I dont get it!  why is it so bloody hard?
Let me give an example: why give search preferance to lets say a surgical-theatre, industrial-plant, etc, WITHOUT PEOPLE??  empty?

search prefs should ofcourse be given to:  With people! especially considering 90% ( except travels) are searches for places including people.

When are they going to learn? I mean who . wants to see a surgical-theatre without SURGEONS?  who want to see an ind-plant without engineers or workers?

As someone who shoots nothing but objects, I want images WITHOUT people to get search preferences, unless the buyer specifies "people" in the search.

As someone who shoots nothing but objects, I want images WITHOUT people to get search preferences, unless the buyer specifies "people" in the search.

Are you being a wally on purpose??
I said places NOT the objects youre shooting. Read the post before answering.

I don't know what a "wally" is either.

If the original post were a bit more coherent, the responses might be more relevant.

maybe he meant Wall-E? he's looking for people because he's tired of being alone (at industrial place, etc etc...)

Apparently, then, I'm a cucumber. So be it.  I thought maybe it was a reference to Beaver Cleaver's older brother, who was a bit of a dope.

As far as I understand, lagereek is complaining that searches result in images without people more than with people. 

I personally find it contrary, and sometimes I wonder how my images are found in searches, as I don't use people.  Try "isolated stethoscope" and you will get tons of images in which stethoscopes are secondary, people being the real subject (as usual, DT is the best at actually showing adequate results).

And plese don't say the search should use the boolen "NOT people" because I would bet most people do not even know this can be used in a search.

I believe many people may want hotel rooms or restaurants or sports arenas without people.

Regards,
Adelaide

Quote from: RacePhoto on March 14, 2009, 10:48

No luck. I still don't understand the complaint either.

It is about the new search options on Dreamstime. You can now look for images with and without people. He thinks this is a useless feature since all customers - according to him - are looking for travel, industrial plants and surgery room images with people.

Oh that explains it, why didn't he just say that? 

There's nothing in the OP about DT or the with people / without people. Which in my opinion is a good way to limit the searches if a buyer wants a photo with No People! Makes for better seacrh results.

Of course some folks will have a person on a mountain, 3 miles away and list it as person in their photo, just so they get more views, which goes back to wasting buyers time with bad matches.

It's just another example of the totally inadequacy of keyword search, which is all todays microstocks really have - those that also have categories  don't have nearly enough of them.   There is just no way to say "I want a picture of a stethescope".  All you can ask for is pictures for which the submitter has applied the keyword "stethescope".  Leaving keywording up to the submitter avoids a huge labor cost but because of the obvious conflict of interest the result is millions of images tagged with misleading keywords.

Maybe some second-generation microstock will come along and do its own keywording from day one - probably by hiring a thousand keyworders in Mamba.  It won't be just keywording either - an image will be tagged with an honest statement of what it contains, as opposed to someone's idea of topics it relates too.

There has to be a lot of "search fatigue" going on today, where a buyer burns out after 12 pages of mostly unrelated images and says screw it, I'll just use the same image I used last year, this microstock thing wastes too much of my time. I want a picture of a stethescope - why is that so hard?

Out of sheer interest, I made a blogpost a few years ago about search strategies, keywords and number of images. That was the time that DT had about one million images online. It's a bit of math and I won't repeat it here but if you make a few assumptions like a set of M equivalued keywords per image and a search based on one or two keywords, the strategy will break down from N = 3 million images on. That means you'll get many hundreds or more of search results. The spam is obviously a disturbing factor but the sheer number of images is the real reason for this breakdown. Now, in 2009, we already have over 5 million images on the top sites...

To tackle this problem there are basically three solutions. One is limit the set of keywords per image, an approach CanStockPhoto tried a year ago and which raised a lot of protest from contributors. The second one is a controlled vocabulary like done by iStock. Same remark. The third solution is less drastic and consists in ranking keywords for relevance, and putting more weight in searches to the first keywords. Fotolia was supposed to do it. The problem is that most sites don't preserve ranking and rerank the keywords alphabetically. Nevertheless, this is one of the right approaches to make searches more relevant.

As for now, relevancy is done by some black magic. Some sites (DT, BigStock) use the repeat of keywords in title and/or description as a boost for relevance. Also the view and sales count enters into the equation, but that is irrelevant since it accumulates by age and for a buyer, the most relevant picture might be amongst the newest. SS makes newness a relevancy criterion but that has the disadvantage that older pictures (which might in fact be more relevant for a particular buyer) get burried. Also, popularity in se is a very bad criterion for being fit for the purpose of a particular buyer.

As for the stethoscope example, I was very amazed that Zymm would allow for the keyword 'doctor' in an image of an isolated stethoscope. This strategy is sortof OK for a database with a few hundred of thousand images but not for a few millions.

Most microstock sites grew out of amateur photo sharing sites and the key - search engine - is treated rather amateurish. The buyer is the victim.

And plese don't say the search should use the boolen "NOT people" because I would bet most people do not even know this can be used in a search.

I think the boolean searching is a good point. It can definitely help in eliminating some stuff. As far as spam, it's a tough call. I've seen IS eliminate terms that have gotten me downloads at Dreamstime. Granted some keywording is just abusive, but I think there should be some room for conceptual terms. Finally, I could see the "peopless" searches being handy. If I'm looking for a good background to Photoshop a product like a car onto, sometimes the people get in the way.

As madelaide mentioned, most buyers don't or won't take the time to understand the workings of a search, and they will not be educated - they are too many and come from many different industries, cultures and mind-sets. And they certainly do not spend time reading these forums. I bought photos for my design firm regularly for six years before becoming a contributor, and it was not until then that I really understood how keyword searching works. Only then did I really care. Prior to that the results were the results and I bought what I could find that was relevant. I never read forums, or tried to learn how it works because I thought I already knew. It has to be easy for the buyers and I believe the only way to achieve it is to overhaul the entire collections - or more likely is that new sites will evolve and take over where this is attended to from the start - with a limit on words, or site-created keywords and maybe some other mechanisms such as the secondary categories for "may be applicable to..." that jim_h suggested.