BM Discussion: Updated Jan 25th

Displaying 161 to 180 of 2050 matches.
Leontura
Member is a Gold contributor and has 10,000 - 24,999 Photo downloadsMember is a Bronze contributor and has 250 - 1,249 Video downloadsMember is a Bronze contributor and has 125 - 1,249 Audio downloadsMember is a Diamond contributor and has 25,000 - 199,999 Illustration downloadsExclusiveExclusive iStockphoto IllustratorExclusive iStockphoto VideographerExclusive iStockphoto Audio Artist
Posted Thu Dec 6, 2012 10:31PM
Back to the issue of new content showing up when multiple keywords are searched on – here’s why it happens:  We store relevance data at a keyword level, e.g. we know which pictures are popular for a search on "copy space" or "vegetable", and for single keyword searches we push up those images in Best Match.  However, for combinations of multiple search terms that reduce the search results significantly (many great examples have been reported in this thread), the amount of data that determines keyword-level relevance is much lower than for single term searches.  In this situation, the search engine falls back to other data to determine what to push up to the top of the search results.  We currently fall back heavily on file age, so you see newer content at the top of the search results.


Why don't you use the average relevance data for the two terms? It seems that you know a file is popular for "copy space" and for "vegetable" but for some reason you don't know it'll be useful for people searching for "copy space AND vegetable". The data is there but you're not using all of it. And the result is newest first. I'm sure if you had a lot more data your method would work better but it seems from the results that you don't.

(Edited on 2012-12-06 22:45:43 by Leontura)
lagereek
Member is a Diamond contributor and has 25,000 - 199,999 Photo downloads
Posted Thu Dec 6, 2012 11:12PM

Whatever happend to the fundamental laws of " keep it simple" and easy. Why not just go for Popular, Relevant and Best-match?


Pple here giving millions of examples doesnt help at all because they are simply judging it from the content of their own portfolios. I think maybe the entire CV needs a re-vamp or needs to be looked at.


All I know is this and I have heard it numerous times. When a buyer start searching here, he encounters a whole heap of collections, this and that and in the end its all confusing. so much so, that he simply might move on and then its lost.


 
shank_ali
Member is a Gold contributor and has 10,000 - 24,999 Photo downloadsExclusive
Posted Fri Dec 7, 2012 12:10AM
Posted By Lobo:

We will have some responses prepared for tomorrow. We will continue to gather inquiries from the community through this thread but we hope to accommodate some clarity on a number of the more frequently asked questions in here.

Your comments are appreciated. Please do try to keep things focused on the OP so we don't get to far off track in here.




Could we have some information, that i believe alot of contributors find unfair,in regards of the relevancy factor of where contributors live.


Getting feedback from our buyers is one thing but this part of BM equation,for me,is not acceptable.
stockwerk
Member is a Bronze contributor and has 250 - 2,499 Photo downloadsExclusive
Posted Fri Dec 7, 2012 12:56AM

In the OP it says "We store relevance at a keyword level." My question is what counts as a "hit" on a keyword that increases the relevance data for that keyword for that file. Does a view of a file from the search result set increase the relevance of those keyword(s) used for the search? If that is the case, the next question is, whether this mechanism is also broken (the same issue as not all views registering correctly)? If it is broken, what is the implication for Best Match?
Imgorthand
Member is a Gold contributor and has 10,000 - 24,999 Photo downloadsExclusive
Posted Fri Dec 7, 2012 1:23AM

There was always an impression that number of views plays some role in BM. Perhaps the number of views that is extremely low for new files since the views are not registering correctly - boosts them unnaturally high in BM. 


Just a speculation.
Gannet77
Member is a Gold contributor and has 10,000 - 24,999 Photo downloadsMember is a contributor and has less than 250 Video downloadsExclusive
Posted Fri Dec 7, 2012 2:19AM
Posted By shank_ali:

Could we have some information, that i believe alot of contributors find unfair,in regards of the relevancy factor of where contributors live.


Getting feedback from our buyers is one thing but this part of BM equation,for me,is not acceptable.

My understanding has always been that the location relevancy relates to where the person doing the search is located and what search terms are used for purchases at that location, not where the contributor lives.
benjikat
Member is a Gold contributor and has 5,000 - 12,499 Video downloads
Posted Fri Dec 7, 2012 2:49AM
Posted By Gannet77:

Posted By shank_ali:

Could we have some information, that i believe alot of contributors find unfair,in regards of the relevancy factor of where contributors live.


Getting feedback from our buyers is one thing but this part of BM equation,for me,is not acceptable.


My understanding has always been that the location relevancy relates to where the person doing the search is located and what search terms are used for purchases at that location, not where the contributor lives.

I was told at a 'lypse that it was the opposite - someone who lives in New York will be assumed to have taken the better and more authentic material of New York. I think that idea is rubbish as I seem to do my better work when focussed on filming away from home.
sodafish
Member is a Diamond contributor and has 25,000 - 199,999 Photo downloadsMember is a contributor and has less than 250 Video downloadsMember is a Black Diamond contributor and has more than 200,000 Illustration downloadsExclusiveExclusive iStockphoto IllustratorExclusive iStockphoto Videographer
Posted Fri Dec 7, 2012 3:05AM
Posted By benjikat:

Posted By Gannet77:


Posted By shank_ali:

Could we have some information, that i believe alot of contributors find unfair,in regards of the relevancy factor of where contributors live.


Getting feedback from our buyers is one thing but this part of BM equation,for me,is not acceptable.



My understanding has always been that the location relevancy relates to where the person doing the search is located and what search terms are used for purchases at that location, not where the contributor lives.


I was told at a 'lypse that it was the opposite - someone who lives in New York will be assumed to have taken the better and more authentic material of New York. I think that idea is rubbish as I seem to do my better work when focussed on filming away from home.

If that's really the case I hope this will be changed, that simply doesn't make sense to me. What about the talented photographers that travel a lot, what about illustrators, ... it's completely unpredictable. 
RapidEye
Member is a Black Diamond contributor and has more than 200,000 Photo downloadsMember is a contributor and has less than 125 Audio downloadsExclusiveExclusive iStockphoto Audio Artist
Posted Fri Dec 7, 2012 3:55AM
Whichever the case on the regional relevancy, it should be scrapped. It can have only a marginal effect on sales but it adds an extra layer of complexity to a BM that has now been demonstrated to be so complex as to be opaque even to its creators.

(Edited on 2012-12-07 05:02:53 by RapidEye)
georgeclerk
Member is a Diamond contributor and has 25,000 - 199,999 Photo downloadsMember is a Bronze contributor and has 125 - 1,249 Audio downloadsExclusiveExclusive iStockphoto Audio Artist
Posted Fri Dec 7, 2012 4:02AM
Posted By Lobo:

...

We will have some responses prepared for tomorrow. We will continue to gather inquiries from the community through this thread but we hope to accommodate some clarity on a number of the more frequently asked questions in here.

Thanks Lobo, and thanks to those contributors who kept insisting that there was a problem with best match until iStock were willing to acknowledge it.

If possible please include information about why the quality of results returned by BM has fallen so heavily in recent weeks/months.  A system that was working well in the past and returning good, relevant results is not doing that now.  Do iStock know why this is, and is the data still available to allow more relevant results?

Just in case iStock don't think there's still a problem needing to be fixed, here's another example.  I know that results differ depending on the user, but nobody could argue that the results I see (on several different systems, etc..) from a photo BM search for 'Asia beach' are optimised in any way for relevancy.   Out of the top 100 results, only two have any downloads (one each).  Every single file number is greater than #21000000.
juqboxmusic
Member is a Diamond contributor and has 12,500 - 99,999 Audio downloadsExclusive iStockphoto Audio Artist
Posted Fri Dec 7, 2012 4:30AM
Just to report on a couple of things when browsing the Audio BM; keywords such as "love", "friendship", "family" & "teamwork" seem to be showing similar sets of results that contributors describe when searching for videos and photos, lots of 0 download files that seem to stay there quite happily without selling, however they could probably sell for other keywords. As far as I can tell, I only noticed this change once the stats started registering cash sales, I may be wrong though, that's just my observation. I'm sure there are many examples of keywords that show the same type of results with audio, but I can confirm from where I am (UK), I'm seeing very strange video & photo results indeed (compared with about 2 months ago), if I was a buyer, and I have been in the past, I would probably change the sort to downloads. 


I think rather than falling back on File Age, it would be best to fall back on Views per DL. That means there is no discrimination against the age of a file, it just simply shows the files that sell well. However at the moment views are not registering in real time, which would probably have to be corrected first before that would work properly.

(Edited on 2012-12-07 04:34:26 by juqboxmusic)
RapidEye
Member is a Black Diamond contributor and has more than 200,000 Photo downloadsMember is a contributor and has less than 125 Audio downloadsExclusiveExclusive iStockphoto Audio Artist
Posted Fri Dec 7, 2012 5:05AM

Posted By juqboxmusic:

I think rather than falling back on File Age, it would be best to fall back on Views per DL.


Better and easier would be DLs/month.
crossbrain66
Member is a Silver contributor and has 2,500 - 9,999 Photo downloadsMember is a contributor and has less than 250 Illustration downloadsExclusive
Posted Fri Dec 7, 2012 5:11AM
Posted By Lobo:

One more thing to keep in mind: customers do not use a lot of search terms , and in general do not narrow down their search results much by using the various search features on the site.

As the number of images in the collection grows, this customers behaviour will no longer be the norm. Searching in millions of images for the right ones is extremely time-consuming if not optimally aided by the search engine. A good functioning BM with multiple search-terms shows good and relevant images on the first pages. This was once a plus for istockphoto - it's of invaluable importance to bring it back fast.
sjlocke
Member is a Black Diamond contributor and has more than 200,000 Photo downloadsMember is a Gold contributor and has 5,000 - 12,499 Video downloadsMember is a Bronze contributor and has 125 - 1,249 Audio downloadsMember is a Silver contributor and has 2,500 - 9,999 Illustration downloadsExclusiveExclusive iStockphoto IllustratorExclusive iStockphoto Videographer
Posted Fri Dec 7, 2012 5:32AM

Posted By benjikat:
I was told at a 'lypse that it was the opposite - someone who lives in New York will be assumed to have taken the better and more authentic material of New York. I think that idea is rubbish as I seem to do my better work when focussed on filming away from home.


No, it is based on buyer's location. Not contributor's.
allyclarkCLOSED
Member is a Silver contributor and has 2,500 - 9,999 Photo downloadsExclusive
Posted Fri Dec 7, 2012 6:17AM

Choo-choo!  The train arriving at platform 1 is the 2000 Gravy Train from iStock Central. This train will terminate here. Will passengers who have not thought through their business model properly kindly take care when they finally stop stamping their feet and step off into the real world.......  


Businesses are affected by decisions outwith their control all the time.  How you react to it separates the winners from the losers.  Not being able to see it as a potential risk or failing to minimise the risk is one thing, but failing to get the message three months down the line???
Sproetniek
Member is a Silver contributor and has 2,500 - 9,999 Photo downloadsMember is a Bronze contributor and has 250 - 1,249 Video downloadsMember is a Bronze contributor and has 125 - 1,249 Audio downloadsExclusiveExclusive iStockphoto Audio Artist
Posted Fri Dec 7, 2012 6:19AM
Posted By lagereek:
Whatever happend to the fundamental laws of " keep it simple" and easy. Why not just go for Popular, Relevant and Best-match?

I've never understood the need for a category "Relevant" - as opposed to what - searching for "Irrelevant"? WTF?
georgeclerk
Member is a Diamond contributor and has 25,000 - 199,999 Photo downloadsMember is a Bronze contributor and has 125 - 1,249 Audio downloadsExclusiveExclusive iStockphoto Audio Artist
Posted Fri Dec 7, 2012 6:33AM
Posted By Sproetniek:


Posted By lagereek:
Whatever happend to the fundamental laws of " keep it simple" and easy. Why not just go for Popular, Relevant and Best-match?





I've never understood the need for a category "Relevant" - as opposed to what - searching for "Irrelevant"? WTF?



If you take Google results as an example (since they do it so well), searching for 'asia beach', you get 510,000,000 results.  So they found half a billion results that have something to do with 'asia' and 'beach'.

They are all 'relevant' results, because they all have some connection to asia and beaches.

But what makes Google's results so good is that they sort them by lots of critera which mean that the first pages of results are optimised to be the most relevant they can produce, based on lots of feedback including choices made by previous Google users, and what they know about the person doing the search.

(Edited on 2012-12-07 06:40:34 by georgeclerk)
Difydave
Member is a Diamond contributor and has 25,000 - 199,999 Photo downloadsMember is a contributor and has less than 250 Illustration downloadsExclusive
Posted Fri Dec 7, 2012 7:06AM

The other thing that makes Google results so good is that generally they are the results the searcher wants, rather than the results that the search engine wants the searcher to see.


It seems to me that continual attempts to put iStock's "preferred" content in front of buyers here has ended up with a system which is over complicated, isn't good for contributors (who are after all's said and done putting their work here to be sold), and perhaps most importantly gives buyers results they dislike.


From both buyer's and contributor's perspectives it seems that it's now gone beyond giving acceptable results.
swilmor
Member is a Diamond contributor and has 25,000 - 199,999 Photo downloads
Posted Fri Dec 7, 2012 7:55AM
Posted By benjikat:


Posted By Gannet77:


Posted By shank_ali:

Could we have some information, that i believe alot of contributors find unfair,in regards of the relevancy factor of where contributors live. Getting feedback from our buyers is one thing but this part of BM equation,for me,is not acceptable.


My understanding has always been that the location relevancy relates to where the person doing the search is located and what search terms are used for purchases at that location, not where the contributor lives.


I was told at a 'lypse that it was the opposite - someone who lives in New York will be assumed to have taken the better and more authentic material of New York. I think that idea is rubbish as I seem to do my better work when focussed on filming away from home.



I find this hard to believe but would like to hear official confirmation from someone at iStock about whether this is true. I am from the US but I live in Spain and have that listed as my residence for iStock. While I do shoot a lot of content in Spain, I also travel to the US every year and shoot a lot of material there. Much of my content is probably more influenced by my own cultural background(US) than that of where I choose to live. This would mean that I am penalized in the BM simply because of where I choose to live regardless of the actual content I produce and upload.


ETA: I've always been under the impression that location relevancy in BM was based on the location of the buyers and had nothing to do with where the contributor was from.

(Edited on 2012-12-07 08:12:31 by swilmor)
Juanmonino
Member is a Black Diamond contributor and has more than 200,000 Photo downloadsExclusive
Posted Fri Dec 7, 2012 8:01AM
We all know the BM is broken, the question is if they are willing to fix it, and if they want to fix it, if they can fix it, and if they can fix it, when they will fix it.
I have been here for 6 years, this is the most serious issue ever.
This thread has been locked.
Displaying 161 to 180 of 2050 matches.
Not a member?Join