Extra Form Fields - Do You Use Them?

[madelaide]

ann,

I meant brackets and quotation marks in keywords are ignored.  RP showed examples in which it does make a difference when the buyer uses "" for searches. 

Having "South American" with quotation marks in my image's keywords doesn't impede the image of being found in a search for American Airlines without quotation marks (which is actually treated by the search engine as American OR Airlines, but they would not appear if the search is for "American Airlines", because the search engine understands the buyer wants this expression.

I'm not sure - RP can possibly enlighten us about this - if the quotation marks in search equal American AND Airlines (with the proximity bringing relevance) or only the perfect match (therefore if I have in my keywords American plane airlines, this image will not return because there is a word in the middle)

[ann]

@madelaide  {{{hugs}}}   I so appreciate your kind perseverance in clarifying this, since keywording can significantly help make or break an image, esp on Alamy:

So it seems I don't really understand why there's such a different # of results when buyer searches red car VS "red car"    

I thought it was because the specific phrase "RED CAR" appears with quotation marks around it in a KEYWORD Box for around 3-thousand images, and the words RED and CAR appear somewhere as Keywords and/or in Captions in 45-thousand images.  

[madelaide]

@madelaide  {{{hugs}}}   I so appreciate your kind perseverance in clarifying this, since keywording can significantly help make or break an image, esp on Alamy:

So it seems I don't really understand why there's such a different # of results when buyer searches red car VS "red car"    

I thought it was because the specific phrase "RED CAR" appears with quotation marks around it in a KEYWORD Box for around 3-thousand images, and the words RED and CAR appear somewhere as Keywords and/or in Captions in 45-thousand images.  

I believe you are right, ann. As I said, I'm not sure exactly how Alamy uses a search with quotation marks, but without them they do an OR search, so images with either one of the keywords are returned, which is obviously not what the buyer wanted.

But again, it is not just Alamy's fault.  I only think they should take advantage of their more advanced keywording system.

[RacePhoto]

@madelaide  {{{hugs}}}   I so appreciate your kind perseverance in clarifying this, since keywording can significantly help make or break an image, esp on Alamy:

So it seems I don't really understand why there's such a different # of results when buyer searches red car VS "red car"    

I thought it was because the specific phrase "RED CAR" appears with quotation marks around it in a KEYWORD Box for around 3-thousand images, and the words RED and CAR appear somewhere as Keywords and/or in Captions in 45-thousand images.  

Close except it's not Red and Car for the one out of quotes, it's Red OR Car

So in the "South American" example, the conclusion is that it's finding something that isn't there, but it's finding two words which ARE there.

Search:
south american = 161,784 Results
"south american" = 32,513 Results

Clearly it's finding different sets images. Someone needs to look at all the words in their keywords, title and caption for both words, and find that the search is not intuitive at all. Both words within the quotes MUST be present. In the general search, no quotes, either word can be present. Did that help a little?

It's not magic, it's really simple logic. Once you get it, the apparent wrong results, suddenly make sense.

Does it seem like I'm repeating myself? Maybe I should kust start hitting my head against the wall, because writing doesn't do anything either.

Enter "south american toad" in quotes. No Images. That's because ALL words must be present or the search won't include any images. If the two words South and American aren't in the search, nothing will show. If someone tells you the search is finding things that are impossible, you might want to check to see if they are correct, because the seacrh is a dumb computer that doesn't think.

Back to the other part, which is also falling on deaf ears. Use at the very minimum three word searches, please, if you want good results. Of course two words searches are going to get bad results, it's a crappy search!

Do we need to get into Stop Words? The words that the search ignores and you don't need to add them in a search and don't want to add them in your keywords either. Waste of time and effort. (now I'm asking for it aren't I?) LOL

and ps, there are other fields for what some have complained, should be available: Probably what the original thread started out asking.
   
Age
   Any
   Baby(0)
   Child(0)
   Teenager(0)
   Adult(0)
   Senior(0)
   See all ...
    
Number of People
   Any
   None(1)
   1 Person(0)
   2 People(0)
   3 People(0)
   4 People(0)
   Groups or Crowds(0)
    
Ethnicity
   Any
   African(0)
   African American(0)
   Caucasian(0)
   Chinese(0)
   Indian(0)
   Japanese(0)
   See all ...
    
Viewpoint
   Any
   Front(0)
   Side(0)
   Rear(0)
   Profile(0)
    
Location
   Any
   USA(0)
   UK(1)
   Europe(0)
    
Orientation
   Any
   Landscape(1)
   Portrait(0)
   Panoramic(0)
   Square(0)
    
Image Type
   Any
   Photographs(1)
   Illustrations(0)
   Cut Outs(0)
    
Date Taken
   Any
   Last 7 days
   Last month
   Last 3 months
   Last 6 months
   Last 12 months
   Last 2 years
   Custom ...
    
Minimum File Size
   Any
   1 MB(1)
   5 MB(1)
   15 MB(1)
   24 MB(1)
   48 MB(1)
   70 MB(0)

ignore the (#) It's just from the image I used for the sample.

[madelaide]

RP,

It's not that we don't understand you, your examples were very clear. It's about Alamy's search tool assumption that one will use quotation marks in a search.

if someone types Mario Andretti, he is not looking for images of Michael Andretti. Unless Mario is in them too, nor for Mario Bros. Now, if no results using the AND operator are found, then the search engine could offer partial matches.

[ShadySue]

and ps, there are other fields for what some have complained, should be available: Probably what the original thread started out asking.

Number of People
   Any
   None(1)
   1 Person(0)
   2 People(0)
   3 People(0)
   4 People(0)
   Groups or Crowds(0)
    
Ethnicity
   Any
   African(0)
   African American(0)
   Caucasian(0)
   Chinese(0)
   Indian(0)
   Japanese(0)
   See all ...
    

But again, these can lead to poor results for the buyer. vide my Japanese drums showing up as Japanese ethnicity.

Also people. For example, my recent good sale had three totally unrecogniseable bits of people. I was going to clone them out, but then I'd have had to mark it as 'manipulated' so not 'strict' editorial - and one, though completely silhouetted, was the driver of the vehicle, so pretty essential (but tiny). You can hardly see them, but they're there and must be accounted for. Fair enough, for MR purposes, but not for search.
But if someone was wanting a photo of 'three people', I can cast iron guarantee they didn't want to see my photo. They wanted three clearly visible people.
For example, someone searches for 'London' and ticks 'three people', they want to see three clear people in London, not a London cityscape and 'spot the indistinguishable people'.
Of course for MR/editorial purposes, you have to state whether or not you have people, buts of people, blurred people, silhouetted people etc. But that doesn't mean that these people should be automatically returned in a search for people.

It's all about the buyer. Not forcing them to try to think of some way of avoiding returns for unrecognisable people.
Steve Krug's Don't Make Me Think should be compulsory reading for everyone involved in search architecture.

[ShadySue]

And today's unfortunate search:
I had a search on 'office party', because I have a photo of the 'head office' of Sinn Fein (caption and the essential keywords) and 'political party' in the main keywords.
On the blog, we had recent remonstration about making sure our keywords don't lead to unwanted searches for the buyer or our AR would suffer. But they're not helping us.
 I agree that we're all suffering in the same way, so it's not as though some ARs will suffer in relation to others.
But the main thing is that it's all about helping, and not p*ssing off the buyer!
And well, clearly two word searches give poor results, but an 'office party' is an 'office party'. It's a 'well-known phrase or saying'.

[ann]

@RacePhoto  - Re:  "Close except it's not Red and Car for the one out of quotes, it's Red OR Car"

Not quite that simple, based on results for below search.

Total results for just red search is significantly more than for red car search.
But if red car = red OR car in searchable fields, then results should be at least 1,312,632.  
<<<Unless I'm totally misunderstanding you, and red OR car is not meant to be same as red AND/OR car. But no one's going to type exactly red car in search box and want results to AVOID all images with both red AND car in searchable fields, so that's not it.>>>

red:             All of Alamy(1,312,632)   Creative(187,468)
car:              All of Alamy(459,632)  Creative(36,705)
red car:        All of Alamy(45,382)   Creative(2,750)   (no quotation marks used in the search)

with good cheer  - Ann

[ShadySue]

And today's worst (not as funny as office party, but it still illustrates the point)
cattle Botswana. My photo was of a "Cattle Egret" in Botswana, both words in the caption and essential keywords, no way out of it. But it certainly wasn't what the buyer wanted.
Hmmmm. I did ethics at Uni (back in the day. I've forgotten it all now!) Interesting question: whether to claw back some money from the excellent charities to invest in marketing and getting a really good search architecture, for possible long-term gain (in a time when returns per image are dwindling). Answer in 5000 words.

[RacePhoto]

@RacePhoto  - Re:  "Close except it's not Red and Car for the one out of quotes, it's Red OR Car"

Not quite that simple, based on results for below search.

Total results for just red search is significantly more than for red car search.
But if red car = red OR car in searchable fields, then results should be at least 1,312,632.  
<<<Unless I'm totally misunderstanding you, and red OR car is not meant to be same as red AND/OR car. But no one's going to type exactly red car in search box and want results to AVOID all images with both red AND car in searchable fields, so that's not it.>>>

red:             All of Alamy(1,312,632)   Creative(187,468)
car:              All of Alamy(459,632)  Creative(36,705)
red car:        All of Alamy(45,382)   Creative(2,750)   (no quotation marks used in the search)

with good cheer  - Ann

I wish I understood what that was. No doubt that you are accurate, but for some reason it should come up with many more for Red Car. Funny that we ended up with that one. What do you get for "Red Car" in quotes.

And ps, it's still a crappy search, I started it, and I'm guilty, but two words is a bad search, waiting to give bad results.

Red = 1,378,075 Results (I have RM and RF checked, not all four boxes)
Car = 459,822 Results (strange that we get different numbers, but no big deal)
Red Car = 45,433 Results (hey look the search is kind of smarter than I thought? red and car?) Not Red OR Car OOPS!
"Red Car" = 3,463 Results (getting better!)

Yes I'll also confess that we assume that someone using the search, knows how to use it. So do you give a gun to someone, and send them out hunting and assume they know how to use it safely? Sorry but the search has the same issues, not as lethal but just as much of a problem. I should have picked a better search!

Part of the argument is, the search should find things that we want and know what the buyer wants (like psychic search engine maybe?) and not give the person searching, what they actually ask for... because it gives bad results. Hey wait, it just gives back, what it does and what it's asked to do.

The problem is the human doing the search, not the search computer. That's been my point from the start.

Yes we can find erroneous search results, we can create bad searches that will give crummy results, but it's the search, not the results that are the problem most of the time. I really don't think that finding some obscure two words that can be confused makes much sense. How many people are looking for that? Wouldn't the logical point be that what you search is what you get?

Ooh Ooh, Michael or Mario or Marco Andretti, what a great search example. No quotes used in any of these:

Mario Andretti = 54 Results (no mario Bros either)
Michael Andretti = 35 Results (no michael jackson or anyone else, just the right one?)
Marco Andretti = 16 Results (no michael, mario, or anyone else, five of my shots)

OK tell me how the search is flawed, when it's producing good results for almost everything we search for, except some very unusual crossed words or dual meanings for two words that might come out in the same search? Office Party! That's a good one! Maybe some other two word search with multiple vague meanings would be another point, but the same.

True Sue: A vague search will give vague results!

[ShadySue]

The problem is the human doing the search, not the search computer. That's been my point from the start.

Ah. The 'myth of the stupid user'.

Originally coined by Jakob Nielsen:
"Wsability is important to me. If the purpose of technology is to make our lives better, then that purpose is subverted when designers fail to pay due regard to the user and the context of use. The software industry has been particularly negligent in this regard, to the extent that a large number of people feel that they are 'too stupid' to handle computers. In effect, these people have been disenfranchised - one of the greatest boons of our times has been denied them because software designers have failed to meet their needs.
For me, usability is about the move to counter this trend. I believe that usability activities can have a direct and beneficial impact on the quality of people's lives. I get enormous satisfaction from doing work that results in products that are easier to use.
I believe that we are far too willing to accept what I call 'the myth of the stupid user'. It's time that we started to reject products that confuse use, are rude to us or in any way belittle us."

"By 'the myth of the stupid user' I mean the belief that users are stupid, and the concomitant inference that inability to use a particular product can be blamed on the user, thus exonerating the product itself and the development team.

Anyone who hangs around with IT professionals knows that many of them consider users to be stupid. Do an internet search for 'stupid users' and you will find thousands of stories about the stupid things people do with their computers. For example, you can read dozens of variations of the story on the person who rang up their ISP and asked 'Is this the Internet?'

This attitude permeates the whole of the IT and web development industry. Many people will be familiar with this attitude from dealing with technical support staff whose technical skills far outweigh their ability to communicate with other humans.

My take on this is that it's a cop-out. Retailers don't treat their customers as if they are stupid. If a shop had a sign that no-one could read, the owners would change the sign, rather than throwing their hands in the air and lamenting the stupidity of shoppers."

See also: http://www.goodexperience.com

Nowadays, if Google does something one way (phrases in "..."), it behooves everyone else to do it the same way. (at the moment. Two, five, ten years hence, who knows).
The visitor should NOT have to learn how to navigate or search in a site. That's the designers' job.

Usability studies with a group of people not connected with the site (possible new buyers would be ideal!) and see how they do searches without instruction other than "think of something you might buy an image of and try to find it", don't help them, don't make suggestions, get them to articulate aloud what they're doing and why (without interrupting), and what they think about the search process and results. Interview them afterwards.
Make changes.

[madelaide]

Red Car = 45,433 Results (hey look the search is kind of smarter than I thought? red and car?) Not Red OR Car OOPS!
"Red Car" = 3,463 Results (getting better!)

(...)

Mario Andretti = 54 Results (no mario Bros either)

"Mario Andretti" = 54 Results

So why there is a difference in the first search with and without quotes, and not in the second?  

In the "red car" results, not everything is a red car either (I saw a few exceptions).

[ShadySue]

In the "red car" results, not everything is a red car either (I saw a few exceptions).

On Alamy, as everywhere else, there are bad keyworders. In fact in the first 72, I only found one which didn't appear to have a 'red car', (the one which shows the red rear lights of a car driving away. I couldn't see the car, but I'm not sure how youd best keyword that image. red trailing light? What would someone likely search on?  rear red car light, maybe, giving rise to 'red car'.

[madelaide]

There was one of a red button on a dashboard (we can't see the color of the car.  And it was in the first page of results (ok, it depends on how it was keyworded).  Also a red present in the trunk of white car.  But I am not complaining of these results, just trying to uinderstand why in one case (red car) the quotation mark made a huge difference but not in the other (Michael Andretti).

[ann]

Most recently, I can't figure how to word description for image of young girl wearing Santa hat so it can be found, but not show up in searches for Santa. (in keywords, Santa only appears in phrase "Santa hat")

RacePhoto:  your Psychic Search Engine - is that what happens when NumbErs and PSYCH get together?

[RacePhoto]

Red Car = 45,433 Results (hey look the search is kind of smarter than I thought? red and car?) Not Red OR Car OOPS!
"Red Car" = 3,463 Results (getting better!)

(...)

Mario Andretti = 54 Results (no mario Bros either)

"Mario Andretti" = 54 Results

So why there is a difference in the first search with and without quotes, and not in the second?  

In the "red car" results, not everything is a red car either (I saw a few exceptions).

Because the search engine doesn't write the keywords!

As for Mario, probably my error when I cut and pasted. Oops, stupid users syndrome. Nope, just that Mario Andretti is unique enough that both produce the same results. Which by the way, is contrary to the contention that the search finds things with one word out of two.

But here I go again. Three Word Searches! Two words aren't enough.

Bad searches make for bad results. Why can't I get that point across, and then people keep coming back with examples of two unrelated vague words, that don't produce an expected, vague and obscure result instead of the logical common things connected to those TWO WORDS! Many words have more than one meaning. There's no point in arguing that the search for ambiguous words and terms, will obviously produce, ambiguous and vague results.

The uneducated user is not a myth. Computers and technology have proven this to be even more true than ever before. I get phone calls and emails daily that document this.

If I perfect the psychic search engine, I'll be richer than Google! Hey wait, there's an idea, try your "office party" in Google and see if you get something about a political office?

Do we want to continue with Red Car, which I confess is not a great search to start with? Yes, red tail light on car will come up. Red Box behind a white car will come up. But most of the time, the highest percentage of the images are Red Cars, aren't they?

Ann, I don't think you can have a word in your keywords and not have it show in searches according to a specific "not" at the same time. If you ask for Santa Hat, you are going to get Santa and Hat results, unless you trust the buyer/searcher to do something as simple as search for - girl wearing santa hat - http://tinyurl.com/26jvbdl  hey look, 805 images of girls wearing santa hats. It's easy. (No quotes were harmed in making this search)

[ann]

  [....]

Ann, I don't think you can have a word in your keywords and not have it show in searches according to a specific "not" at the same time. If you ask for Santa Hat, you are going to get Santa and Hat results, unless you trust the buyer/searcher to do something as simple as search for - girl wearing santa hat - http://tinyurl.com/26jvbdl  hey look, 805 images of girls wearing santa hats. It's easy. (No quotes were harmed in making this search)

Thanks for posting that link to search results for girl wearing santa hat, RacePhoto - since my photos showed up on first page, I guess their keywords are okay. Earlier today, I took the word Santa out of title, just referring to hat as Red and White Christmas Hat, but change didn't kick in yet. I'm going to put it back to original wording.

[donding]

Earlier today, I took the word Santa out of title, just referring to hat as Red and White Christmas Hat, but change didn't kick in yet. I'm going to put it back to original wording.

Doesn't it take 24 hours for it to be updated in the search engine after the change? I think I read that somewhere.

[ann]

Earlier today, I took the word Santa out of title, just referring to hat as Red and White Christmas Hat, but change didn't kick in yet. I'm going to put it back to original wording.

Doesn't it take 24 hours for it to be updated in the search engine after the change? I think I read that somewhere.

I think you're right. Since I saw RacePhoto's post within hours of the change, I was able to refer to images' live online description and essential keywords when re-editing the info     (?I thought your username was dongding, like a bell in reverse.)

[donding]

Earlier today, I took the word Santa out of title, just referring to hat as Red and White Christmas Hat, but change didn't kick in yet. I'm going to put it back to original wording.

Doesn't it take 24 hours for it to be updated in the search engine after the change? I think I read that somewhere.

I think you're right. Since I saw RacePhoto's post within hours of the change, I was able to refer to images' live online description and essential keywords when re-editing the info     (?I thought your username was dongding, like a bell in reverse.)

Nope it's donding....I picked that name up many years ago when my sister's screen name was Dordo and I tryed to take the screen name of dingdong. I couldn't use dingdong so adopted donding...
Real name Donna...

[madelaide]

The only reason that occured tome for the Mario Andretti vs "Mario Andretti" is that maybe all the images have "Mario Andretti" with quotation marks, and when exact matches are found the search engine doesn't look further - but this is just a theory.  

I was checking another example (sorry, RP, two words, as in the case of most people's names): Westminster Cathedral (a catholic church in London, not Westminster Abbey). Most images are the Cathedral. Some are the Abbey but the caption says Westminster Cathedral, so people who keyworded them did not know it well.
Westminster Cathedral: 1284 results
"Westminster Cathedral": 589 results, still some are the Abbey with wrong caption

Julian Assange: 74
"Julian Assange": 73

Smithsonian Museum: 1998, the first pages I saw seemed all relevant
"Smithsonian Museum": 354

Desert Edelweiss: 6 images
"Desert Edelweiss": 4 images (all mine )

[RacePhoto]

Entertaining isn't it? With a bunch of us testing and trying, it teaches us how to write better keywords for the Alamy searches. I've been doing all kinds of experiments since I started uploading there. Things like putting keywords as the first two in a box and then moving them to last to in the same box. Take a different photo (identical subjects from same day, same event, same driver Etc.) and change the order of those two words. Wait at least a day or past Midnight here = 7AM there, one day later and see what it does for the placement and order they show up.

The question about changes reminded me of this. They resort the database once a day that I can see. I haven't tested hour by hour.

Back to those other two, word order and proximity and they do make a difference. So it's not just finding words and what box they are in.

Then the one that I think is most favorable for all of us. The diversity algorithm. Where all of the one persons images don't show up in a pack. We get spread out across the search, which means all of us get better views for the same search, not just the person with the highest rating or most photos of that subject.

Something about the way they are counting views and zooms has also changed. My view went through the floor a few months ago. Either that or people just stopped searching for everything I shoot, suddenly? My CTR right now is .74

With all of that, Alamy rank means less and less and I don't really think we should worry so much about errant views from poor searches. We all get them.

It's been fun!  

The only reason that occured tome for the Mario Andretti vs "Mario Andretti" is that maybe all the images have "Mario Andretti" with quotation marks, and when exact matches are found the search engine doesn't look further - but this is just a theory.  

I was checking another example (sorry, RP, two words, as in the case of most people's names): Westminster Cathedral (a catholic church in London, not Westminster Abbey). Most images are the Cathedral. Some are the Abbey but the caption says Westminster Cathedral, so people who keyworded them did not know it well.
Westminster Cathedral: 1284 results
"Westminster Cathedral": 589 results, still some are the Abbey with wrong caption

Julian Assange: 74
"Julian Assange": 73

Smithsonian Museum: 1998, the first pages I saw seemed all relevant
"Smithsonian Museum": 354

Desert Edelweiss: 6 images
"Desert Edelweiss": 4 images (all mine )

[ann]

You put things nicely in perspective, RacePhoto, and, yes, it is both helpful and fun to share trials and tribulations of keyword searches.

Reading some of madelaide and ShadySue's search results experiences reminds me of fun project my writing students would do. They'd illustrate a sentence with a misplaced modifier, and then illustrate the sentence with the error fixed, to show how incorrect wording unintentionally alters meaning.

     NO:   Hundreds of thousands of miles away, Pat looked through his telescope
     at the lunar eclipse.  

     YES:  Looking through his telescope, Pat saw the lunar eclipse
     hundreds of thousands of miles away.

     (still have the great drawings of Pat far away in space looking through telescope,
     and then Pat safely here on earth...)


search:  "lunar eclipse"  telescope

results: http://tinyurl.com/25fyhfg  

Entertaining isn't it? With a bunch of us testing and trying, it teaches us how to write better keywords for the Alamy searches.
[....]

Something about the way they are counting views and zooms has also changed. My view went through the floor a few months ago. Either that or people just stopped searching for everything I shoot, suddenly? My CTR right now is .74

With all of that, Alamy rank means less and less and I don't really think we should worry so much about errant views from poor searches. We all get them.

It's been fun!