Amiga.org
Coffee House => Coffee House Boards => CH / Science and Technology => Topic started by: mr_a500 on April 18, 2006, 05:19:59 PM
-
Is it my imagination or are search engines getting dumber?
Google Advanced Search "Find with the exact phrase" returns results not containing the exact phrase. An example, entering a phrase like "this is it" will return over 9000000 entries, but only about 20 contain the actual phrase "this is it" - the rest may have the word "this" or "is" or "it" which is NOT the exact phrase (this is just an example - I'm not searching for "this is it" ;-)).
Another annoying thing is searching for something with non-alphanumeric character like "stuff.text" or "Airplane!" or "$45" (again, just examples). Instead of a list of the exact thing, the special character is ignored or when in the middle, text is split into two words (stuff.text becomes "stuff" "text").
I searched for the game "One on One" on Planetemu today as a test and it returned 20 pages of every entry containing "one" somewhere in a word in the title! (words including "Bone", "Gone", "Money")
Even on Amiga.org, I searched for "filemanager" and it returned every post with the word "file" and "manager". No, I asked specifically for "filemanager".
Maybe the problem is that search engines are trying to be too clever - but instead you end up with dumb results.
-
Eh? (http://www.google.se/search?hl=sv&q=%22this+is+it%22&meta=)
-
If you look carefully, you'll notice that on the first page, hits #3,4,9 and 10 do not contain the search string "this is it" (check on the pages). If you look at search page 2, most of them do not contain "this is it".
If the search worked the way it should (only the actual string), then all the pages without the string should not be there.
I found even more "search stupidity" today. On EAB, if you put 2 words in the search, it does an "OR" instead of an "AND". That makes it bloody impossible to find something containing only both words.
-
Interesting point. I noticed that when I am searching for technical data, the very argument that you present is true.
So, here's a question: what search engine is the most efficient at returning correct strings?
Miked
-
So, here's a question: what search engine is the most efficient at returning correct strings?
I don't know but I'd certainly like to know. It's bloody impossible to search for a simple string when you have to sift through thousands of unrelated pages that don't even have the damn string.
Hmm... maybe I'll search for "bloody search engine that actually works properly". It'll probably return pages about "bloody body discovered - worker run over by steam engine - harness not actually attached properly".
-
While I agree that Google seems to be getting less relavante with a lot of the results from my searches, I don't think "this is it" is a good string to search for. Anytime I have used the word "and" in a search string, I am told in the results "and" is a common word and won't be used in the search... it seems to me that "this", "is", and "it" are all pretty commonly used too, so maybe Google's engine is confused on exactly what to show you, even if you do select the "exact phrase" option.
But, there have been several times in the last couple months where I have wanted to show somebody something that I didn't bother to bookmark and that I was easily able to find over and over in Google just by doing a search. Next time I looked for it nothing close come up again. I know I searched the exact terms again because FireFox's cache showed me it as an autocomplete in the field. Couldn't for the life of me fine any of it on Google. Went to Yahoo which I hadn't don in ages... first search, first page listed. Go figure.
-
My point was that it shouldn't matter what is in the exact string, but it should be taken exactly as typed and only return results containing the string exactly as typed. If they're going to break it up and decide what is common or not, it is not exact. I should be able to search for "jalv2^B*6-* and the -&_M78NB!@Cx`" if I want. Otherwise they shouldn't say "Find with the exact phrase".
It's the opposite of the way it used to be. Remember the old days when you tried to search for something and if you just put one too many spaces it wouldn't find it? Yes, that's a pain in the ass when you want to search for something "similar to what you typed", but it's very useful if you know only one page in a million has it EXACTLY as you typed.
Modern search engines should also be smart enough to allow you to search with strings containing special characters and not just be limited to alphanumeric. (come on, it's just text - not rocket science)
-
JetFireDX wrote:
Went to Yahoo which I hadn't don in ages... first search, first page listed. Go figure.
I am no longer a fan of google. They used to return very relavent searches. I agree with you- I am becoming a fan of Yahoo once again.
-Miked
-
mr_a500 wrote:
Modern search engines should also be smart enough to allow you to search with strings containing special characters and not just be limited to alphanumeric. (come on, it's just text - not rocket science)
Quite...I'm annoyed by how google handles ø,å,æ,ö,á etc.
-
mr_a500 wrote:
My point was that it shouldn't matter what is in the exact string, but it should be taken exactly as typed and only return results containing the string exactly as typed. If they're going to break it up and decide what is common or not, it is not exact.
I agree with you there, but perhaps using common words like that, even with "exact phrase" selected causes it to not do an exact search, it may be a bug. You might want to write them a report about it. It couldn't hurt afterall.
Another feature to be able to search with would be a tolerance setting for how close the words you searched for should be in a page... what good is it doing a search for some terms when they are used in 5 different unrelated places in a page spread out over 20 or 30 posts in a forum? Not very relevant. Something like "Find X,Y, and Z within 1 paragraph". I also wouldn't mind if the page results took more than 0.23 seconds to return all of the relevant hits. It's okay, take a full second and do a better job at searching. Using a second to find me better results is going to save me time over getting crap back in 0.23sec and having to do more searching.
-
Another feature to be able to search with would be a tolerance setting for how close the words you searched for should be in a page... what good is it doing a search for some terms when they are used in 5 different unrelated places in a page spread out over 20 or 30 posts in a forum?
Exactly! I was having major problems just yesterday searching for multiple words (using "+" to try to make sure only pages with all words come up) and I got lots of pages containing those words but in totally unrelated places.
Example: a search for a piece of baroque music by composer, piece number, orchestra, conductor, musician and other info to specifically identify a recording returned lots of huge webpages (600K-3Mb!) with unrelated hits all over the page - like right conductor, wrong orchestra and piece; right orchestra and musician, wrong composer; right piece number, everything else wrong... and so on. I never did find that damn recording.
"Search terms within same paragraph" or something similar would certainly have helped.
-
Google really *IS* starting to suck bad.
I'm not even talking about the crap and fluf that find's it's way into the results, but simple text searches that used to return 10's of thousands of hits won't return anything anymore.
Also, I frequently need to search for strings where "", *, and | arn't enough to deal with the variations I need to express.
Is there a search engine that has better pattern and string tools?
Here's a search that used to routinely return 50,000+ results, now only returning 5.
"haven't|not been|managed * to make|get this|that thing * yet"
-
Yes, Google is making some changes. Has anyone noticed - in the last few days, Google made a change to the links returned from searching. Now, they attach a refer string to the link. I hate this. This is one of the reasons I switched from Altavista to Google in the first place. Altavista used to do this and it was extremely annoying because whenever they made a little change, bookmarked links would stop working (because of the damn refer string).
Why????
-
this search seems to work a little better
this is it (http://www.google.com/search?as_q=&num=10&hl=en&btnG=Google+Search&as_epq=%22this+is+it%22&as_oq=&as_eq=&lr=&as_ft=i&as_filetype=&as_qdr=all&as_nlo=&as_nhi=&as_occt=any&as_dt=i&as_sitesearch=&as_rights=&safe=images)
-
www.snap.com is nice :-)
-
www.snap.com is nice
Not from what I saw. I'm using a real Amiga on dialup. The snap.com site requires javascript, then loads a 275K page and 275K again with every search page. It didn't seem to find anything either.
-
use a "popular" browser on relatively powerful computer on broadband. gives you a nice preview of the page in the side