Admin: Search Word Length

[SLang]

Tried a search awhile ago... you have your search criteria limited to more than 3 letters. Though I see the point of this to max. search resources, it does cause some conflicts.



When searching for common diesel things such as "steering box", "max power", "max egt", "#11 plate install", etc it becomes bothersome. How much would it hurt to back the limitation down to more than 2 letters. I could also see this being a problem with "max HP", etc. But it's better than now.



Good site just trying to keep you up on the little thing you do not have time to test.

 

[Archive]

I agree. The search is too restrictive this way. I would like to see it changed to less words.

 

[Joe G.]

It doesn't support quoted strings either. I tried that. If it did you could enter "Max Power" and the search would be for a nine character string. It appears that the search gets its speed from only considering whole words in a search. When I tried the above it complained the Max had only three characters. I could certainly stand a slower search if it would find what I want. The search speed should not be such an issue that capability is lost.

 

[paccool]

I agree with the above. Maybe I'm doing something wrong, but had better luck with the old search engine.



Stan

 

[admin2]

Search

The search gets its speed because it is database driven. Really has nothing to do with the way the words are put together.



The most likely reason you're not coming up with good results is the search engine defaults to searching posts less than a day old and to one forum only. The old system defaulted to all threads in all forums. If you expand your search to include all dates and all forums you're likely to get better results.



The system will not go to a 3 letter minimum or less. The search index would get huge. You need to learn the current system using the wildcard feature. To search for "oil", use "oil*". T



The current search is far more powerful than the old but unlike the old, you have to use logic (ie, boolean logic) to search.



Examples, to search for Bob's truck search for "Bob's*truck" or "Bob and truck". The old system would take your search literally and could not use wildcards or boolean logic. While it had its pluses, it was very limited when it comes to complex searchs.



Ken

TDR Admin

 

[rickhans]

Where does one go to find all of the rules and special characters required to do boolean searches. Although I have used boolean equations extensively, I am not familiar with using * for "and". I would like to read a complete set of search instructions for this type of search system.

 

[admin2]

Boolean

The VBulletin boolean search options are listed on the search screen:



Advanced query: Join words with AND, OR and NOT to control you search in more detail.

Add asterisks (*) to use wild cards in your search (*bullet* matches vBulletin etc. )



Examples:



If you want to search for posts containing oil and synthetic but you don't want posts about Amsoil:



oil AND synthetic AND NOT amsoil



If you want a search for posts about Dodge or Ford diesel engines but not Chevy engines:



dodge OR ford AND diesel AND engine AND NOT chevy



If you want to search for posts with Bob, Bobby or Bobet:



bob*



If you want to search for motor oil (getting around the 4 character limit for oil):



motor oil*



OR



motor AND oil*



As to the search limitation of 4 characters... .



The current search index file is over a gigabyte in length. For every character smaller, it grows expontential. Imagine its size if the search index file included every occurance of "of", "the", "etc", "will", "for", "it", "how", "is". For every word occurance in the index, 35 bytes of storage is used. Just the word "the" would take over 16. 8 megabytes drive space (and RAM during searches).



MySQL, the database used by VBulletin, does not support row level locking. In laymens terms, that means that while a search is going on, all posts and edits come to a screeching halt until the search is done. Waiting 5 seconds for your post to complete isn't a big deal but how many of you would raise the dead screaming if it took 3-5 minutes to post/edit?



Lets put the 4 character limit to rest, its not going to change unless someone has a $50,000 database server to donate.



Ken

TDR Admin

 

[Joe G.]

Most search engines I've worked with ignore common articles and other "punctuation" words like the, and, etc, will, for, ... . If I understand the way this works then you can search for words like since, which, otherwise, after, other, like, more, thanks, ignore, ... . That doesn't make any sense at all. For example, this reply consists of a lot of words longer than three characters that no one in his right mind would want to search for. If this search engine does not search for these sort of words then adding real words like oil would not be a big deal in storage.

 

[admin2]

Re: Boolean

Originally posted by admin

Lets put the 4 character limit to rest, its not going to change unless.....

 

[Joe G.]

Re: Boolean

Originally posted by admin





As to the search limitation of 4 characters... .



The current search index file is over a gigabyte in length. For every character smaller, it grows expontential. Imagine its size if the search index file included every occurance of "of", "the", "etc", "will", "for", "it", "how", "is".



Ken

TDR Admin

So the index only contains four character words or longer. If that is the case then how in the world can a wild card search find anything like oil with oil* if oil isn't in the search index to start with?



Then the search locks up the database during search???? Sounds like a really lousy choice of a search engine to me! If sure looks like user capability and usability have been sacraficed for ease of programming. More user hostile software.

 

[admin2]

Put it to rest...

The search locks the search index table, not the database. The search is part of the VBulletin package so we have to live with it. Its MySQL, the database that VBulletin uses, that causes the table to lock.



No amount of debate is going to change the fact that the search has a 4 character limit.



Ken

TDR Admin