Either attack that via custom score or a filter query which includes
all friend and friends of friends. For jetslide / jetwick(open source)
I've implemented this that way.
But as Paul noted this can become quite big. A solution would be to
attach a marker array "forUsers" to every document. This way you have
hopefully less big arrays in the documents and one a simple term
filter while searching. You could also do some boosting e.g. add
"twitterUserName":9 to the document and then do a sort or boost
against that field...
On 2 Dez., 16:47, Paul Loy ketera...@gmail.com wrote:
that sounds like possibly a custom score:http://www.elasticsearch.org/guide/reference/query-dsl/custom-score-q...
But be careful, friends of friends can become pretty big pretty quick
On Fri, Dec 2, 2011 at 7:39 AM, Benoit Domingue bedin...@gmail.com wrote:
I'm kinda new to elasticsearch, but so far, it's been a really good
Now I i'd like to boost results that according to a social graph...
For example, if some content is from my friend or my friend's friend,
I'd like to boost it.
How would that be possible using elasticsearch ?
Would I need an index for each users ?
How would you guys set the mapping ?
Thanks a lot !