A question about recall and snippet assignment method selection

classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

A question about recall and snippet assignment method selection

seyfullahd
I want to consult you something about the selection of snippet assignment strategy.

Let's assume that, we followed a modified algorithm and get much higher recall values (on average 20 percent increase for instance and so close to 1), but relatively lower precision values. In such a case, since our real purpose is increasing recall first and also precision later then, is that reasonable to change the snippet assignment algorithm so that it assigns only the snippets that provides the higher recall values.

I mean when we change the algorithm as I told, the precision becomes always 1 and recall is the same as we get before.

Is this, changing the assignment strategy according to case of the highest recall result, reasonable?

Thanks in advance,

Seyfullah
Reply | Threaded
Open this post in threaded view
|

Re: A question about recall and snippet assignment method selection

seyfullahd
Maybe I'm right, but only if it is possible :)
Reply | Threaded
Open this post in threaded view
|

Re: A question about recall and snippet assignment method selection

Dawid Weiss-2
In reply to this post by seyfullahd
I don't think I follow what you're trying to achieve. In any case,
balancing recall and precision is probably an art more than science.
It'll depend on your target audience, what their expectations are, how
committed they are to sifting through results/ clusters to get to the
right answer. For the general search engine public I'd say precision
and ranking matter more since most people will only look at the
topmost search results (similar case applies to cluster labels).

Dawid

On Sun, Jul 14, 2013 at 10:11 PM, seyfullahd <[hidden email]> wrote:

> I want to consult you something about the selection of snippet assignment
> strategy.
>
> Let's assume that, we followed a modified algorithm and get much higher
> recall values (on average 20 percent increase for instance and so close to
> 1), but relatively lower precision values. In such a case, since our real
> purpose is increasing recall first and also precision later then, is that
> reasonable to change the snippet assignment algorithm so that it assigns
> only the snippets that provides the higher recall values.
>
> I mean when we change the algorithm as I told, the precision becomes always
> 1 and recall is the same as we get before.
>
> Is this, changing the assignment strategy according to case of the highest
> recall result, reasonable?
>
> Thanks in advance,
>
> Seyfullah
>
>
>
> --
> View this message in context: http://carrot2-users-and-developers-forum.607571.n2.nabble.com/A-question-about-recall-and-snippet-assignment-method-selection-tp7578213.html
> Sent from the Carrot2 Users and Developers Forum mailing list archive at Nabble.com.
>
> ------------------------------------------------------------------------------
> See everything from the browser to the database with AppDynamics
> Get end-to-end visibility with application monitoring from AppDynamics
> Isolate bottlenecks and diagnose root cause in seconds.
> Start your free trial of AppDynamics Pro today!
> http://pubads.g.doubleclick.net/gampad/clk?id=48808831&iu=/4140/ostg.clktrk
> _______________________________________________
> Carrot2-developers mailing list
> [hidden email]
> https://lists.sourceforge.net/lists/listinfo/carrot2-developers
>

------------------------------------------------------------------------------
See everything from the browser to the database with AppDynamics
Get end-to-end visibility with application monitoring from AppDynamics
Isolate bottlenecks and diagnose root cause in seconds.
Start your free trial of AppDynamics Pro today!
http://pubads.g.doubleclick.net/gampad/clk?id=48808831&iu=/4140/ostg.clktrk
_______________________________________________
Carrot2-developers mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/carrot2-developers