Mailing List Archive: kinosearch: discuss

wildcard searching

Index | Next | Previous | View Threaded

jscofiel at nd

Sep 14, 2007, 6:06 AM

Post #1 of 4 (1096 views)
Permalink

wildcard searching

I am having a problem getting wildcard searching to work. I understand
that it is implemented in Lucene, but could not find any good
documentation on its implementation in Kinosearch. I apologize if it is
readily available, just point me to it.

Basically I attempt to pass a query such as "Mar?" which in my
understanding should return things like "Mary" or "Mark". Instead, I
get no hits. It's the same with *. Any suggestions would be greatly
appreciated...

Thanks,

John

_______________________________________________
KinoSearch mailing list
KinoSearch [at] rectangular
http://www.rectangular.com/mailman/listinfo/kinosearch

matthew.oconnor at socialtext

Sep 14, 2007, 9:21 AM

Post #2 of 4 (1034 views)
Permalink

Re: wildcard searching [In reply to]

On Sep 14, 2007, at 6:06 AM, John Scofield wrote:

> I am having a problem getting wildcard searching to work. I
> understand that it is implemented in Lucene, but could not find any
> good documentation on its implementation in Kinosearch. I
> apologize if it is readily available, just point me to it.
>
> Basically I attempt to pass a query such as "Mar?" which in my
> understanding should return things like "Mary" or "Mark". Instead,
> I get no hits. It's the same with *. Any suggestions would be
> greatly appreciated...

KinoSearch doesn't have Wildcard searching. Socialtext implemented
very simple Wildcard searching, in the obvious way. It's slow and
icky, and usually stemming gives better results. If you really must
have it then I can point you at some code.

-matthew

_______________________________________________
KinoSearch mailing list
KinoSearch [at] rectangular
http://www.rectangular.com/mailman/listinfo/kinosearch

jscofiel at nd

Sep 14, 2007, 9:32 AM

Post #3 of 4 (1046 views)
Permalink

Re: wildcard searching [In reply to]

>
> On Sep 14, 2007, at 6:06 AM, John Scofield wrote:
>
>> I am having a problem getting wildcard searching to work. I
>> understand that it is implemented in Lucene, but could not find any
>> good documentation on its implementation in Kinosearch. I apologize
>> if it is readily available, just point me to it.
>>
>> Basically I attempt to pass a query such as "Mar?" which in my
>> understanding should return things like "Mary" or "Mark". Instead, I
>> get no hits. It's the same with *. Any suggestions would be greatly
>> appreciated...
>
> KinoSearch doesn't have Wildcard searching. Socialtext implemented
> very simple Wildcard searching, in the obvious way. It's slow and
> icky, and usually stemming gives better results. If you really must
> have it then I can point you at some code.
>
> -matthew
>
> _______________________________________________
> KinoSearch mailing list
> KinoSearch [at] rectangular
> http://www.rectangular.com/mailman/listinfo/kinosearch

Makes it hard to find when it doesn't exist! I think I saw something on
a socialtext list that mentioned wildcard searching. hmmm. I guess I'll
just play with the stemming and see what I think. If not, I may ping you
for some hacks.

Thanks,

John

_______________________________________________
KinoSearch mailing list
KinoSearch [at] rectangular
http://www.rectangular.com/mailman/listinfo/kinosearch

marvin at rectangular

Sep 15, 2007, 8:30 PM

Post #4 of 4 (1047 views)
Permalink

Re: wildcard searching [In reply to]

On Sep 14, 2007, at 9:21 AM, Matthew O'Connor wrote:

> KinoSearch doesn't have Wildcard searching. Socialtext implemented
> very simple Wildcard searching, in the obvious way. It's slow and
> icky, and usually stemming gives better results. If you really
> must have it then I can point you at some code.

KS doesn't have it in core, and I don't plan to add it. It's slow
and icky in Lucene, too -- compared with everything else. Because
people don't automatically guess that such would be the case or
understand why it's the case, some inevitably end up with unforeseen
performance challenges. Talking these people down on the mailing
lists consumes a fair amount of energy.

In absolute terms, wildcard searching can be quite quite useful and
performance can be acceptable for smaller collections. It would be a
nice thing to have on CPAN under the KSx namespace. As a separate
distribution, people would be psychologically prepared for behavior
unlike that of the KS core.

Marvin Humphrey
Rectangular Research
http://www.rectangular.com/

_______________________________________________
KinoSearch mailing list
KinoSearch [at] rectangular
http://www.rectangular.com/mailman/listinfo/kinosearch

Index | Next | Previous | View Threaded

Interested in having your list archived? Contact Gossamer Threads