nms-cgi-devel

Re: [Nms-cgi-devel] Simple Search problems

[Olivier D. <dr...@sh...>]

On Thu, Jan 31, 2002 at 11:07:27PM -0500, Joseph F. Ryan wrote:
> I was fixing a typo in the simple search config description when I decided 
> to give it a try.  It didn't work.  Detaint_dirname killed the script 
> (there were no directory names which should have killed it), and I couldn't 
> get the recursion to work with ('*.html, */*.html, */*/*.html, 
> */*/*/*.html).  Did we decide on another method?    Perhaps we should have 
> recursion on by default if $Emulate_Matts_Code is set to 0, and keep the 
> old method if it is set to 1.

I wasn't there when the scripts were implemented but crashers are
definetly issues that need to be addressed. ('*.html, */*.html, */*/*.html, 
*/*/*/*.html) ... is that just a typo or are you missing a ' before the
closing bracket?

> Also, I'd like to implement sorted results if $Emulate_Matts_Code is set to 
> 0.

If you can do better than google :o) Just kidding. Sorted results would
be AWESOME! What would you base the sorting on? I've tried to implement
a sort myself for search results and it's pretty hard to get it right.
There is more to take into account than just the frequency of keywords
in a document (although that's a pretty good start -- for 1 word search
it's pretty much the only thing to do, for multiple words you have two
keywords or more to count frequency -> which will give better results?
10-1 or 4-4? Also as the google theory says, how many words in between
those two words -> closer = better. *insert your favorite sorting method
here*, etc, etc ... can get pretty complicated if you really want it)

-Oli

-- 
+----------------------------------------------+
| Olivier Dragon        dr...@sh... |
| Software Engineering II, McMaster University |
+----------------------------------------------+

Re: [Nms-cgi-devel] Simple Search problems

[Olivier D. <dr...@sh...>]

----- Forwarded message from "Joseph F. Ryan" <rya...@os...> -----

>From oli  Fri Feb  1 08:20:13 2002
Envelope-to: oli@localhost
Delivery-date: Fri, 01 Feb 2002 08:20:13 -0500
To: Olivier Dragon <dr...@sh...>
From: "Joseph F. Ryan" <rya...@os...>
Subject: Re: [Nms-cgi-devel] Simple Search problems

At 11:48 PM 1/31/2002 -0500, you wrote:
>On Thu, Jan 31, 2002 at 11:07:27PM -0500, Joseph F. Ryan wrote:
>> I was fixing a typo in the simple search config description when I decided
>> to give it a try.  It didn't work.  Detaint_dirname killed the script
>> (there were no directory names which should have killed it), and I 
>couldn't
>> get the recursion to work with ('*.html, */*.html, */*/*.html,
>> */*/*/*.html).  Did we decide on another method?    Perhaps we should have
>> recursion on by default if $Emulate_Matts_Code is set to 0, and keep the
>> old method if it is set to 1.
>
>I wasn't there when the scripts were implemented but crashers are
>definetly issues that need to be addressed. ('*.html, */*.html, */*/*.html,
>*/*/*/*.html) ... is that just a typo or are you missing a ' before the
>closing bracket?

it was just an example.  it is actually: ('*.html', '*/*.html', 
'*/*/*.html', '*/*/*/*.html')



>> Also, I'd like to implement sorted results if $Emulate_Matts_Code is 
>set to
>> 0.
>
>If you can do better than google :o) Just kidding. Sorted results would
>be AWESOME! What would you base the sorting on? I've tried to implement
>a sort myself for search results and it's pretty hard to get it right.
>There is more to take into account than just the frequency of keywords
>in a document (although that's a pretty good start -- for 1 word search
>it's pretty much the only thing to do, for multiple words you have two
>keywords or more to count frequency -> which will give better results?
>10-1 or 4-4? Also as the google theory says, how many words in between
>those two words -> closer = better. *insert your favorite sorting method
>here*, etc, etc ... can get pretty complicated if you really want it)

Well, since it is simple, I was just planning on sorting by hits, nothing 
as complex as what I have for advanced search (which reminds me, I need to 
work on that some more)

----- End forwarded message -----

-- 
+----------------------------------------------+
| Olivier Dragon        dr...@sh... |
| Software Engineering II, McMaster University |
+----------------------------------------------+