Search limitations on new forum software

[LCHIEN]

The threads on HF 2HP DCs are legendary in their length and detail.
But they are now almost impossible to find.

Why? because the new Search has some interesting limitations.
When you do a Search on HF 2HP DC assembly the little fine print says
"HF, 2HP and DC are ignored because they are too common or too short".

Needless to say only a few applicable hits occurred and lots of extraneous "assembly" of everything else were tagged and had to be sorted thru. With 4 very specific search terms I should have gotten modest number of very close-matching results.

Perhaps that (short words) might be a problem with normal discussion groups but us with our common acronyms in very wide use won't be able to find topics dealing with things like BS, TS, SMT CMS etc.

I think the elimination of short words should be limited to words like
to, for, the, of, is, etc. or the search function will be severely limited and we will lose a lot of good posts.

[Xamu]

Hi Loring,
I feel your pain, man. From what I know about this software, it builds an index of virtually every word in every post - dirty words can be filtered before indexing. Makes searching pretty fast.

Since 3 letter "words" are not typically that useful, the index only starts at (I believe) 4 letter words. That said, the software does not make a distinction between three letter "words" and three letter "acronyms."

Short of modifying the actual forum software there's not a whole lot I think Sam can do about it. And if he did, I believe the site would take a performance hit and the table with the index would be massive.

[Ken Massingale]

Loring, I did a little testing on SawMill Creek which uses the same software and find the same isues.
Worse maybe. In their General Woodworking forum there is a recent thread which contains "HD 10% Weekend". An Advanced search returns only one result, Creeker's Weekend Accomplishments....
The search threw out the HD and 10% and only queried on Weekend.
Hopefully there is a solution.
Ken

Just saw this on the vBullitin site, http://www.vbulletin.com/forum/showt...in+your+search

[Sam Conder]

Yeah, it would bloat the DB badly. We're gonna have to live with this limit.

[Ken Massingale]

Yeah, it would bloat the DB badly. We're gonna have to live with this limit.

To clarify Sam, I don't have a problem with this. Overall vBulletin is a vast improvement. IMO users will have to be educated and acclimate to not using abbreviations or live with the results, or lack of.

A thought, is there Admin capability to do a replace in the database? Replace HD with Home Depot, HF with Harbor Freight, DC with dust collector?
ken

[Sam Conder]

To clarify Sam, I don't have a problem with this. Overall vBulletin is a vast improvement. IMO users will have to be educated and acclimate to not using abbreviations or live with the results, or lack of.

A thought, is there Admin capability to do a replace in the database? Replace HD with Home Depot, HF with Harbor Freight, DC with dust collector?
ken

Hehehe... I am a bit selective on global find and replace actions since the Craftsman Torque Wrench debacle.

[LCHIEN]

Yeah, it would bloat the DB badly. We're gonna have to live with this limit.

I really think it severely limits the usefulness of searches. Wish there was some way around it. I have always felt the post base is a rich resource. With a good search tool it can be exploited rapidly in answer to someones question even if they fail to search themselves. Knowing a few key words from memory- that such a post exists - can help a lot.
But if the search tool won't take key words we commonly use, then
it'll be like looking at 100 shoe boxes filled with index cards and no sort order... I think the thread dates have been monkeyed with, at that, in the DB rebuild.

One big problem is, you don't even notice the small print that keywords were ignored. At least I didn't until I could not understand the search results relevance to the search terms. I tried again and then saw it. The number of people who searched before and said they couldn't find anything will be tripled.

I sure hope something comes to you. I sure don't like living with it.

[tfischer]

Just another vote that I hope we can find a workaround. We use more 2- and 3-letter acronyms around here than many forums, with all the tool and store abbreviations and the like...

Much of that now becomes unsearchable...

Is there possibly a way to have a list of allowable 2- and 3-letter terms that do get indexed? That would still screen out the myriad of other short words and not cause too much bloat...

Just a thought,
-Tim

[Stormbringer]

One big problem is, you don't even notice the small print that keywords were ignored. At least I didn't until I could not understand the search results relevance to the search terms. I tried again and then saw it.

Loring, don't feel bad. I took me about eight attempts before I saw the fine print that "Jet" was too short.

Not all two or three words that we would search for are acronyms. here's my example.

I know , since I remember reading it awhile back, that someone posted a review on the new "Jet bar clamps". Entering that returns a gazillion threads since "Jet" and "Bar" are too short to be included in the search. You can only imagine how many hits came back with the word "clamp" associated with it.

So, if at all possible, here's another vote for refining the search capabilities. If the site had to go down for a day or two, I could live with that.

Thank you,
Greg

[dlminehart]

I agree that the search problem is a serious one. The best work around I've been able to come up with is to not bother using the one within vBulletin. Instead, use google's. If you want to look for HF DC, for instance, enter the following into a google search:

site:www.bt3central.com HF DC

Give it a try. I'd think that, if google can index and quickly search the entire worldwide web, vBulletin could index and search a forum!

[JR]

Great tip, Dave! That's definitely the workaround we needed.

I don't know about comparing vBulletin to Google like that, though. Google's got the special sauce, man! Expecting vBulletin to compare is a bit much to ask.

JR

[Ken Massingale]

I agree that the search problem is a serious one. The best work around I've been able to come up with is to not bother using the one within vBulletin. Instead, use google's. If you want to look for HF DC, for instance, enter the following into a google search:

site:www.bt3central.com HF DC

Give it a try. I'd think that, if google can index and quickly search the entire worldwide web, vBulletin could index and search a forum!

-----------------------------------------------------------------------
It's not that vBulletin can't do the indexing, it's the gloated database that would result. This 'feature' is discussed in many threads in the official vBulletin forum.
Ken

[Sam Conder]

-----------------------------------------------------------------------
It's not that vBulletin can't do the indexing, it's the gloated database that would result. This 'feature' is discussed in many threads in the official vBulletin forum.
Ken

Exactly. Making this comparison is not valid because BT3Central has one small server running both the web server and the SQL database server. It is a small Pentium 4 1.7GHz with 1gb of RAM. I have no idea what Google has but I would say there are literally hundreds if not thousands of servers involved.

[uglystick]

Sam,

There are ways to add the Google search directly on the site (called Google Free?)

There's an example here on the left side of the page.

Just a thought! Trying to give people options...

[dlminehart]

To follow up on Kendall's post, see http://www.google.com/searchcode.html for how to embed free google site search in one's own site.