Boosting popular news


(Eugene Strokin) #1

Hello, I remamber Brandon Sterne announced Social News aggregator,
which is based on ES, while ago. I have a question about boosting
popular news. I have similar task for my project, I need to boost some
documents based on user's like/dislike actions in last N days.
Could you give me some advice where should I start? And if this advice
will be on some real application (like sharengeti.com) example it
would be double grate)))
In general I've started by creating a new document type called Like,
which would have Document ID and User ID, and date when Like is
happened. Now I can store and index Likes of my users.
But now I need to count likes of my documents from last N days, and
use the number as a boost factor. I prefer to do this on a fly, but I
don't know how, if possible at all. Could you please give me some
advice here?
In worst case I was planning to have some kind of job, which would
calculate the likes hourly and update a document with the number, but
this would make relevantency of search not real time, and updating
documents each hour is really what I don't want to do.

Thanks in advance,
Eugene S.


(system) #2