NayesDog

RSS/Web news reader with Naive Bayes powered recommendations

• doglib.py is preprocessing lib.
• naylib.py is ML lib.
• facelib.py is web interface lib.

Dependencies:

feedparser for RSS beautifulsoup4 for web scraping

Installation:

To install latest release (pip of python2.7, pip2 on my system):

pip2 install nayesdog

To install development version:

pip2 install git+https://github.com/MLdog/nayesdog

Usage:

• To run nayesdog you only need to run nayesdog in a terminal

• Default config files are stored in ~/.nayesdog
  • config.py: configuration file. Modify this file to include new RSS feeds or web scrap news, or remove the existing ones.
  • tables.py.gz: Trained model, containing the word counts that are used by the Naive Bayes Classifier. You can copy your model, use it somewhere else and share it.
  • .previous_session: A hidden file that stores the state of your session. If you have problems, try to erase this file.

• By running nayesdog with --config option you can have different nayesdogs trained for different purposes and different RSS feeds.

Example configuration can be found at https://github.com/iprokin/dotfiles/tree/master/.nayesdog.

Python library:

You can import the nayesdog library into python2.7 projects with import nayesdog

To-do

• Each time nayesdog is run, preprocess_html loads all urls even they were previously loaded. This unnecessary work and resulting delays should be avoided.
• Make dropdown menu to lie above "Toggle images" and "Train" buttons.
• Add UI toggle for showing titles only / full content / summarized content
• Save the last feed open and the last folder open
• Upload last version Pypi
• Parse HTML (One more dependency)
• Summarization: https://github.com/neopunisher/Open-Text-Summarizer
• Topic modeling and word search according to topic distance and likability
• Visual search of documents ordered by topics
• Test and spot bugs
• replace shelves for cross-compatibility (?)
• being able to enter feed names that contain spaces!
• Move "toggle images" function to config.py instead of having button?
• Should we remove deleted article also from WordCount dict?