Great to hear :)

This could be applied to any other industry if there are lots of feeds. For web development news there are hundreds of blogs all with RSS feeds so they get imported, manually, into the system.

Manually there are a few controls, adding new RSS feeds the system discovers and populating the tag/category area.

The tag category area is pretty long now. Each tag is related to a category, e.g. jquery is related to the javascript category and has a score of 0.8. In this category will also be hundreds of other keywords with scores. The higher the score the more possibility it has going into the category.

When the system finds a new article from the RSS feeds it will scan through all the tags, regardless which category. It'll then add up the scores and which ever category got the best score it'll be added to that.

The hardest part was shit articles. Readwriteweb and techcrunch are a good example. These sites talk about apple and none-related web dev news so I have a "shitbin" category. This category probably has the most tags. It'll have things like apple, microsoft, SLA and hundreds more. It's the category I have to manually spend the most time adding tags to so the content on the site keeps relevant.

Also have a bunch more tools for finding popular new tags. This tool will give me a list of common words found in articles that haven't been categorised. It'll find words like "blue" but I have to put these into an ignore list.

Scoring is done by a number of ways including how many people tweeted it, how many results in google and how many people liked it on Facebook. This score is hidden. Something with a popularity of 200 could have a score of 15 and never show up on the front page. The threshold is 35. Once the article gets a 35 score it can be promoted to the front page.

Loads of other things going on as well that I've probably missed :)