generated with dictionary data (not included.) This takes a LONG time. Do not remove this file. You will be sorry.

Once the network is initialized, it will prompt the user for a tweet input. When given, it will print the "spamminess" score. Typing exit will exit the loop. will:

Generate a graph of the "spammiest" words using a snapshot of the leaderboard file. The graph is included at img/spammiest.png, so there is no need to rerun this.

Data sources:

You can’t perform that action at this time.