Machine Learning AdBlocker

By Klaas Schoenmaker, Justin Smael and Marjolijn Stam

This program uses over 100.000 links crawled from the top-500 most-visited websites. After being trained with AdBlock Plus's EasyList, our model is able to predict whether a link is an ad-related link or not with 95% accuracy.

Try the model

You can try the model by running the prediction script that uses a saved version of the model (data/trained_model.pkl). You pass the URL to block as a parameter to the script.

For example, run:

$ python predict.py "https://www.google.com/"

Files

tag.py: Takes the URLs in the CSV and maps all of them to a True/False value based on whether EasyList blocks it or not.

model.py: Trains and tests the model using the URLs and a TF-IDF weight. Model is saved in data/trained_model.pkl.

predict.py: Uses the saved model to predict a specified url.