Page 1 of 1

[DONE] Search function limitations / common words list

PostPosted: Mar 16th, '12, 22:46
by doktor5000
I was just searching for something, and look what i got in return:

https://forums.mageia.org/en/search.php wrote:The following words in your search query were ignored because they are too common words: rolling release.
You must specify at least one word to search for. Each word must consist of at least 3 characters and must not contain more than 50 characters excluding wildcards.


What can be done about this? Can those two please be removed from this common words list?

Re: Search function limitations / common words list

PostPosted: Mar 17th, '12, 08:56
by wobo
There is no "list of common words". It's just that both words are used too often. If the number of results for a search word exceeds a certain limit the search function gives this error.
It's one of the shortcomings of the forum search. I already requested to implement "Google search for this forum" as we are using for MandrivaUser.de, but....

Temporary workaround: use Google extended with forums.mageia.org

Re: Search function limitations / common words list

PostPosted: Mar 17th, '12, 16:40
by doktor5000
What limit would that be, do you happen to know?

Re: Search function limitations / common words list

PostPosted: Mar 18th, '12, 07:07
by wobo
In MandrivaUser.de it is set to 5% (occurrence of search word in 5% of indexed posts), that's the default. I don't know if it is the same at Mageia (don't have access to Mageia forums ACP).
Admin can check in the ACP, I doubt that the default has been changed. In "General - Server Configuration - Search function".

Admin has 2 options:
1. re-create the index via ACP (normal maintenance procedure, but I doubt that it was ever done since the forum exists, I do this at MandrivaUser.de every 2 months (roughly) ) or
2. increase the value to 10% or even "0" (which means no limit at all).

Re: Search function limitations / common words list

PostPosted: Mar 22nd, '12, 11:40
by maat
It's already been set to 10% long ago... let's raise it to 30% then

By the way a few thoughts if i'm not mistaken:
1 - Option 2, iow changing threshold, needs also index re-creation to get words out of "common words list"
2 - Re-indexation is, only needed in case of parameter change, modding or updating with effect on search behavior, or heavy change in forum content (mass deletion for example).
3 - Using reindexation for search performance reasons is something rather common... but it's better dealt with at database level using defragmentation/optimization (with less effect on global performance for huge boards) and it will also depend of the database on which the board relies.

Please test if you like

Re: Search function limitations / common words list

PostPosted: Mar 22nd, '12, 12:03
by wobo
@1. Of course, this goes without saying.

BTW: I did not experience any effect on global performance during re-indexing. That there will be an impact on the global performance when the forum reaches a certain size is only logical, no need to test. But I admit that the forum I am doing this has only a couple of thousand registered users and around 270K posts, so it is not a large board. This one here isn't either.

Re: Search function limitations / common words list

PostPosted: Mar 5th, '13, 21:16
by doktor5000
I have to dig out this one, there is again a problem with the search function, i'Ve not yet encountered that with others forums, sorry :/
I've just searched for "systemd plot" (without the quotes) as i know it is at least contained in two posts (as i've posted it myself) and the normal forum search and neither the advanced search show ANY results. How can that be.

Doing a custom google search for "systemd plot site:forums.mageia.org/en" (again without the quotes) shows me the relevant threads.

Can we please get this fixed, and what is the reason for that?

Re: Search function limitations / common words list

PostPosted: Mar 27th, '13, 23:20
by doktor5000
Ping? Just stumbled about this again, seems we need an re-indexing more often. Searching for the string "dacentec" yields:

Code: Select all
Information

No posts were found because the word dacentec is not contained in any post.


Which is not correct, google search for "mageia forum dacentec" yields as first result, and already in the first post of the thread
viewtopic.php?f=23&t=4490

Re: Search function limitations / common words list

PostPosted: Mar 28th, '13, 00:09
by oj
Do quotes work here the same as they do in your 'standard' search engines, eg google etc?

Re: Search function limitations / common words list

PostPosted: Mar 28th, '13, 00:40
by Ken-Bergen
doktor5000 wrote:Which is not correct, google search for "mageia forum dacentec" yields as first result, and already in the first post of the thread
viewtopic.php?f=23&t=4490
As your post that I'm quoting is found in a search it would appear that the forum search doesn't look in code blocks.
I don't know if this is intended or a bug.

Re: Search function limitations / common words list

PostPosted: Oct 8th, '13, 21:29
by doktor5000
I've to dig this up one last time before beginning to ignore this completely.

Searching for canberra-gtk-module yields:
No posts were found because the word canberra-gtk-module is not contained in any post.
Google search via site:https://forums.mageia.org/en "canberra-gtk-module" yields 301 results, most of them containing the string without code tags.
From the 301 results, not one is found by the forum search function.

Can we please either add a custom google search, or fix the search function?

maat?

Re: Search function limitations / common words list

PostPosted: Oct 9th, '13, 00:35
by maat
doktor5000 wrote:I was just searching for something, and look what i got in return:

https://forums.mageia.org/en/search.php wrote:The following words in your search query were ignored because they are too common words: rolling release.
You must specify at least one word to search for. Each word must consist of at least 3 characters and must not contain more than 50 characters excluding wildcards.


What can be done about this? Can those two please be removed from this common words list?


Can you test if there is an ipmrovement please ?

Re: Search function limitations / common words list

PostPosted: Oct 9th, '13, 01:22
by Ken-Bergen
maat wrote:Can you test if there is an ipmrovement please ?
A definite improvement as a search for " rolling release" now returns 228 hits.

@ doktor5000
It would seem the search function doesn't like dashes in a word as
"canberra-gtk-module" returns no hits
but
"+canberra +gtk +module" returns 30 hits including this thread.

Re: Search function limitations / common words list

PostPosted: Oct 9th, '13, 21:17
by doktor5000
It got a bit better, although text inside code tags doesn't get parsed - can anything be done about this?

Re: Search function limitations / common words list

PostPosted: Oct 9th, '13, 23:32
by Ken-Bergen
doktor5000 wrote:although text inside code tags doesn't get parsed - can anything be done about this?
I'm not seeing that.

Re: Search function limitations / common words list

PostPosted: Oct 15th, '13, 12:08
by viking60
From the advanced search help text
Search for keywords:
Place + in front of a word which must be found and - in front of a word which must not be found.

That influences the use of "-"
Google search is really better than phpbb's no matter what you do (Yes they are spying - but they do deliver these annoyingly good results).
Image

Re: Search function limitations / common words list

PostPosted: Oct 15th, '13, 20:40
by doktor5000
viking60 wrote:Google search is really better than phpbb's no matter what you do (Yes they are spying - but they do deliver these annoyingly good results).

Yes, that's why we had a custom google search over at mandrivauser.de forum, as many users liked that better.
And as that will probably never be implemented in this forum *sigh* i've just created a smart search keyword bookmark-thingie for a google search for
%s site:https://forums.mageia.org/en

Works fine here.

Re: Search function limitations / common words list

PostPosted: Oct 16th, '13, 10:44
by viking60
Yup that works just fine here too - confirmed.
Code: Select all
site:https://forums.mageia.org/en google search

Brings me right back here :D
I have implemented it under the Advanced search in prosilver and directly in the header in subsilver2.

I am done commenting that you and other knowledgeable people do not get the rights to implement this simple stuff - it only triggers my Berserk temper Image