Skip to content
Go to file

Latest commit


Git stats


Failed to load latest commit information.
Latest commit message
Commit time


A personal NLP and ML-based scoring proxy for RSS and Atom feeds. Run with

gunicorn -w 7 -t 3600 -b app:app

or equivalent (it's a flask app).

Entry points are /feed/<url> to proxy the feed at url and /learn to retrain the model. It adds a feedback bar atop each entry that says "Time Well Spent" or "Time Wasted". You need to click only if you agree. If you disagree, it means you and the algorithm agree, and no further action is needed. It needs some positive and negative feedback before training the model for the first time. The title is prefixed with a score which is 100*estimated probability of being interesting. But in fact you could try to teach the model any binary classification. I just use it to prioritize my reading without skimming forever. It also adds highlighting to each sentence with intensity proportional to the probability of "interestingness", and a superscript with the actual number. This may be annoying to most, but it helps me understand what the algorithm is doing. Installation is a work in progress. The hard part is installing Infersent as a subdir of the root of the repo. See also the blog entry for additional information.


AI-powered scoring proxy for RSS or Atom feeds




No releases published


No packages published
You can’t perform that action at this time.