Author Topic: Best Match 2.0 (Read 44535 times)

vonkara · « **Reply #25 on:** December 09, 2008, 19:27 »

Quote from: madelaide on December 09, 2008, 19:20

Maybe they will (finally) also evaluate relevance by searching title and description.

Regards,
Adelaide

LOL..You really like how DT search work

I was searching an image today on IS and I caught myself to think it would be easier to search by title as you say

vonkara · « **Reply #26 on:** December 09, 2008, 19:31 »

Quote from: cdwheatley on December 09, 2008, 19:20

Does this mean? that if an image for example: a picture of a woman sitting in a chair on the grass.
Say it has 3 keywords "woman, chair, grass"
A buyer searches for the keyword "woman" and buys the image. Now the image will rank higher under the keyword "woman" but will not rank higher under the keywords "grass and chair". Another happy shopper buys the image by searching keywords "woman, chair" now image moves up best match for "woman and chair" but not for keyword "grass". Individual keyword ranking would be great as long as the image is not penalized for having the relevant keyword "grass".

It would be ok, but again if it's not penalizing the sales made from exposure. Like the ones from lightboxes or the file of the week and everything

PeterChigmaroff · « **Reply #27 on:** December 09, 2008, 19:32 »

I find it interesting that it all comes down to keywords, lack of perceptible noise, artifacts, filtering, shadows etc. and seems to have little to do with the actual image.

Peter

Sean Locke Photography · « **Reply #28 on:** December 09, 2008, 19:36 »

Quote from: madelaide on December 09, 2008, 19:20

Maybe they will (finally) also evaluate relevance by searching title and description.

Regards,
Adelaide

Doubtful, since they aren't translatable, and easily spammable.

Sean Locke Photography · « **Reply #29 on:** December 09, 2008, 19:38 »

Quote from: averil on December 09, 2008, 18:57

My interpretation of keyword ranking is a little different, so it will be interesting to see how the system is actually implemented. I see it as the order of a keyword in a list (and maybe the number of words in the list) determines how much weight it has, not some arbitrary? weighting determined by istock. But maybe I'm wrong

Yes, you're wrong, in this case:
"A long time ago we developed an algorithm to rank keywords on each file. Since then, we've been tracking data for every single file on iStock. Guess what? It works."

vonkara · « **Reply #30 on:** December 09, 2008, 19:42 »

Quote from: sjlocke on December 09, 2008, 19:36

Quote from: madelaide on December 09, 2008, 19:20
Maybe they will (finally) also evaluate relevance by searching title and description.

Regards,
Adelaide

Doubtful, since they aren't translatable, and easily spammable.

It work at DT at least. I never saw spamming in the title. And I think you can't put 2 times the same word in the description or title. And more I think that if you have more than 5 words in the desc.. they get less relevent. But I must find the right article to be sure

cdwheatley · « **Reply #31 on:** December 09, 2008, 19:43 »

Quote from: Vonkara on December 09, 2008, 19:31

Quote from: cdwheatley on December 09, 2008, 19:20
Does this mean? that if an image for example: a picture of a woman sitting in a chair on the grass.
Say it has 3 keywords "woman, chair, grass"
A buyer searches for the keyword "woman" and buys the image. Now the image will rank higher under the keyword "woman" but will not rank higher under the keywords "grass and chair". Another happy shopper buys the image by searching keywords "woman, chair" now image moves up best match for "woman and chair" but not for keyword "grass". Individual keyword ranking would be great as long as the image is not penalized for having the relevant keyword "grass".
It would be ok, but again if it's not penalizing the sales made from exposure. Like the ones from lightboxes or the file of the week and everything

Good point. Maybe they could just limit the best match rating of files to those that have been found by direct keyword search. I know on dreamstime a lot of sales have no keyword search associated with the file purchase. Get hit for self promotion would suck to say the least.

I thought I read that this was only going to be part of the best match also.

borg · « **Reply #32 on:** December 09, 2008, 20:12 »

We will see!

But, this is still "zero sum game"...

Without more buyers... $:-\$

shank_ali · « **Reply #33 on:** December 10, 2008, 02:25 »

Quote from: sjlocke on December 09, 2008, 19:38

Quote from: averil on December 09, 2008, 18:57
My interpretation of keyword ranking is a little different, so it will be interesting to see how the system is actually implemented. I see it as the order of a keyword in a list (and maybe the number of words in the list) determines how much weight it has, not some arbitrary? weighting determined by istock. But maybe I'm wrong

Yes, you're wrong, in this case:
"A long time ago we developed an algorithm to rank keywords on each file. Since then, we've been tracking data for every single file on iStock. Guess what? It works."

It's ok quoting MrThompson but has anyone thought about WHY istockphoto is changing best match and tiering the collection.
IMO they are nervous... are you sitting down ....$1.1 million payed a week to contributors which is about 30%-40% payout.So we can assume istock makes$ 2 million from OUR work.
I have looked through the other micro sites collections these past 2 months and if istock still thinks it offers the best images at the best prices they are mistaken.
I can only keep uploading what i consider to be stock worthy images to istockphoto but i certaintly dont view the company through 'rose titnted spectacles ' anymore.

DanP68 · « **Reply #34 on:** December 10, 2008, 04:26 »

Quote from: melastmohican on December 09, 2008, 17:43

Are we going to see this keyword ranking somewhere? So we can remove keywords that drags us down or replace them with better ones.

I interpret the ranking to work a little differently. It seems to me that images will rank separately for each keyword. For instance, say you have an image of a cow, and keywords: cow, milk. If buyers DL your image after searching for "cow," your image will be ranked higher for every new search for the word "cow." If buyers fail to DL your image after searching for "milk," your image will be ranked lower for every new search for keyword "milk."

I could be wrong, but I believe that is what the new system purports to do.

bittersweet · « **Reply #35 on:** December 10, 2008, 07:21 »

Quote from: DanP68 on December 10, 2008, 04:26

Quote from: melastmohican on December 09, 2008, 17:43
Are we going to see this keyword ranking somewhere? So we can remove keywords that drags us down or replace them with better ones.

I interpret the ranking to work a little differently. It seems to me that images will rank separately for each keyword. For instance, say you have an image of a cow, and keywords: cow, milk. If buyers DL your image after searching for "cow," your image will be ranked higher for every new search for the word "cow." If buyers fail to DL your image after searching for "milk," your image will be ranked lower for every new search for keyword "milk."

I could be wrong, but I believe that is what the new system purports to do.

That would seem to create feedback loops where the same files continue to be downloaded and each download works to secure the spot in the front. Just because another image contains the word "milk" but is NOT chosen to be downloaded, does not mean that the keyword is not appropriate for that image. I don't think there should be a "punishment" attached to not being selected for download.

I'm sure we won't know really know anything until it's up and running, and even then there will probably be a ton of the usual speculation. I hope it's an improvement!!

I always think it's funny that whenever iStock makes some big change, so many people think it is a knee jerk reaction to their impending doom, and not that it could possibly be a project that has been in development over a period of time (in this case, years... or so they say).

Despite their faults, they continue to be the trailblazers in this industry.

StockManiac · « **Reply #36 on:** December 10, 2008, 07:22 »

Quote from: DanP68 on December 10, 2008, 04:26

I interpret the ranking to work a little differently. It seems to me that images will rank separately for each keyword. For instance, say you have an image of a cow, and keywords: cow, milk. If buyers DL your image after searching for "cow," your image will be ranked higher for every new search for the word "cow." If buyers fail to DL your image after searching for "milk," your image will be ranked lower for every new search for keyword "milk."

But my question is what happens if a buyer purchases the image after searching for "milk", but then your image gets Wiki'd and they remove the word "milk" from your image!!!!

DanP68 · « **Reply #37 on:** December 10, 2008, 08:01 »

Quote from: whatalife on December 10, 2008, 07:21

Despite their faults, they continue to be the trailblazers in this industry.

If "trailblazing" refers to independents burning in anger, then I think you hit the nail on the head.

Otherwise, I don't see them doing anything so special. The InfiniteCollection and EVO are old news, and defunct LuckyOliver did the Sideshow 2 years ago, so whatever midstock offering iStock intends will be at best the fourth entrant into microstock.

As far as their search engine goes, all they really did was copy the Getty model. And so far, I can't say the search results I get at IS are any better or worse than the search results I get at Shutterstock or anywhere else. The new incarnation sounds promising, but I'll believe it's the best in the business when I see it.

bittersweet · « **Reply #38 on:** December 10, 2008, 08:19 »

(deleted)

vonkara · « **Reply #39 on:** December 10, 2008, 13:45 »

Quote from: DanP68 on December 10, 2008, 04:26

I interpret the ranking to work a little differently. It seems to me that images will rank separately for each keyword. For instance, say you have an image of a cow, and keywords: cow, milk. If buyers DL your image after searching for "cow," your image will be ranked higher for every new search for the word "cow." If buyers fail to DL your image after searching for "milk," your image will be ranked lower for every new search for keyword "milk."

I could be wrong, but I believe that is what the new system purports to do.

If it manage to work as we say, then I hope IS will understand that speeching of myself, only 60% of my DT sales are made with relevent keywords. Others are from exposure (without keywords) and weird or not fully relevent keywords, sometimes not even in the list. Then I don't know if managing some lightboxes or beeing on front page will be advantaging anymore.

They said they was working on this since a while, I hope they took care of every small details, because it could be easily a mess. Sometimes the more simple solutions are the best, like the DT and SS search engine work.

helix7 · « **Reply #40 on:** December 10, 2008, 15:37 »

I wrote something in the istock forums about this, but feel compelled to post a less edited version here:

The backslapping at istock over this best match change has me completely baffled. In one thread you've got people talking about how best match killed their sales for the last few months, and then they pop over to the best match 2.0 thread and are saying "Yay best match change again!" Last time I used the phrase "blind loyalty" over there it wasn't exactly received well, but I can't think of any better example of such a phenomenon than this.

As always, best match changes are done with the buyers in mind, and rightfully so. I think istock is making the right move here, catering to buyers even if it pisses off some contributors. But doing so with best match changes often means that many of us end up disappointed with the result. best match changes have been and will remain to be done for the buyers, and definitely not for us.

I'm as hopeful as anyone that this works out positively for everyone. But as a contributor I'm not singing istock's praises for this move and giving them the standing ovation, and I'm baffled as to why anyone else is. Sure it's possible that things could get better for those who suffered because of the last change. But there's an equal chance that things will take another dive for those same people after this one.

The fact of the matter is that best match can hurt a contributor just as easily as it can help them. No doubt, this next best match shift will hurt some people, including some of the same folks who are applauding and cheering "Go istock!" over this announcement. While we all hope the best match change will help everyone in a positive way, we also need to be a little more realistic and not be so surprised when next month there is another thread about best match killing sales with posts by some of the same people who are cheering for istock in today's thread.

crazychristina · « **Reply #41 on:** December 10, 2008, 15:51 »

There seems to be the assumption that an improved search engine will significantly improve sales. I personally think that istock has a rather distorted view of what makes good search. Banning conceptual keywords, for a start. Somehow I doubt that they'll be dragging too much market share from other sites with this move. If they do manage it, no doubt others will revamp their search engines to follow suit.

lisafx · « **Reply #42 on:** December 10, 2008, 17:48 »

Quote from: averil on December 10, 2008, 15:51

There seems to be the assumption that an improved search engine will significantly improve sales. I personally think that istock has a rather distorted view of what makes good search. Banning conceptual keywords, for a start. Somehow I doubt that they'll be dragging too much market share from other sites with this move. If they do manage it, no doubt others will revamp their search engines to follow suit.

You are probably right about this. It does seem like istock's constant messing with the best match is just a way to redistribute the pieces of an ever shrinking pie.

vonkara · « **Reply #43 on:** December 10, 2008, 18:30 »

Could someone explain me what this sentence exactly mean in Istock language? Taken at the end of the first post
http://www.istockphoto.com/forum_messages.php?threadid=80951&page=1

We will begin pulling leavers and turning dials to slowly turn up ranking in best match starting next week

I tought I was understanding english

jsnover · « **Reply #44 on:** December 10, 2008, 18:35 »

I think it means that they will slowly phase in use of the keyword ranking code. Given their inability to "fix" the banishment of vectors in searches even after they promised there'd be a quick fix for that, I can understand why they'd want to see what happened in practice with the live database.

The current best match doesn't produce a particularly nice mix of search results - it's better than the version where everything new was up front, but it's still got too many chunks of similar material together and the insane use of high views brings forward images that shouldn't be up front.

The only reason for any "yea" responses is that they are actually trying to do something about this. I don't understand all the up front praise before anyone's seen how this actually works. That is the only thing that will matter in the end - what it does, not what their process was in coming up with it or intentions in coding it.

CofkoCof · « **Reply #45 on:** December 10, 2008, 18:42 »

Quote from: Vonkara on December 10, 2008, 18:30

We will begin pulling leavers and turning dials to slowly turn up ranking in best match starting next week

I expect the results to be very interesting. Might have a few good laughs about it. I did some thinking the other day and it will be very hard to get the parameters right. Much much harder than with best match 1.0. And we all know they couldn't do it even with many tries. But then again, do they need to get them right?

vonkara · « **Reply #46 on:** December 10, 2008, 18:53 »

Thanks jsnover.

I'm aware of what it would be. I have images of food on table with landscape background. If the search work with how keywords are matching together for example, my images who have landscape and food as keywords would not match that much anymore.

It could work in part this way. They already have linked most of the keywords between them in the CV. It could explain when they say to have worked on this since a long time

Sean Locke Photography · « **Reply #47 on:** December 10, 2008, 22:15 »

Quote from: helix7 on December 10, 2008, 15:37

The fact of the matter is that best match can hurt a contributor just as easily as it can help them. No doubt, this next best match shift will hurt some people, including some of the same folks who are applauding and cheering "Go istock!" over this announcement. While we all hope the best match change will help everyone in a positive way, we also need to be a little more realistic and not be so surprised when next month there is another thread about best match killing sales with posts by some of the same people who are cheering for istock in today's thread.

I don't see this as a best match change. This isn't factoring newness from 21% to %30 or something. This is a fundamental change. They're incorporating other buyers' satisfaction with a keyword/image relationship to present future buyers with the images that may actually best suit their needs. It's a relevancy factor, and that hasn't been there before. I don't think the minority of sales from people who search for "cow" but actually decide to buy a picture of a cookie for some reason, will throw things off too much.

The point is to show the buyers the best results with a relevancy factor, and that saves them time. Those who save time are happy, and come back, and buy more.

That's how I see it. It's what we've been asking for forever.

crazychristina · « **Reply #48 on:** December 10, 2008, 22:26 »

One of the biggest problems with best match (IMO) is the positive feedback factor (in the systems sense). If an image that sells gets preferential placement in search as a result, then it will continue to sell well. It is probably an attempt to prevent this that images that take off have been getting squashed pretty quickly lately (in some recent version of the best match). It will be interesting to see if this remains a significant problem with the new algorithm in place.

traveler1116 · « **Reply #49 on:** December 10, 2008, 22:34 »

Do you think a lot of exclusives will quit being exclusive if the search actually finds the real best match? A lot of exclusives have had a pretty big jump in sales since the best match change so this "massive shift", I think thats what they are saying, would probably hurt the exclusives much more than nonexclusives.

MicrostockGroup Sponsors

Author Topic: Best Match 2.0 (Read 44535 times)

vonkara

vonkara

PeterChigmaroff

Sean Locke Photography

Sean Locke Photography

vonkara

cdwheatley

borg

shank_ali

DanP68

bittersweet

StockManiac

DanP68

bittersweet

vonkara

helix7

crazychristina

lisafx

vonkara

jsnover

CofkoCof

vonkara

Sean Locke Photography

crazychristina

traveler1116

Related Topics

Sponsors

Microstock Poll Results

Sponsors