https://github.com/the/interesting
News feed denoiser
https://github.com/the/interesting
Last synced: 4 months ago
JSON representation
News feed denoiser
- Host: GitHub
- URL: https://github.com/the/interesting
- Owner: the
- License: mit
- Created: 2017-07-09T22:39:33.000Z (almost 9 years ago)
- Default Branch: master
- Last Pushed: 2024-07-12T16:10:42.000Z (almost 2 years ago)
- Last Synced: 2024-07-12T18:23:36.424Z (almost 2 years ago)
- Language: Python
- Size: 267 KB
- Stars: 1
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE.txt
Awesome Lists containing this project
README
Interesting
===========
Collect tech news articles and tweets in one place and filter out the interesting content by tagging items as either interesting or boring (vote up/down).
Installation
------------
Ensure you have Python 3 installed. The application is tested with Python 3.6.1 on macOS Sierra.
If you are on macOS and use homebrew:
brew install python3
Create virual environment and activate it:
python3 -m venv .env
source .env/bin/activate
Install dependencies:
python -m pip install -r requirements.txt
Configuration
-------------
You can activate different data sources in `config.yml`. If you want to activate Twitter you need to create a Twitter application with your own account on https://dev.twitter.com. Copy the required keys and secrets into the config.
The application will automatically collect data from all the sources that are configured in the config under source.
Running the application
-----------------------
Execute `run.sh` (or see its content how to do it manually).
Point your browser to http://localhost:5000.
Content of the enabled sources will appear in the corresponding sections. By up- and downvoting items the systems learns the user's interests. For items to appear in the _interesting_ section for a particular data source one needs to have at least one upvote and one downvote. The more data is gathered the more accurate the prediction becomes (200+ tagged items for a data source is a good start).
Classifier evaluation
---------------------
There is an extensible list of classifier pipelines defined in `classifier.py` that is used for evaluation. Run `python eval.py` to get cross validation metrics for each pipeline (precision, recall, etc.).