Author Topic: Best match (Read 13849 times)

traveler1116 · « **Reply #25 on:** November 15, 2008, 17:36 »

I didn't mean just yuri's but there are a lot of nonexclusives that have some of the best work out there. The point was that if exclusivity is more important than getting the best images the results will suffer. I highly doubt buyers care at all whether an image is exclusive to istock it still can have been sold to thousands of other businesses, the highest quality collection and best search results seem to be most important. Getting a good collection by having exclusives on it is part of it but putting their images before other higher quality more salable images doesn't make IS more useful for buyers does it? It is IS's decision to do what they want and it really doesn't matter too much to me, I would much rather sell images at DT or anywhere else where I will get higher commissions, so if buyers leave IS and I get more sales from other sites I'd be happy.

loop · « **Reply #26 on:** November 15, 2008, 18:27 »

If you don't understand the point about uniqueness in a market that offers almost exactly the same pics everywhere, I'll just add that I disagree.

And, well, the fact is that anybody can have anything both ways. It's a choice.

hoi ha · « **Reply #27 on:** November 15, 2008, 21:30 »

Quote from: sjlocke on November 15, 2008, 13:38

Quote from: Eco on November 15, 2008, 12:51
There is no relevancy involved in Best Match, as has been said thousands of time. As for benefitting the buyer, I hardly see any complaints from buyers about the returns from the sort. They don't come onto the forum screaming about how bad it is.

Quote

Yes, as Sean says, "best match" on istock has nothing to do with relevancy as has been repeated to the contributors repeatedly - I actually find that disingenuous as a buyer - istock can sort its searches any way it wants to ... that is their business ... but when a buyer goes to istock and sees "best match", they assume that is what it means ... i.e. the best match for the key words they have input and that is relevancy - I would bet 95% of the buyers on istock assume that best match says what it implies ... a search based on the relevancy of their keywords ...

CofkoCof · « **Reply #28 on:** November 17, 2008, 00:56 »

Quote from: sjlocke on November 15, 2008, 13:38

Quote from: Eco on November 15, 2008, 12:51
Madelaide got it right. Best match should essentially be employed for the benefit of the BUYER - to get the most relevant picture for the search term(s) used. IS apparently see best match as a tool to benefit CONTRIBUTERS and in its latest incarnation openly and undeniably to benefit their exclusive photographers.

There is no relevancy involved in Best Match, as has been said thousands of time. As for benefitting the buyer, I hardly see any complaints from buyers about the returns from the sort. They don't come onto the forum screaming about how bad it is.

So I guess there is no keyword spamming problem since buyers don't come to the forum screaming about it. There's life out there outside of the forums also you know

Besides this thread is about how we feel best match should look like and I think most would agree relevancy should (if could) be a part of it.

Sean Locke Photography · « **Reply #29 on:** November 17, 2008, 04:28 »

Oh, I totally agree relevancy should be a factor. I've argued for weighted keywords and sales tracking from searches as much as anyone, if not more.

But we ain't there yet.

Aetherial · « **Reply #30 on:** November 17, 2008, 04:49 »

I also think that some weighing system for keywords should be employed. When you allow 50 or 100 or whatever keywords people rush to fill all the slots because that increases the chance of image being seen and consequentially spamming. if you reduce that to 10 or less people start complaining how thats not enough.

IMHO problem is that there is no boundary between conceptual and descriptive keywords. for instance, if you have a boy and a girl with a bouquet of flowers, i pressume most people would tag it with "love, relationship, romance etc..." to broaden the chance of it being used for that purpose. And then it also gets "Valentine, anniversary, gift" etc etc... is it spamming? on one hand it is, because it is arguable that the picture represents solely one concept and that it represents it clearly. on the other hand, the buyer searching for Valentine picture for next magazine issue might and most probably will include Valentine and then if your image doesn't have it as a keyword it will never be found. what should the image tagged with Valentine have? Who is to say that? kisses, roses, hearts, people in love....?
i think that categories are useless, they just waste contributors time when uploading and can also be misused. I think that you should give slots for up to 5 keywords, but no more for describing concept that photographer had in mind when he made the picture. those keywords should be ordered by importance. the buyer could check whether he wants to search by preconcepted photos or just go after whats in the picture. the rest should strictly be descriptive keywords describing only what actually is on the photo and have no weighing. maybe splitting them into major keywords (are there people in the picture, how many, full body or closeup, what objects) and minor (what hair color, what composition, what clothing).

now, how to rearrange the 3 million photo collection....

CofkoCof · « **Reply #31 on:** November 17, 2008, 08:04 »

Quote from: sjlocke on November 17, 2008, 04:28

Oh, I totally agree relevancy should be a factor. I've argued for weighted keywords and sales tracking from searches as much as anyone, if not more.

But we ain't there yet.

Yeah I know, I've read one of your replies on the IS forums about it.

BTW here's what I wrote in the topic "How's your month shaping up" and is talking about the keyword relevance search:

Quote from: CofkoCof

Mostly agree with you but let me offer another idea for the algorithm, that has been mentioned a few times and I think would solve at least two problems at the same time: it (should) return good results and it would help against keyword abuse. Here it goes: sites would have to track the keywords that lead to the download of a particular image. For each keyword you would have a factor of relevance. This way you would get good (because they sold many times) relevant (because their relevance factor is high) images at the front. Ofc you would have to put in other factors like age to give newer images a chance at the start. Also sometimes people go to your portfolio and don't download image on a keyword but just by browsing. You could maybe add a small factor like 0.1 to all keywords if an images get's downloaded like that.

However it looks like this would be too hard to implement (also you would need to track a lot of data, have a seperate field in the database for all keywords on a image). So I don't see it coming anytime soon.

lisafx · « **Reply #32 on:** November 17, 2008, 17:47 »

CofkoCof, I think your plan for tracking the keywords that led to sales is really good, and I am betting that as this industry matures something like that will become the standard.

But for the reasons you mentioned I don't think it will happen any time soon.

I also doubt istock will be the one to implement it. They seem to be hampered by shaky infrastructure. I suspect that may be the underlying cause of a lot of the peripheral issues (buggy search results, site slowdowns/outages, no more live stats, etc.) that are always cropping up.

Whiz · « **Reply #33 on:** November 17, 2008, 17:57 »

Quote from: lisafx on November 17, 2008, 17:47

CofkoCof, I think your plan for tracking the keywords that led to sales is really good, and I am betting that as this industry matures something like that will become the standard.

But for the reasons you mentioned I don't think it will happen any time soon.

I also doubt istock will be the one to implement it. They seem to be hampered by shaky infrastructure. I suspect that may be the underlying cause of a lot of the peripheral issues (buggy search results, site slowdowns/outages, no more live stats, etc.) that are always cropping up.

I thought Dreamstime already did this?

Sean Locke Photography · « **Reply #34 on:** November 17, 2008, 18:41 »

The sales relevancy idea is nothing new. As I mentioned we've been throwing that and manually weighted keywords around as suggs for a couple of years.

Aetherial · « **Reply #35 on:** November 18, 2008, 04:13 »

I'm very interested to know, based on your experiences, what of these is closest to (statisticaly) average buyer:

a) do they want to buy an image that is proven with many sales (which means that it is a strong image but also appearing on many many other places)?

b) do they want to buy an image that hasn't been used much (low number of dls)?

c) they don't care

it seems to me that pushing good sellers on top of search which is kind of standard search engine behavior forces those images on buyers leaving much other material unseen. if it doesn't sell quickly it gets burried. maybe it was the best option in the beginning when image database had less than 1 million images (which is zilch compared to macro stock libraries), but now maybe more than 80% of content never gets noticed. what is the point of having 3,4 or 5 million images that are not easily available to buyers. how deep does the majority dig into search results? page 10, page 50...? I don't think so.

if you would shuffle search results to enable equal exposure to both old and new, sold and unsold files and in the same time tighten the reviewing standards to be even more strict to weed out bad photos, it would yield better, healthier image library and in the same time contributors would know to play the quality game over quantity game, because accepted images would pay off even if not noticed immediately, no need for large series of same images.

Sean Locke Photography · « **Reply #36 on:** November 18, 2008, 07:26 »

MicrostockGroup Sponsors

Author Topic: Best match (Read 13849 times)

traveler1116

loop

hoi ha

CofkoCof

Sean Locke Photography

Aetherial

CofkoCof

lisafx

Whiz

Sean Locke Photography

Aetherial

Sean Locke Photography

Related Topics

Sponsors

Microstock Poll Results

Sponsors