Search terms

[JoshEarl]

For some reason, the search tool rejects search terms that are "too short."

I've tried searching for things like "jaw" and "boss" when I was looking for specific info, and the search tool just won't do it.

Can we change this, or is there some really good technical reason why it has to be this way?

Thanks for all the hard work you do to keep SRP humming along...

Josh

[azjoe]

For some reason, the search tool rejects search terms that are "too short."

I've tried searching for things like "jaw" and "boss" when I was looking for specific info, and the search tool just won't do it.

Can we change this, or is there some really good technical reason why it has to be this way?

Thanks for all the hard work you do to keep SRP humming along...

Josh

Well the short answer is yes it can be changed, yes there's a good technical reason why, and no we're not going to change it.

The long answer is that the system is set to only search for words which are 4-characters long or longer. So "boss" should have returned a result (and did so when I tried it). But "jaw" won't. The number 4 is easily changed... it's a configuration parameter (the default is 4, btw). But the ramifications of changing it are great, particularly to a smaller number, eg 3.

In lay terms, the way the vBulletin search works is that it stores each unique word that's 4-characters or longer in a table and then creates an index entry for EACH occurrence of that string in another table. So "boss" has an entry in the the word table and each occurrence of "boss" has an entry in the index table.

The index table would grow significantly (and hence slow down searching and increase the server load and memory use) if we were to require 3-character words to be indexed... that would mean every occurrence of words like "the", "and", "etc" and so forth would get an index entry. Eg, I've used the word "the" about 10 times so far in this paragraph alone, so that would require ten index entries if 3-character searching were enabled. There's 75,733 posts on the forum as I'm writing this... imagine how many occurrences of "the" there must be in all of those... and every one would require an entry in the index table. Now, think about how many other 3-character words and abbreviations there are... thousands, probably. So there has to be a limit and empirically 4 seems to be the best choice per the guidance the forum software people have given.

[JoshEarl]

Joe,

Fair enough.

One other question: Would it be possible to set the indexing software to just exclude certain words? Seems like maybe a short list--maybe 30 to 50 common words--would cut out most of the junk.

If not, I'll just live with it.

Thanks for the quick reply.

Josh

[azjoe]

Joe,

Fair enough.

One other question: Would it be possible to set the indexing software to just exclude certain words? Seems like maybe a short list--maybe 30 to 50 common words--would cut out most of the junk.

If not, I'll just live with it.

Thanks for the quick reply.

Josh

How bout a compromise... use a wildcard? That way you can make a 3-letter term a 4-letter. Do a search on "*jaw" or "jaw*". You'll get a few more results than if you could search on just "jaw" but it's not bad... and sometimes it actually helps (eg, jaw* picks up "jawline" in the search).

[JoshEarl]

Joe,

That's a great solution.

I wish I could remember the other search term I'd had trouble with--I guess it wasn't "boss." (Which was a reference to a barber hone I have.)

I suppose there really aren't that many three-letter shaving terms out there...

The search engine is pretty decent in every other respect. No complaints here.

Thanks,
Josh

[b0000urns]

For short search terms you can also use google.
Example: pin site: