Future of Web Apps (FOWA)
Session: The Future of News
Note by Phil Wilson:
Digging to date just based on overall popularity. Is this useful for the audience. Providing content that's valuable to them.
Recommendations can be based on similar users - they fill the gap between people you know and the great unwashed - this is what services like last.fm do.
- this is old ground, can't imagine it's really new to anyone in the audience. The traditional problem here is never showing you anything new. A good recommendation recently was injection on news coverage that you don't normally look at to ensure you see content you normally wouldn't.
Invented their own graph-based backend!
Authentication, reputation and trust.
Shows that this isn't a solved problem and there's probably a lot of money to be made. Recommendation is the search of the noughties.
Show the impact you have - when you mod up or down, let the user know how many people they've affected.
Recommendation API which news services could integrate to show stories the engines know you'll like (if they exist).
Contrary to my expectations, Digg is taxonomy+graph DB based. Backend in Python.
Your recommendations for different topics are stored separately so you don't just have an aggregate recommendation score, because people suck at recommending some things and are good at recommending others.
Recommendations can be based on similar users - they fill the gap between people you know and the great unwashed - this is what services like last.fm do.
- this is old ground, can't imagine it's really new to anyone in the audience. The traditional problem here is never showing you anything new. A good recommendation recently was injection on news coverage that you don't normally look at to ensure you see content you normally wouldn't.
Invented their own graph-based backend!
Authentication, reputation and trust.
Shows that this isn't a solved problem and there's probably a lot of money to be made. Recommendation is the search of the noughties.
Show the impact you have - when you mod up or down, let the user know how many people they've affected.
Recommendation API which news services could integrate to show stories the engines know you'll like (if they exist).
Contrary to my expectations, Digg is taxonomy+graph DB based. Backend in Python.
Your recommendations for different topics are stored separately so you don't just have an aggregate recommendation score, because people suck at recommending some things and are good at recommending others.
