MicrostockGroup Sponsors


Author Topic: Wierd Getty Search Results  (Read 2392 times)

0 Members and 1 Guest are viewing this topic.

« on: September 10, 2011, 15:55 »
0
I've been looking at Edstock content and all the keywording etc. - so I hop over to Getty Images and poke around.  I do one of my typical searches and ask for "ice hockey player".   The search works pretty much okay, but there's all these photos of what looks to me like a female lion - then I read it's actually a panther (what do I know).  I look at the keywords and none of these cat photos contain my search words ice or hockey or player.  what?

Then today it comes to me.  The big cat comes up in the search has been labelled a Florida Panther.  There happens to be a pro NHL ice hockey team in Florida, yes, the Florida Panthers; florida and panther are in the keywords.  That's pretty.... intelligent.

Wow, looks like a whole lot more goes into the search than we can see on the surface.


ShadySue

« Reply #1 on: September 10, 2011, 15:58 »
0
That's what happens when you don't disambiguate.
Although I thought iStock's DA was based on Getty's; weren't these files properly DAd?
« Last Edit: September 10, 2011, 16:00 by ShadySue »

« Reply #2 on: September 12, 2011, 12:41 »
0
pretty intelligent maybe, but not wise. Which I guess sums up a lot of my complaints w/ the Getty "family" these days.

I mean really, if you want a picture of a mountain lion, why would you search on "ice hockey player"?

« Reply #3 on: September 13, 2011, 09:36 »
0
I have this weird theory that the disambiguate process that began at IS a few years ago consigned a huge number of images to some bottomless pit. I know my sales at IS dropped 40-60% immediately thereafter and never recovered. And, yes, I did go through the torturous process of updating my keywords although it took a few months. It didn't seem to make any difference at all in my sales level that remained far below past comparable months.

« Reply #4 on: September 13, 2011, 10:21 »
0
My memory sucks.  Wasn't Istock disambiguation before the sale?

ShadySue

« Reply #5 on: September 13, 2011, 10:26 »
0
My memory sucks.  Wasn't Istock disambiguation before the sale?
I don't think so. I joined in late 2006, when Getty had already bought iStock, and everyone was whining about DAing their files - especially the long-established members with already several hundreds or thousands of files.
According to Wikipedia, "On February 9, 2006 iStockphoto was acquired by Getty Images for $50 million USD. ... ... On September 18, 2006 iStockphoto displayed the first results of the Getty Images ownership. A new keywording taxonomy called a Controlled vocabulary was borrowed from Getty Images and implemented on iStockphoto to control and manage keywords and searches, and to provide multilingual searching abilities."
« Last Edit: September 13, 2011, 10:54 by ShadySue »

ShadySue

« Reply #6 on: September 13, 2011, 10:30 »
0
According to Wikipedia, "On February 9, 2006 iStockphoto was acquired by Getty Images for $50 million USD.  ...
Ha! I skimmed the rest of the Wikipedia entry on iStock and it is all very postive, anything negative is reported in a very short way. Previously, there was mention of the issues concerning contributors, but they've all been expunged. I guess Lobo or someone else from The Company is always jumping in editing out dissenting views.
However, there's a banner at the top saying, "This article is written like an advertisement. Please help rewrite this article from a neutral point of view." They wish!

« Reply #7 on: September 13, 2011, 10:48 »
0
My memory sucks.  Wasn't Istock disambiguation before the sale?

No. DA'ing was a result of adopting Getty's CV. We had a prior iStock-invented conversion from old to new categories (which I think was how they were going to try and solve the problem Gettty thought the CV addressed), so those of us who had been around a while had already been through one "new and improved" conversion already.

There were also bulk editing tools promised to contributors to help them with the conversion, and they were never released. They never helped in any way to educate contributors about the structure of the CV (which some of us did by walking the tree in the edit image dialog to learn how it was put together). There were massive amounts of contributor hours put into that effort. Hence the complaining.

johngriffin

« Reply #8 on: September 13, 2011, 15:35 »
0
You can see how we are doing it now at Cutcaster when you want to break down a search term i.e. disambiguate it. Try a image search for girl and then click on the keyword box for girl in the upper left to see a breakdown of meanings.

Then you can do a search for "hood" and see the dropdown in the autosuggest for the different meanings or if you select nothing from the autosuggest you can see how if you hover over the keyword that is in the upper left, you can then select your meaning. What do you guys think?

« Reply #9 on: September 13, 2011, 17:25 »
0
This is a huge area, perhaps worthy of its own topic if you're going to try and make a smart search out of "regular" keywords, but...

Taking your example of a search for girl, I picked Little Girl from the meanings and a series of images of this woman were included. Girl is not a keyword on the image, but with little black dress and young woman I suppose your search assumed little girl. Incorrectly.

If I try to search for little girl blond I get this image and this one., neither of whom are little girls. You could have something simple that excludes any image with nude or sexy keywords from the results for Little Girl (SS won't allow those combinations to be entered on images).

There have been a number of innovative approaches to search, including using visuals (more like this one or this group) and making assumptions based on the sets of keywords. I'm no fan of Getty's CV (a big problem is that it's so inflexible and good keywords on less popular topics or places are hard to do), but if you're going to try a CV on the fly, it'll have to deal with a lot of complex issues to make it work well.


 

Related Topics

  Subject / Started by Replies Last post
9 Replies
3585 Views
Last post May 02, 2007, 17:03
by hospitalera
5 Replies
2079 Views
Last post March 13, 2011, 21:51
by Read_My_Rights
2 Replies
1553 Views
Last post April 12, 2011, 20:11
by redwater
0 Replies
639 Views
Last post January 18, 2013, 12:14
by grsphoto
2 Replies
1202 Views
Last post February 15, 2013, 16:29
by Pixart

Sponsors

Microstock Poll Results

Sponsors