Skip to content
master
Switch branches/tags
Code

Latest commit

 

Git stats

Files

Permalink
Failed to load latest commit information.
Type
Name
Latest commit message
Commit time
app
 
 
bin
 
 
 
 
db
 
 
lib
 
 
log
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Build Status

News aggregator app

This is a Ruby-on-Rails app for running (German) news aggregator websites. Today, it powers:

Reasoning

I want to follow news of those two areas but struggle with RSS, as it is too much for me too process - I want to see the most "relevant" sources at once, without investing too much time. Other sources, like Twitter + Reddit I found too noisy to follow.

This is why I created that app

News fetching + scoring algorithm

The admin of the apps curates a list of trusted sources. Those will regularly checked for new content. Following news sources are supported:

  • RSS/Atom feeds (FeedSource)
  • Podcast via RSS/Atom (similar as FeedSource but different visual)
  • Twitter Streams
  • (in planning) RedditSources - subscribe whole /r/'s

In similar fashion, the app checks popularity of the news in social network, that means:

  • Facebook likecount (as reported by Facebook Like Button)
  • Twitter retweets + favorites (as reported by Twitter API)
  • XING + LinkedIn shares (as reported by regarding Widgets)
  • Reddit total score sum in all subreddits (if exists)
  • Each of those sources is configured with a different value (e.g. Facebook likes are more common, so less value than XING share)

The admin of the sites can give a Source individual:

  • Base factor (that means, how much "Likes" any link of that website is worth, can also be negative too remove noise from some sources)
  • Multiplicator, e.g. 0.2, 1.0, 2.0 - each like will be multiplied by that number -- Some sources have a much higher reach and can be leveled out so the news are more broad

Altogether, the score is calculated regularly for fresh links. For Display on the homepage, the freshness is also important - the older the link, the more the score is reduced.

Topics

The topic matching is very simple - just simple keyword lists. That means, the categorization is far far from perfect or even good. It might be an area of further development :)

Newsletter

It is possible to subscribe via E-Mail. Then, once per week on sunday, you will receive a Mail with from the selected topics.

Development

As it is a Rails app, you can try to run it yourself. First make sure to have installed:

  • Linux or macos
  • Ruby 2.2+
  • Postgresql 9.5+
  • Redis (For Background jobs) 3.0+
  • nodejs 10.0+ and yarn
git clone https://github.com/zealot128/filter-app.git
cd filter-app
bin/setup

To run the app + background worker + webpack run:

foreman start

If anything fails, then you might need to adjust the config/database.yml to configure the correct database or run the individual commands

Otherwise: you can check out the running app on localhost:5000