rgbstock
Yet another SEARCH query from me - Sorry
1. fishmonk2 November 2010, 13:39 GMT +01:00
I typed in "customer" & was amazed to find lots of unrelated images, mostly cars.
On further inspection I could see 'custom' as being the keyword that flagged up these images.
When I searched for 'news', however, I got a different image to searching for 'newspaper'.
Does our search engine know that -er is a suffix?
2. lennie3 November 2010, 17:26 GMT +01:00
Yes, it, among other things, has rudimentary understanding of the english language. Most searchengines all use it at the basic level, it is called stemming. They also all seem to all use a variant of the same BM25 weighting scheme (BM sounds really fancy, but it is just short for Best Match). But that is a whole other story. :-)
They all need tuning.
Custom and customer is a very good example where it does not work. I'll see if I can put it on a blacklist.
Looks like I didn't have an exception list for this situation yet, I'll have to change a few things.
3. fishmonk3 November 2010, 21:40 GMT +01:00
Lennie, thanks for indulging my ignorance yet again.
That's great - now my results make sense to me. I had not realised that the search engines were so involved. I am more & more impressed with the site as time goes by - you have done a great job!
4. lennie4 November 2010, 9:33 GMT +01:00
Wow, slow down, I haven't actually fixed the problem yet. :-)
I think shortening useful to use might also be such a word.
5. Ayla874 November 2010, 9:58 GMT +01:00
helpful -help
wonderful -wonder
those ful-words may be full of surprises when you search for them :)
--Michael
6. lennie4 November 2010, 10:20 GMT +01:00
I've moved some stuff around, so I could make the change.
I think I'll do it soon.
Especially when I have enough guts to actually put the change online. ;-)
When I think I know what the real impact is.
It usually takes longer to do changes then you first expected.
Michael, I do think wonderful is a lot less worse then custom/customer, atleast wonder/ful is kind of related.
7. micromoth4 November 2010, 13:13 GMT +01:00
Unfortunately English is full of such pitfalls. Take "minute" for example. It can mean "very small" and it can mean "one-sixtieth of an hour". Same spelling. I would think it's impossible to weed out all the potential mismatches in the search results. The best one could achieve is to target the worst problems as they become apparent.
8. Ayla874 November 2010, 13:32 GMT +01:00
Kevin, you are absolutely right. I experience such problems quite often when I'm searching for images. This is not only a question of optimizing the search engine but also of giving tipps to everyone how to use it. To urge photographers to use thesauruses and use synonyms when describing their keywords can't be repeated often enough. Someone with better English than I should try to write a little FAQ on searching and finding, I'm sure this would be a great hit for the next newsletter.
--Michael
9. lennie4 November 2010, 18:00 GMT +01:00
We've been thinking of adding synonyms to the searchengine, but that is possibly a bigger project.
10. Ayla875 November 2010, 6:17 GMT +01:00
Lennie, sounds like a great idea. In the meantime, would it be easier to add automatic synonym suggestions in the photo management/tagging area?
--Michael
11. fishmonk5 November 2010, 21:44 GMT +01:00
Is Lennie beginning to regret ever touching a computer?
13. fishmonk6 November 2010, 10:21 GMT +01:00
Yes, just that!
Because your list is getting longer not shorter as time goes on...
14. lennie6 November 2010, 16:03 GMT +01:00
Somehow that is something I've grown accustomed too. Usually my list of 'immediate' things to do is just as long at the start of the week as when it began, even though I crossed some things of the list.
I've come to the conclusion: you are as busy as you make yourself be busy. If that makes any sense in English. You are in control of your own todo (short) list. :-)
