MicrostockGroup Sponsors


Author Topic: Flagged keywords - what ???  (Read 18158 times)

0 Members and 1 Guest are viewing this topic.

« Reply #50 on: September 11, 2009, 14:00 »
0
Yeah, I don't really see the problem with this.  If someone wants to spend their time for $.02 what do I care?  Seems to be some paranoia in the air.
fred

Fred, I'm well aware that not everybody feels like me.

However you can read clearly that others don't agree with you either.

Removing "cat" from a lion image is ridiculous. Why?

Simple: somebody wants an image of a lion or a panther or a jaguar why would it be wrong to use the keywords "wild" and "cat"? So somebody looks for "wildcat" and somebody looks for "wild cat" which is totally appropriate in my opinion but maybe I'm just some crazy lunatic who dares questioning a fantastic flagging system...  :P

And how many millions of images would have to be re-keyworded to fall into the proper quotation category??? Talking about disambiguation here aren't we?

This is literally shooting birds with cannons (not CANONs)  :-X

I'm more than happy to see the flagging system being USED appropriately for images that contain severe keyword violations. But from my experience looking at the flags that I got, too many people just take shots in the dark to get $.02 - that is what I'm referring to about wasting DT's resources.

Well taking your last statement first.  My understanding is that no one gets paid until the reviewers decide if the flag was appropriate.  If the DT reviewers get upset at too many mis-flagged words I am sure they will take whatever action they think is necessary.

Yes, this is disambiguation which is a reasonable process and cost wise this seems a very reasonable way of doing it.  No sweat from contributors (unless they are spamming) and little reviewer time. 

If you are not spamming what are you worried about?  Some twit flagging one of your keywords?  So what? If they flagged it wrong the reviewers will catch it if they caught one of your mistakes you have a chance to fix it.

Just don't fall victim to the "confirmation bias" inherent in what you read about the process.  You are likely to only see posts about stupid keyword flagging errors because people like to complain about it.  The corrected keywords will be a lot more frequent but few people willl post some story about how someone flagged one of their stupid keywords.

fred   


« Reply #51 on: September 11, 2009, 15:28 »
0
So the system is actually more open and fair than it has been in the past - but if you don't spam keywords you have nothing to worry about - being flagged has no impact whatsoever on a contributor unless he/she is spamming. If you are not spamming, you have not been harmed in any way. 
p
  spam and confirmation bias are highly subjective which is why i dont think this system is a good use of time.  the lion - cat example is excellent, and part of a broader category.  in these discussions, many seem to think there's only 1 way t search and 1 type of searcher - but buyers are going to differ in their knwledge, needs and searching ability, so we have to be ready for all f them -- when i post a orang or gorilla picture, i use 'monkey' as a tag - scientifically incorrect, but check out how often you hear it while taking thse pix in the first lace!

a lot of the confusion could be eliminared if sites used the description in the search [some do, most dont].  then specific info lke panthera, lcatin details, etc can still be found, and the tags can be reserved for more exact keywords. 

i incude detailed location & sci names w all my images - but on many sites no one will find my peru pix unless i include 'peru' in the tags too.  somene viewing a market scene might object to using Peru [one reviewer actually removed images claiming location was irrelevant for stock!] and flag the image, wasting a revieweer & my time.  yet someone who needs market scenes for a blog/article on peru wouldnt want images from bolivia

essetially the keyword system has to cover a wide range of ssearches and each site has to determine whether they want to control type I or type II errors - you can't easily do both -- so do you have as system that fails by returning some images that are not relevant?  or fails by NOT returning some images that ARE relevant?

« Reply #52 on: September 11, 2009, 16:01 »
0
I had the word 'labourer' flagged, this is British English and was appropriate for the image. The person who flagged was British and had no images. He didn't flag the American version of the word. I deleted the word because he obviously needs the money more that I do. Not impressed with this, very annoying.

- Do we get to know who flagged the image?
- Does DT tell us images have been flagged, or do we need to find it?

lisafx

« Reply #53 on: September 11, 2009, 16:11 »
0

- Do we get to know who flagged the image?
- Does DT tell us images have been flagged, or do we need to find it?

Yes, the flags show up as "comments" on the images.  And like other comments, the person commenting is shown. 

« Reply #54 on: September 11, 2009, 20:58 »
0

i incude detailed location & sci names w all my images -

while i agree that scientific names are relevant keywords, detailed location of where your images were shot may not be relevant at all ... I assume you are referring here to images that show the geographical location? Because if you shoot a cup and saucer isolated on white, for example, the geographical location is not relevant at all and should not be in the keywords. 

« Reply #55 on: September 12, 2009, 12:38 »
0

i incude detailed location & sci names w all my images -

while i agree that scientific names are relevant keywords, detailed location of where your images were shot may not be relevant at all ... I assume you are referring here to images that show the geographical location? Because if you shoot a cup and saucer isolated on white, for example, the geographical location is not relevant at all and should not be in the keywords. 


exactly - that's one extreme that doesnt need location -- however, esp'ly for editorials, there's a large middl ground where location  deends on the user -- eg, a picture of a moose with a forest background, or snowplows working dring a snowstorm - someone looking for those objects won't care where it was taken.  but someone doing a project on grand tetons and wyoming won't be interested in a moose from glacier NP or a snowplow in  oregon.

this is where mst keywrd complaints fail - the reviewer is thinking in too narrow a sense

s

« Reply #56 on: September 14, 2009, 08:28 »
0
IS tried to implement a similar system but it had several flaws.

1. Groups of contributors would create "ratings rings". Person A (no images - blank account) would flag the word "apple" on an image of an apple on white. This would push the image back in the search results (can't reward "spammers"). Then Person B would upload images of an apple on white which would now come earlier in the search because the other images were pushed back.

2. Relevant keywords were often removed because the reviewer didn't know what they were looking at. The previously mentioned Panthera Leo case is a good example. If I search for "large Cat" I won't find that image because "cat" was deemed to be irrelevant. The door swings both ways on this one. You may have contributors who don't know to use Panthera Leo for a Lion but you also have reviewers who are looking at the image of the lion and saying "It's not the astrological sign 'Leo' so it's an irrelevant keyword".

3. Reviewer overload. Once this catches on - particularly by those trying to dupe the system - the reviewers get so bogged down that it takes months before they can review things. IS had to set up a separate set of reviewers just for flagged keywords. When they get overloaded they stopped having the time to properly research the flag and just make snap decisions. This is part of what leads to number 2 on this list. Long wait times result in frustrated buyers (they just told you that the keyword was irrelevant why is still showing up a day later?) and pissed/panicked contributors (my image was flagged - I can't edit my image - why is it taking so long to realize that the flag is irrelevant - how is this affecting my sales/search placement?)

4. IS didn't pay flaggers. Instead you got a "wiki warrior" icon during a "wiki spree". Several people asked that the icon be removed because they were getting retaliation flags. ie you flagged my apple so I'll flag yours. Others requested that it be removed because they didn't want to be associated with the program when it really started going down hill with all the complaints of invalid flagging and reviewers removing relevant keywords.

5. Language barriers caused multiple issues. Contributors who didn't speak English well felt that they were getting punished for bad keywords (girl instead of woman) because flagging the keyword caused it change location with in the search results. Reviewers who didn't speak the same language as the contributor had a hard time judging subjective keywords. ie removing "knickers" or "bloomers" from a set of girls "panties" because they didn't know that all 3 words mean the same thing.

6. Personal experience. Every time someone wikied one of my images I would write a note to support explaining that the removed keywords were relevant and the words would be reinstated within a day or two. I don't know if DT has a system in place to reinstate keywords that should be there or not but they should. Reviewers don't know everything. It's not physically possible to know everything. And they don't usually have the time to properly research the difference between "wigwam" and "teepee".

« Reply #57 on: September 14, 2009, 08:44 »
0
@azurelaroux

Well said.

This addresses concerns I had all along.

Microstock is evolving all the time so many things have changed. Prices, commission, quality requirements and download numbers... (unfortunately).

The business hasn't become any easier and it will be important for the agencies to ensure that their contributors are working with them and not against (e.g. by keyword spamming).

It's impossible to edit out every single, not directly relating keyword out of every image online but notorious hardcore spammers should feel some serious consequences.

It's not the "lion" example that causes buyers to freak out. It's the contributors that use the keyword apple when the image shows a pear.

I think at some point the agencies will implement something like a "3 strikes and you're out" system. If spammers don't get punished for their actions they will keep doing it.

And locking someone's keywords it no punishment in my opinion...

« Reply #58 on: September 17, 2009, 06:54 »
0
I did not know there were that many problems with the IS plan, azurelaroux. I have noticed a decline in keyword rejections. They have moved on to disabling for lack of model/property releases now.

I also discovered another thing this morning. I had another image flagged on DT for a bad keyword. And I agree, the keyword needed to be removed. I went in and cleaned up the keywords on the image. Went back to the comments page to thank them, and the comment was gone and so was the comment that I got a couple of days ago from my first flagged image. Maybe they are removing them now because flaggers are already getting retaliation?

Yeah, this is probably going to end up like istocks plan.

But as a buyer I know it is very frustrating to search for a cow and get pictures of hotels in there (just an example, you get my drift). Something needs to be done.
« Last Edit: September 17, 2009, 06:57 by cclapper »

lisafx

« Reply #59 on: September 17, 2009, 16:57 »
0
I just had another batch of keywords flagged.  Some 4 year old images of spaghetti sauce ingredients (tomatoes, garlic, etc.) flagged for "pasta" and "sauce".  Those words are definitely spammed by today's standards and I went ahead and removed them. 

But I also images of a woman undergoing treatment for breast cancer flagged for the word "breast".  Without being able to key in phrases together, words can appear to be spam individually, but when taken as a phrase they can be legitimate. 

I don't blame the person who flagged them.  If I were looking at a search for "breast" and my bald chemotherapy lady showed up I would think it was spam too.

There really should be some better way to include phrases if individual words are going to be deleted. 

« Reply #60 on: September 17, 2009, 17:05 »
0
There really should be some better way to include phrases if individual words are going to be deleted. 

Composed keywords is something sites should consider - but may it's too late.  StockXpert used to, but then they moved the opposite way.

« Reply #61 on: September 17, 2009, 17:11 »
0
I'm hearing a lot of suggestions how the system could be improved.

However I'd consider the aforementioned cases as given requirements for a proper stock image agency to have IF they implement a flagging system.

After all the discussion here I still believe this flagging feature is not working efficiently at all.

Again, I am all for removing keywords that are not even remotely related to the subject in the image. But I also received flags that I had to ignore because the reported keywords directly relate to the image.

There is so much time wasted on both the contributors' and agency's side with this. The agency should handle it on their own.

« Reply #62 on: September 17, 2009, 17:20 »
0
If an image is flagged, does it lose something in search results while they have not been reviewed yet?

If you find a wrong flagging, like Lisa mentioned, what are we supposed to do, is there some way to automatically explain it to DT reviewers before they cut the keyword? 

Has anyone had the experience of having a wrong flagging accepted by DT?

« Reply #63 on: September 17, 2009, 17:43 »
0
If an image is flagged, does it lose something in search results while they have not been reviewed yet?

If you find a wrong flagging, like Lisa mentioned, what are we supposed to do, is there some way to automatically explain it to DT reviewers before they cut the keyword? 

Has anyone had the experience of having a wrong flagging accepted by DT?

Good questions.

I'm not sure if images will be pushed back that have pending flags. I strongly don't hope so, because that would cause a major flood of flags of contributors who want to knock out the competition even if it's just for a little while.

If you think your flagged images were flagged inappropriately (correct keywords) then there is nothing further you have to do. It was mentioned on the DT message boards that the reviewer will most likely agree that it was an unnecessary flag anyway. I'm holding my breath on my flagged images to see if how the reviewers will decide. That's why I think this system is flawed. Too much time spent on unnecessary reviewing.

Wrong flagging has occurred as it was mentioned in this thread earlier before. At least I consider the lion - big cat thing as wrong flagging. It's not like someone used the keyword "sexy girl" for an image showing a lion...


« Reply #64 on: September 17, 2009, 18:50 »
0
So, am I supposed to be reciprocal and flag the flagger's bad keywords?  Would that be the best way to show my sincere gratitude?

« Reply #65 on: September 17, 2009, 18:56 »
0
So, am I supposed to be reciprocal and flag the flagger's bad keywords?  Would that be the best way to show my sincere gratitude?

That's what was trying to say. When I checked the images of the people who flagged mine I could have flagged almost their entire portfolio if their interpretation of my wrong keywords would have been applied.

Naturally, this power will lead to some nasty situations between contributors. Watch it.

« Reply #66 on: September 17, 2009, 19:33 »
0
Wrong flagging has occurred as it was mentioned in this thread earlier before.

Yes, but has anyone experienced having correct keywords that were inadequately flagged being removed by DT.  THAT would be a concern, and it has happened in IS.

« Reply #67 on: September 19, 2009, 18:01 »
0
Actually I had a image flagged for the keyword business .. it was a financial concept involving money. LOL yeah my new plan is that every time somebody with no experience, a crap portfolio and no sales attempts to correct my keywording I'm going to flag 10 of their images for every 1 of mine flagged .. just because I can.  ;D

abimages

« Reply #68 on: September 20, 2009, 04:07 »
0
Actually I had a image flagged for the keyword business .. it was a financial concept involving money. LOL yeah my new plan is that every time somebody with no experience, a crap portfolio and no sales attempts to correct my keywording I'm going to flag 10 of their images for every 1 of mine flagged .. just because I can.  ;D

LOL I think most peoples reactions would be similar to yours. However such action could land you in hot water with the DT admins!

« Reply #69 on: September 20, 2009, 04:20 »
0
This is not DT, but please look at this FT video clip and look at it's keywords....

http://www.fotolia.com/id/16993659

« Reply #70 on: September 20, 2009, 15:37 »
0
Actually I had a image flagged for the keyword business .. it was a financial concept involving money. LOL yeah my new plan is that every time somebody with no experience, a crap portfolio and no sales attempts to correct my keywording I'm going to flag 10 of their images for every 1 of mine flagged .. just because I can.  ;D

LOL I think most peoples reactions would be similar to yours. However such action could land you in hot water with the DT admins!

I don't see how .. I wasn't talking about just going in and flagging words at random


 

Related Topics

  Subject / Started by Replies Last post
10 Replies
5865 Views
Last post February 03, 2011, 04:11
by Punit Patel
1 Replies
2473 Views
Last post July 09, 2007, 08:42
by snem
0 Replies
2318 Views
Last post August 05, 2008, 23:37
by bbettina
25 Replies
10856 Views
Last post April 03, 2011, 12:25
by stockastic
12 Replies
3544 Views
Last post November 14, 2012, 10:42
by dbvirago

Sponsors

Mega Bundle of 5,900+ Professional Lightroom Presets

Microstock Poll Results

Sponsors

3100 Posing Cards Bundle