This bot takes a bit of initial training from the web browser to select the kinds of articles that are relevant. After that training, the script actively searches for keywords and favorites tweets. Other services have proven that auto-favoriting tweets is an effective way to grow following. Typically these favorites convert to followers at about 1-2%. By using a new algorithm, I (hopefully) have made a better-converting selection, thus boosting the effectiveness of the 250 fav per day limit.

The Machine Learning:

  • Identifies relevant searches
  • Combs through tweets returned and orders by follower to followe ratio & activity metric.
  • Reads tweets and any links and ranks their significance based on Naive Bayesian Classification.
  • Favorites highest ranking tweets

The theory behind this approach is that high activity and ratios near 1.0 are the best candidates for following back from the favorite. This has been proven to be true during my experience growing my personal following.

Built With

Share this project:
×

Updates