Ask three keywording companies for Getty or Alamy keywording, or keywording appropriate for your own web site, and you're guaranteed to get three very different results.
A client of ours recently shopped around some of the biggest names in keywording to see just how standards differed. Although asking for exactly the same keywording "standards" from each company, they found remarkable differences in what was offered under the same banner.
The areas they found the most variance in were (in order of importance):
1. Specificity of keywords - Some companies delivered keywords which were too general, and did not capture the essence of each image. Getting to grips with what makes an image special is what delivers the strongest search results.
2. Over-keywording - Call it keywording spam if you want to be more brutal. Some of the companies just didn't know when to stop, seeing all sorts of possibilities in the images which rendered the keywording less effective, not more.
3. Poor conceptual keywording - The ability to include appropriate concepts such as victory, beauty in nature or romance, is crucial for images or video to be found and sold. Deep knowledge of English is needed to get the best results, and it showed.
4. Confused presentation - One company in particular provided keywords in such a bewildering array of lists of synonyms, which seemed to overlap, that it was hard to understand what it all meant.
As a general rule, the more money you pay for your keywording the less likely you are to find inadequacies in these areas. Excellent English language skills of the keyworders are vital.
It is best in every case to shop around and get samples done to your specific requirements. Take the time to make sure the company knows what those are. It is a good sign if you are able to get into a dialogue with the keywording company and find out what they understand by the particular standard required.
Saturday, 12 December 2009
Getting Short Changed With Your Keywording Standard?
Posted by
Kevin Townsend
at
8:44 PM
0
comments
Labels:
avoiding pitfalls,
keyword spam,
keywording standards
Friday, 4 December 2009
Piksee Software Speeds Live Image Workflow
News and celebrity photo agencies will soon have a time-efficient way to add keywording to their live image production, thanks to the advent of Piksee workflow software.
By taking a fresh approach to the problem of organising and processing image sets, Piksee - developed by Keedup - can handle large volumes of pictures with minimal loss of time, whilst also reducing internet data traffic by 90%.
Where news and celebrity images once went out to clients and sub-agents with hardly any metadata, there is now the opportunity to get the information into the IPTC fields of the images from the beginning.
The breakthrough with Piksee is that it allows keywords to be added to high-resolution images without them having to leave our clients’ servers. Previously the only way to keyword images where they stood, was by using a VPN or browser system, normally one image at a time which is extremely slow.
To do the keywording itself, it is far faster to use Keedup's customised keywording systems, but that has meant having high-resolution images sent by FTP, and returned the same way. Piksee uses a system of sampling the high-res images and writing back only the tiny amount of actual metadata. It also avoids having to use clumsy and slow FTP.
Keedup developers studied the sort of production line principles used by companies such as Toyota so that image sets were “assembled” automatically and processed in an orderly fashion, syncing seamlessly with Keedup’s keywording software.
When they studied the workflow process they found time was being lost in numerous areas such as in the time it took for an FTP server to recognise an image was ready to be sent. Plus when large numbers of images were being transmitted, sets of images often took too long to be received because part of the set was stuck way back in the queue.
Piksee has been written in Java to operate with Mac and PC, and the emphasis is on versatility and simplicity. It can be installed and operated with minimum of fuss. If the keywording software were to be changed or updated, Piksee would work with that. No complicated database or software is needed at the client end.
Everything is controlled by Keedup's Image-Eye Server which is similar to the All Seeing Eye software used by gaming companies to allow players to interact from anywhere in the world. So distance is not a barrier to speed.
This also means it is easy for Piksee to be plugged into existing client workflows, something that photo agencies are very concerned about.
Beyond the speed and efficiency gains, Piksee gives clients an easy way to monitor progress with their own client window, and allows detailed performance monitoring by Keedup staff.
Piksee has taken a year to write and test, and is already being deployed for one of Keedup’s major clients.
For more information, contact Kevin Townsend kevint@keedup.com
By taking a fresh approach to the problem of organising and processing image sets, Piksee - developed by Keedup - can handle large volumes of pictures with minimal loss of time, whilst also reducing internet data traffic by 90%.
Where news and celebrity images once went out to clients and sub-agents with hardly any metadata, there is now the opportunity to get the information into the IPTC fields of the images from the beginning.
The breakthrough with Piksee is that it allows keywords to be added to high-resolution images without them having to leave our clients’ servers. Previously the only way to keyword images where they stood, was by using a VPN or browser system, normally one image at a time which is extremely slow.
Piksee in operation at Keedup's Auckland office
To do the keywording itself, it is far faster to use Keedup's customised keywording systems, but that has meant having high-resolution images sent by FTP, and returned the same way. Piksee uses a system of sampling the high-res images and writing back only the tiny amount of actual metadata. It also avoids having to use clumsy and slow FTP.
Keedup developers studied the sort of production line principles used by companies such as Toyota so that image sets were “assembled” automatically and processed in an orderly fashion, syncing seamlessly with Keedup’s keywording software.
When they studied the workflow process they found time was being lost in numerous areas such as in the time it took for an FTP server to recognise an image was ready to be sent. Plus when large numbers of images were being transmitted, sets of images often took too long to be received because part of the set was stuck way back in the queue.
Piksee has been written in Java to operate with Mac and PC, and the emphasis is on versatility and simplicity. It can be installed and operated with minimum of fuss. If the keywording software were to be changed or updated, Piksee would work with that. No complicated database or software is needed at the client end.
Everything is controlled by Keedup's Image-Eye Server which is similar to the All Seeing Eye software used by gaming companies to allow players to interact from anywhere in the world. So distance is not a barrier to speed.
This also means it is easy for Piksee to be plugged into existing client workflows, something that photo agencies are very concerned about.
Beyond the speed and efficiency gains, Piksee gives clients an easy way to monitor progress with their own client window, and allows detailed performance monitoring by Keedup staff.
Piksee has taken a year to write and test, and is already being deployed for one of Keedup’s major clients.
For more information, contact Kevin Townsend kevint@keedup.com
Saturday, 28 November 2009
Keywording Standards? Nobody Knows
So what is the industry standard for keywording? When that question is raised, the appropriate answer is - in the words of Agatha Christie's Miss Marple - "nobody knows".
Due to it's unregulated do-it-yourself nature, keywording standards have developed to meet the needs of the particular photo or video libraries concerned, rather than as part of an industry-wide standard.
The result is some front runners (notably the so-called Getty standard), but many, many other standards with different sorts of rules and controlled vocabularies, or no set vocabularies at all.
Whilst the Getty Images standard has much to recommend it - particularly its emphasis on including only the most pertinent terms - the original keywording submitted by photographers and photo libraries (often with the help of professional keywording contractors) is designed primarily to fit into the Getty system - being absorbed into a much larger vocabulary. The result is the inclusion of some slightly peculiar terms such as human hand and full suit (jacket and pants). This keywording is therefore not ideally suited to normal intuitive searching by ordinary people more used to looking for hands and suits.
Outside of Getty, there are numerous standards such as for Alamy and Corbis. The irksome thing is that the difference between these standards is often stylistic rather than substantive. For instance, ages and ethnicities are often referred to with different terms which essentially mean the same thing, but are unacceptable to other companies. Some keywording standards involve multiple fields containing keywords, whilst most only have one - more complication.
Cynical observers might think that this is an attempt by the large photo libraries to make it difficult for images to be offered to competitors, by making it expensive to re-keyword images to submit in alternative keywording styles.
More likely, it is simply a case of no regulatory body and the historical rise and fall of various players in the market and with them their particular ideas about keywording.
Ironically, given that keywording is so un-standardised, people often have strong views about what constitues good keywording. We recently had a client ask for keywording to an "impressive standard" - and who wouldn't want that. The only problem is that no such standard exists, and if you asked two people in the photo business to define it, they wouldn't come up with the same answer even if they could actually define a standard.
In amongst all this is an ongoing battle between two major camps: those who believe less is more (eg Getty) and those who believe more is more. So the same set of keywords can be praised for being inadequate because they are over-keyworded with too many words of minor relevance, or not keyworded enough because the keywords do not include every single possible word that might be associated with the image.
The easiest way to view a standard is a set of rules setting out the constraints and prescriptions for how keywording should be entered. These rules include such imperatives as "no plurals", or "only use verbs in present tense continuous" (running as opposed to run). They may even go so far as to tell you exactly what words you can use - a controlled vocabulary. Any words not in the vocabulary are not permitted.
Controlled vocabularies give a lot of consistency, but run into a major problem - language is a moving and growing target. Words that are in common usage are changing all the time, whilst new words enter the language every day. These new words are not necessarily slang or common words, more likely they are things such as place names, people's names, brands and other proper nouns. Even sheer size of the language is a problem. Whilst the contents of the Oxford dictionary might be considered "the English language", it is only a fraction of all the words available when you add in those place names, species and so on.
In attempting to make sense of such a difficult thing as language, keywording companies sometimes talk (with religious zeal in some cases) about keywords and synonyms as if there is a set of "main" words, and then their synonyms. This is obviously nonsensical to anyone with high school English. Two words such as "pants" and "trousers" are synonyms of each other, but deciding one is the primary word is totally arbitrary.
In getting advice about keywording, or picking a contractor to do your keywording, be wary of people who try to tell you there is an industry standard to which they adhere - this is just a sales ploy. Likewise, be worried about people talking about keywords and synonyms - this is more hocus pocus.
Whilst it goes against the grain in some ways, a keywording company which has an honest appraisal of the uncertainties of keywording is far more likely to do a good job than one which tells you it will meet the industry standard (the one that doesn't exist). One way tounderstand what sort of company you are dealing with is to ask them if there is an industry standard or get them to explain how they deal with the expanding language. The more ready they are to work with you to fit the sort of style you want, the more likely they are to give you what you want. If they have developed their own standards and vocabularies that is a good sign as it shows at least the company realises that standards ar not set in stone, and have doen the hard yards producing a standard of their own.
Due to it's unregulated do-it-yourself nature, keywording standards have developed to meet the needs of the particular photo or video libraries concerned, rather than as part of an industry-wide standard.
The result is some front runners (notably the so-called Getty standard), but many, many other standards with different sorts of rules and controlled vocabularies, or no set vocabularies at all.
Agatha Christie's Miss Marple - "nobody knows"
Whilst the Getty Images standard has much to recommend it - particularly its emphasis on including only the most pertinent terms - the original keywording submitted by photographers and photo libraries (often with the help of professional keywording contractors) is designed primarily to fit into the Getty system - being absorbed into a much larger vocabulary. The result is the inclusion of some slightly peculiar terms such as human hand and full suit (jacket and pants). This keywording is therefore not ideally suited to normal intuitive searching by ordinary people more used to looking for hands and suits.
Outside of Getty, there are numerous standards such as for Alamy and Corbis. The irksome thing is that the difference between these standards is often stylistic rather than substantive. For instance, ages and ethnicities are often referred to with different terms which essentially mean the same thing, but are unacceptable to other companies. Some keywording standards involve multiple fields containing keywords, whilst most only have one - more complication.
Cynical observers might think that this is an attempt by the large photo libraries to make it difficult for images to be offered to competitors, by making it expensive to re-keyword images to submit in alternative keywording styles.
More likely, it is simply a case of no regulatory body and the historical rise and fall of various players in the market and with them their particular ideas about keywording.
Ironically, given that keywording is so un-standardised, people often have strong views about what constitues good keywording. We recently had a client ask for keywording to an "impressive standard" - and who wouldn't want that. The only problem is that no such standard exists, and if you asked two people in the photo business to define it, they wouldn't come up with the same answer even if they could actually define a standard.
In amongst all this is an ongoing battle between two major camps: those who believe less is more (eg Getty) and those who believe more is more. So the same set of keywords can be praised for being inadequate because they are over-keyworded with too many words of minor relevance, or not keyworded enough because the keywords do not include every single possible word that might be associated with the image.
The easiest way to view a standard is a set of rules setting out the constraints and prescriptions for how keywording should be entered. These rules include such imperatives as "no plurals", or "only use verbs in present tense continuous" (running as opposed to run). They may even go so far as to tell you exactly what words you can use - a controlled vocabulary. Any words not in the vocabulary are not permitted.
Controlled vocabularies give a lot of consistency, but run into a major problem - language is a moving and growing target. Words that are in common usage are changing all the time, whilst new words enter the language every day. These new words are not necessarily slang or common words, more likely they are things such as place names, people's names, brands and other proper nouns. Even sheer size of the language is a problem. Whilst the contents of the Oxford dictionary might be considered "the English language", it is only a fraction of all the words available when you add in those place names, species and so on.
In attempting to make sense of such a difficult thing as language, keywording companies sometimes talk (with religious zeal in some cases) about keywords and synonyms as if there is a set of "main" words, and then their synonyms. This is obviously nonsensical to anyone with high school English. Two words such as "pants" and "trousers" are synonyms of each other, but deciding one is the primary word is totally arbitrary.
In getting advice about keywording, or picking a contractor to do your keywording, be wary of people who try to tell you there is an industry standard to which they adhere - this is just a sales ploy. Likewise, be worried about people talking about keywords and synonyms - this is more hocus pocus.
Whilst it goes against the grain in some ways, a keywording company which has an honest appraisal of the uncertainties of keywording is far more likely to do a good job than one which tells you it will meet the industry standard (the one that doesn't exist). One way tounderstand what sort of company you are dealing with is to ask them if there is an industry standard or get them to explain how they deal with the expanding language. The more ready they are to work with you to fit the sort of style you want, the more likely they are to give you what you want. If they have developed their own standards and vocabularies that is a good sign as it shows at least the company realises that standards ar not set in stone, and have doen the hard yards producing a standard of their own.
Friday, 20 November 2009
New Zealand the Least Corrupt Country in the World
New Zealand has sneaked ahead of Denmark and Sweden as the least corrupt country in the world.
Rankings produced by Transparency International in its latest Corruption Perceptions Index show New Zealand has gained a mere 10th of a point to top the ladder. Major western countries the United States and Britain came in at 19th and 17th least corrupt, while economic powerehouses India and china earned the dubious honour of being at 84th and 79th place respectively.
Last year, Denmark, Sweden and New Zealand shared top place with a score of 9.3. This year New Zealand has moved up to 9.4, with Denmark remaining at 9.3. Sweden slipped to 9.2 to share third with Singapore while Switzerland is next at 9.0.
New Zealand- Home of keywording company Keedup Ltd
Transparency International, a Berlin-based organisation, says these scores “reflect political stability, long-established conflict of interest regulations and solid, functioning public institutions.”
The vast majority of the 180 countries included in the 2009 index score below five on a scale from 0 (perceived to be highly corrupt) to 10 (perceived to have low levels of corruption).
“Fragile, unstable states that are scarred by war and ongoing conflict linger at the bottom of the index,” Transparency International says.
These are: Somalia, with a score of 1.1, Afghanistan at 1.3, Myanmar at 1.4 and Sudan tied with Iraq at 1.5.
“These results demonstrate that countries which are perceived to have the highest levels of public-sector corruption are also those plagued by long-standing conflicts, which have torn apart their governance infrastructure.”
In other top 10 placings, Australia and Canada have moved up from ninth to eighth, while Sweden has slipped from first equal to third and Iceland has dropped to eight equal from seventh.
Among major countries, the UK remains at all-time low score of 7.7, ranking it 17th out of 180, against 16th in 2008. The US is one place behind at 7.5, up from 7.3 last year and placing it at 19th. Japan is tied with the UK at 7.7, up from 7.3 last year.
Subscribe to:
Posts (Atom)


