Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [jdt-ui-dev] Proposal Relevance System

I’m working in content assist proposals atm. The (future) findings may be applicable to quick fix too.

We’ll collect a few completion statistics (what was applied when, where, proposal rank and the like) and train an optimized ranking system.

I’ve read that you want to sort your quickfix proposals by the likelihood that a proposal fixes the problem. I haven’t read how you’ll determine the likelihood. Is this something you’d like to “learn”? If so, you may consider sending some quickfix events (e.g., using the IEventBroker) we can track and actually learn what was relevant and provide you the missing numbers.

Best,
Marcel


On 14.08.2012, at 12:40, Deepak Azad wrote:

Marcel, are you talking about relevance for quick fix proposals or content assist proposals?

Deepak

On Tue, Aug 14, 2012 at 10:27 AM, Marcel Bruch <bruch@xxxxxxxxxxxxxxxxxx> wrote:
Hi,

I’m currently experimenting with a bunch of new relevance computers that take into account different context information to put the most relevant proposals on top.

The coded relevance computation makes it (sometimes) hard to see how the different base relevances and relevance bonuses contribute to the overall relevance. Is there any external (non-code) documentation that captures some of your thoughts you made when setting up the relevance computation?


Thanks,
Marcel

--
Eclipse Code Recommenders:
 w eclipse.org/recommenders
 w about.me/mbruch

_______________________________________________
jdt-ui-dev mailing list
jdt-ui-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/jdt-ui-dev

_______________________________________________
jdt-ui-dev mailing list
jdt-ui-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/jdt-ui-dev

-- 
Eclipse Code Recommenders:
 w eclipse.org/recommenders
 w about.me/mbruch


Back to the top