Forum search kinda blows
by Orion Elenzil · in Site Feedback · 07/21/2006 (8:41 am) · 60 replies
Sorry for the harsh subject, but there's a ton of really valuable information in the forums which is really difficult to find because search is so hit-or-miss.
here's an example,
an article posted a month and a half ago to the TGE Private Forums. (sorry, non-sdk owners)
the article is titled "How to skip transmitting datablocks altogether", and has both the words "transmitting" and "datablocks" in the body.
so if you search for "transmitting datablocks" by just typing in the two words (without quotes) into the little yellow "search" field at the upper right, you get a page with a few hits, none of which are the page in question. even if you click the "show all results" link, it's still not found.
it's hard to imagine a better-conditioned target page.
variations:
- adding plusses seems to have no effect
- adding quotes reduces the results to two, neither of which are the target.
so i'm not sure what the deal is.
is google just not scraping the forums well enough ?
does google not have a license to the TGE private forums ?
is google scraping with an extremely low frequency ?
are post subjects included as a searchable field ?
is this just forums, or is there similar spotty-ness with TDN, resources, blogs, etc ?
here's an example,
an article posted a month and a half ago to the TGE Private Forums. (sorry, non-sdk owners)
the article is titled "How to skip transmitting datablocks altogether", and has both the words "transmitting" and "datablocks" in the body.
so if you search for "transmitting datablocks" by just typing in the two words (without quotes) into the little yellow "search" field at the upper right, you get a page with a few hits, none of which are the page in question. even if you click the "show all results" link, it's still not found.
it's hard to imagine a better-conditioned target page.
variations:
- adding plusses seems to have no effect
- adding quotes reduces the results to two, neither of which are the target.
so i'm not sure what the deal is.
is google just not scraping the forums well enough ?
does google not have a license to the TGE private forums ?
is google scraping with an extremely low frequency ?
are post subjects included as a searchable field ?
is this just forums, or is there similar spotty-ness with TDN, resources, blogs, etc ?
About the author
#2
07/21/2006 (10:45 am)
@Orion: I have noticed the exact same thing. It seems as if the google index needs some tweaking.
#3
07/21/2006 (11:21 am)
Same thing here.
#4
07/21/2006 (1:06 pm)
It isn't perfect, but I'm still recovering from the pre-google days...
#5
07/21/2006 (2:11 pm)
Heh, true dat.
#6
--Rick
07/21/2006 (3:37 pm)
@google-mini ...humm, interesting. The google Mini is a black box (actually google blue) with only a few knobs. It indexes everything every night at 2am PST. It decides exactly what to crawl/re-crawl and what to save in the index. It does have access to all the forums. It does highly value back links just like google proper and most forum threads don't have many back links. Did try a few other angles on the search but had no luck.--Rick
#7
07/21/2006 (4:15 pm)
Let me say this... it may not be perfect, but it could be much worse.
#8
07/21/2006 (4:20 pm)
IMO, the only improvement that would be handy on the current system would be a search filter.
#9
search for "dsprintf".
you'd think it would turn up this page, seeing as "dsprintf" turns up as a well-conditioned string in the body numerous times, and in the subject, of course.
or another example, search for "vsnprintf". again, it appears in the message body very well-conditioned:
the only way i was able to go back and find this post was to go and search thru the subject lines in my own profile, which is kind of a hassle as i've apparently now created 100+ threads, and if all i remembered was a keyword in the body and not the subject, i'd be hosed.
seriously, it's a bit broken, i think.
07/28/2006 (11:22 am)
Another example,search for "dsprintf".
you'd think it would turn up this page, seeing as "dsprintf" turns up as a well-conditioned string in the body numerous times, and in the subject, of course.
or another example, search for "vsnprintf". again, it appears in the message body very well-conditioned:
Quote:
..possibly vsnprintf didn't compile..
the only way i was able to go back and find this post was to go and search thru the subject lines in my own profile, which is kind of a hassle as i've apparently now created 100+ threads, and if all i remembered was a keyword in the body and not the subject, i'd be hosed.
seriously, it's a bit broken, i think.
#10
07/28/2006 (11:34 am)
Didn't the old search function allow you to specify whether to search only resources, forums or the entire site? That would be the best addition to the search as it is right now. That way I can stop getting pages and pages of .plans that have nothing to do with what I'm looking for.
#11
07/28/2006 (11:56 am)
You can do that now. Just click search button but leave the search field empty. You will be taken to a new screen where you can choose your search category.
#13
Thanks Mr. GG Webmaster.
07/28/2006 (12:09 pm)
The search categories have recently been updated. Plans (blogs) and resources used to share the same category, which did suck, but have now been separated. Thanks Mr. GG Webmaster.
#14
it is so frustrating that a search for "vista" does not return the single most relevant thread on the site:
www.garagegames.com/mg/forums/result.thread.php?qt=54666
that thread contains "vista" in both the subject and the first post in the body,
and has been up there for a month.
i was only able to find this thread because fortunately i'd marked it as "watched" back at the begining of december, so my email inbox had a bunch of updates in it.
c'mon guys. that's ridiculous.
01/03/2007 (3:21 pm)
Bump.it is so frustrating that a search for "vista" does not return the single most relevant thread on the site:
www.garagegames.com/mg/forums/result.thread.php?qt=54666
that thread contains "vista" in both the subject and the first post in the body,
and has been up there for a month.
i was only able to find this thread because fortunately i'd marked it as "watched" back at the begining of december, so my email inbox had a bunch of updates in it.
c'mon guys. that's ridiculous.
#15
I too check threads manually because of this, but it's a hassle in the long run.
01/03/2007 (3:31 pm)
I agree, I rarely use the search feature anymore because it's so bad.I too check threads manually because of this, but it's a hassle in the long run.
#16
01/04/2007 (8:30 pm)
This was brought up in the TGB forum recently too. I mentioned there that I really would like a search this forum and a search this thread option instead of having to add TGB + search keyword and hope it limits the search to just to the TGB forums. Also I too have noticed once in a great while keywords do not show up any results.
#17
01/05/2007 (12:34 pm)
I just searched for "ballistics" in all documents and only 4 were returned. But when I 'limited' the search to just forums, 27 were returned. ????
#19
this really ought to be fixed as it confuses everyone.
01/05/2007 (12:44 pm)
Just noticed that if I ask Google to display the "less relevant" ones from all document search, then the all documents search displays 38 matches. so it thinks they are more relevant if searched in the category in which they reside.this really ought to be fixed as it confuses everyone.
#20
On my forum, I can type in things in the search and find no results, but then I load up google.com directly, and search, and I can find things on my forum.
Also, there are some things google does not index at all, ever, for reasons unknown. I thought it was just the free invision boards they did this with, but apparently they do it to everyone.
01/20/2007 (8:01 am)
Google does that with ALL forums.On my forum, I can type in things in the search and find no results, but then I load up google.com directly, and search, and I can find things on my forum.
Also, there are some things google does not index at all, ever, for reasons unknown. I thought it was just the free invision boards they did this with, but apparently they do it to everyone.
Torque Owner Philip Mansfield
Default Studio Name