pancakes

MicrostockGroup Sponsors


Author Topic: Why Does Shutterstock seem to break up mult-word keywords?  (Read 3047 times)

0 Members and 1 Guest are viewing this topic.

wds

« on: July 25, 2015, 21:46 »
0
It seems when I upload files with embedded IPTC keyword data, multi-word keywords seem to be broken up.
For example, the keyword "isolated on white" gets broken up into three keywords: "isolated", "on", "white".
What may I be doing wrong here?


PZF

« Reply #1 on: July 26, 2015, 04:31 »
+1
That's just how it is.......! Always now.

wds

« Reply #2 on: July 26, 2015, 06:30 »
0
And it doesn't cause search problems?

cuppacoffee

« Reply #3 on: July 26, 2015, 07:38 »
+1
It depends on how you define "problems". Other sites break up phrases too. If the searcher uses all the words of the phrase in their search the image will show up. No one site has the best search algorithm and with all the keyword spam and so many images it is a wonder any images show up in their proper places. Although many will disagree, defining your image with fewer keywords is a good way to improve sales. It is very important to have the most relevant words in your list. Skip the fluff. When people have 50 keywords and only two match the buyer's search words, they will be further down the page than someone with 10 keywords of which two match. It's a matter of percentage, 2 out of 50 is 4%, 2 out of 10 is 20%. The system sees the 20% match as more relevant.
« Last Edit: July 26, 2015, 07:41 by cuppacoffee »

ShadySue

« Reply #4 on: July 26, 2015, 08:14 »
0
Illustration of the issue on SS - Blue Whale:
http://www.shutterstock.com/cat.mhtml?lang=en&language=en&ref_site=photo&search_source=search_form&version=llv1&anyorall=all&safesearch=1&use_local_boost=1&autocomplete_id=&search_tracking_id=IReSg7sW5EBGHm0o0YzSTw&searchterm=Blue+Whale&show_color_wheel=1&orient=&commercial_ok=&media_type=photos&search_cat=&searchtermx=&photographer_name=&people_gender=&people_age=&people_ethnicity=&people_number=&color=
There are Blue Whales and many other species, some spammed/misidentified, but mostly which have the word 'blue' (water or sky) and the word 'whale', split from any other whale, mainly humpback, also sperm etc. So if someone has (correctly) put e.g. 'whale shark' and 'blue water' / 'blue sky' (not in quotes) these pics show on a search for blue whale. Particularly annoyinngly, you can't see the allocated title on the SS search.

Search should be better on iS because of the CV, and is cleaner in this particular case:
,%2235%22:[%22blue+whale%22]}#6b69bec]http://www.istockphoto.com/photos/blue+whale?facets={%2234%22:[%221%22,%228%22],%2235%22:[%22blue+whale%22]}#6b69bec
but so much keyword spamming goes on there, that you don't always get the clean search you'd hope for (e.g. the illo of elephants and a gnu). At least you get to see the title which shows what is claimed (e&oe!), though I see at least one illustration of a sperm whale silhouette with a blue texture titled as 'blue whale', as it is a blue whale, though not a Blue Whale.  ::) (Actually, it seems that is a misidentification case rather than deliberate spam.)
« Last Edit: July 26, 2015, 08:36 by ShadySue »

cuppacoffee

« Reply #5 on: July 26, 2015, 09:09 »
0
What everyone has to recognize is that a reviewer doesn't look at every image and then decide how it should be placed in any search. A system is in place that can only analyze data - words - no matter if they are in the title, description or list of keywords. Different fields of data are given more or less weight by different sites. Intelligent keyword search analysis is evolving but when there is a ton of data one word or phrase is weighed against the whole. The importance of each keyword is weighed differently against those rules and the result is supposed to help buyers (more buyers, more sales) by displaying the most relevant images first. For example, if a buyer searches for onion and photos of single onions are most often purchased when that search is made, more weight will be given to those images than the countless food shots of prepared dishes which also contain the word onion becausee it is an ingredient. So sales, matter in bringing an image to the top. That goes against the theory by some that new images sell better. I don't have an answer but it has to be remembered that machines, not people establish where an image shows up in any search. All data goes into a form (similar to a spreadsheet which is easier to understand if you are not a database expert) and then a set of rules is established to sort those words by order of importance. The meaning of "Importance" is tweeked by each site. Each site does it differently no matter if you enter a phrase or a series of words. The world of search technology is evolving but no matter the code behind the system the axiom garbage in, garbage out still applies. Too much garbage overloads the system and it's too late to go back.

wds

« Reply #6 on: July 26, 2015, 09:14 »
0
Here's the weird thing. It seems that if you upload a file with the (IPTC) keyword: "isolated on white", by the time the file is submitted, there will be three separate keywords: "isolated", "on", and "white". Yet you can go into the content editor and add the complete keyword "isolated on white" and it remains as one keyword as intended. So it would seem that the fact that the system breaks up multi-word keywords into individual keywords is in fact an error. It would also seem that in order to fix this, the contributor would have to individually manually edit every file (with multi-word keywords) in the content editor after it has been accepted. :(

« Reply #7 on: July 26, 2015, 09:53 »
0
how do you enter your keywords?

I do mine in lightroom and have always used 'multi words' or whatever you want to call them, and they've always survived the upload process into the SS content editor staying as "national park" or "evergreen tree" for example. I just checked.

maybe it's the program you are using. if the software automatically converts a space between words into a comma or semi-colon, that would probably be the source of the issue. Have you tried making quotations around your 'multi-words' in your software? That should fix the issue.

wds

« Reply #8 on: July 26, 2015, 09:59 »
0
how do you enter your keywords?

I do mine in lightroom and have always used 'multi words' or whatever you want to call them, and they've always survived the upload process into the SS content editor staying as "national park" or "evergreen tree" for example. I just checked.

maybe it's the program you are using. if the software automatically converts a space between words into a comma or semi-colon, that would probably be the source of the issue. Have you tried making quotations around your 'multi-words' in your software? That should fix the issue.

I use Lightroom also. In fact, on initial upload and when viewing in SS's "images to be submitted" the keywords aren't broken up, it is only upon actual submission that this happens.
I also do all my uploades via ftp.
« Last Edit: July 26, 2015, 10:02 by wds »

« Reply #9 on: July 26, 2015, 10:04 »
0

I use Lightroom also. In fact, on initial upload and when viewing in SS's "images to be submitted" the keywords aren't broken up, it is only upon actual submission that this happens.
I also do all my uploades via ftp.

ahh i see, right you are... just goes to show how much I look at my images after they've been accepted.


 

Related Topics

  Subject / Started by Replies Last post
8 Replies
3506 Views
Last post October 30, 2008, 06:42
by fotorob
3 Replies
3501 Views
Last post August 02, 2011, 08:19
by Slovenian
3 Replies
2419 Views
Last post January 28, 2013, 20:32
by Suljo
2 Replies
1805 Views
Last post January 29, 2015, 17:53
by Uncle Pete
17 Replies
2830 Views
Last post December 18, 2018, 16:47
by MatHayward

Sponsors

Microstock Poll Results