010607.html | WebReference


((((((((((((((((( WEBREFERENCE UPDATE NEWSLETTER ))))))))))))))))) June 7, 2001


This newsletter is sponsored by: American Express Corporate Portal Evaluation & Selection Seminar __________________________________________________________________

This week open publisher Tony DeYoung looks into visual search engines (VSE). VSE use pictures to locate similar pictures, and can dramatically increase relevance over keyword searches.

http://www.webreference.com *- link to us today http://www.webreference.com/new/ *- newsletter home http://www.webreference.com/new/submit.html *- submit article

New this week on WebReference.com and the Web:

1. FEATURED ARTICLE: Visual Search Engines 2. OTHER VOICES: * The Net Effect: May the Best Interface Win! * The next phases of Web development * Grow Your Site, Keep Your Users * GPL, Legally Speaking 3. NET NEWS: * Failure to act fast takes people out of their domain * Landmark case of Internet publishing * Save As XML Plug-In for Macintosh * Mighty morphin' Internet billboards are coming to a mall near you

/-------------------------------------------------------------------\ What's new on the Net? How about American Express? When you work at American Express Technologies, you'll be driving the innovations that move a global financial leader forward at the speed of light. The same people who pioneered the online wallet are always looking for new ways to leverage the power of the Internet. If you've got vision, creativity and the burning desire to work with the top people and technology in the business, this is the place for you. http://ilib.com/!webref6775/www.americanexpress.com/ITcareers


Like what you see? Get our front page e-mailed to you every business day with our HTML newsletter. Just send an e-mail to:


or for this text newsletter:


Spread the word! Feel free to send a copy of this newsletter to your friends and colleagues, and while you're at it, snap a link to WebReference.com.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 1. FEATURED ARTICLE: Visual Search Engines

Search technology may be the foundation of the Internet, but if you're looking for rich media content, today's text and keyword- based searches are woefully inadequate.

For photography, graphics, logos, audio or video, text descriptions rarely convey the most valuable information. For example, if you're looking for a certain shade of blue sky, it's nearly impossible to find a match in a stock photo library using only textual descriptions. Or if you are using a search engine to look for photographs of tigers on the Web, what keywords do you to get less than 10,000 search results? And how do you ensure that those search results are relevant and worth filtering through?

E-commerce sites face the same difficulty. Every try to find something on eBay? Even after you locate one1 item that looks like it might be interesting, how do you find other similar items to comparison shop? Invariably you have to manually search through scores of mostly irrelevant or undecipherable photographs. Shoppers like me have a low tolerance for this and quickly leave. Maybe this is one of the reasons why only 3% of today's online lookers become buyers.

Even with the most meticulously keyworded content, searching with text inevitably means that a Web site visitor must know about the keywords used by that site or master a complex syntax for specifying non-trivial searches. (e.g. I wrote to a stock photo site looking for help on finding an image of a cowboy. They responded back that I should type "cowboy =man cowboy =male cowboy =portrait cowboy =close." Now that is intuitive...NOT.) And of course text-based search engines only work if you speak fluent English.

As Web site content grows exponentially in non-textual types of information, it is apparent that text-based search engines are becoming less equipped to provide good results.

In response to this deficit, several university labs and commercial software companies have developed tools that allow a visual search of images and products. With visual search, a user can make selections based on images rather than text.

Most of these systems operate in a similar way: the user performs a query by choosing an image which is somewhat similar to the desired images and then the engine does a pattern recognition search using global/local comparisons of color, shape or texture. So for example, you find a sample of a sunset and then ask the search engine to find images with similar red and gold colors.

This approach works when the entire image scene is distinctive and relevant, but it gives a lot of obviously wrong results for complex images or large databases.

I recently began experimenting with a beta release of a visual search engine (VSE) toolkit that implements an object-based approach to improve the accuracy of visual search results. The Java-based toolkit is by eVision (http://www.evisionglobal.com) and is available as a free download. Even if you are not a top notch Java programmer, you can produce basic visual search applications using just the high-level API.

The approach that eVision takes is to:

1) treat photographs as a collection of objects, rather than as one big undifferentiated image and 2) make it easier for users to more clearly specify what they mean by "this" when they tell the search engine to "find something that looks like THIS".

>Object-based searches

When we look at photographs, we look for patterns and objects. We identify a photograph that is 10% brown and 90% green as a brown horse in a grassy field. So when searching for similar images, we would not be confused by a photograph of a green river dotted with 10% brown fallen tree branches. But general-purpose VSEs could identify a horse in the field and tree branches in a green river as very similar. They look at the image as one big undifferentiated group of RGB values.

An object-based VSE like eVision, tries to first identify the objects in an image before doing a comparative search. While it can't attribute the meaning of horse to the brown object, it can say that the photo is composed of two distinct objects - a brown one with a particular shape and a green background. Then it runs visual comparisons to other images based on these regions.

For example, with the photo of a horse in a green field and a color similarity search, a general-VSE would say "This photo has 90% green in it and 10% brown so find photos that have this same proportion of colors." eVision would say "This photo has two objects in it, 1 object is 100% green and the other is 100% brown so find photos that contain a 100% green object and a 100% brown object." For the non-object way, you would get horses in fields, a forest (trunks are brown), brown scum in a green river, a green lawn covered in 10% dog droppings etc. With eVision you would get horses in fields, a horse-sized dog on a grassy background etc. The latter matches are certainly closer to the sample image and much much more like the way humans see things. We see objects, not distributions of color.

/-------------------------------------------------------------------\ Get it right the first time with the right tools & attend the Corporate Portal Evaluation & Selection Seminar. This seminar addresses in detail the roles and categories of corporate portals, selection and evaluation mechanisms, and an in-depth examination of leading portal products and platforms. Come learn the framework for making informed purchasing decisions and anticipating implementation issues. Register today: www.imgevents.com/portal_seminar2 7/18-Atlanta; 7/26-NY,NY


>The importance of the User Interface

One difficulty with visual search is that while a person knows what objects are important when they are looking for similar images, the VSE generally has no idea. Similarly when a VSE returns search results, we have little feedback on how the computer came up with them. This can get frustrating when the results are not what we expect.

eVision helps resolve this problem by letting the user see what the VSE interprets as objects (the object map). And it lets the user specify which objects in the object map are relevant to a search.

So in our horse in a field example, if the grassy field is unimportant to the search, the user can select only the horse object and tell the VSE to search for horse shaped objects, regardless of background. Similarly, by looking at the object map that the VSE displays, the user can understand how it came up with its search results and use this information to refine the search.

Even with the assistance of object maps, some significant UI problems still can tax the stamina of the end user of visual search. One of the toughest challenges I have encountered is knowing what features to search on. For example, when I find a sample image of a horse and want to search for other horse images, should I search on color, or shape, or texture or 3D shadowing - or some combination of features? Does it matter to me if the horse is brown or spotted? Facing forward or to the side? I've played with this for a while and I'm still getting the hang of it. This is obviously a user interface issue that developers more clever than I can help address.

Other issues include the speed of analyzing new images. It is a slow process. Maybe better hardware would accelerate this. On the other hand, the visual searching itself is exceptionally fast. Using my sample PhotoDisc 500 image database, visual search results came up almost as fast as keyword search results.

The current beta version of the eVision toolkit targets photographs and graphics. But the Web site mentions that the same technology can search video (by searching on video keyframes) and audio (by looking at audio waveform signatures). Imagine being able to go to amazon.com, listen to a song that you like and tell the Amazon VSE to find other music that had a similar audio signatures - in other words, find other music that I also might like. The implications are pretty compelling.

No search tool is perfect, but searching the Web or a media database for images or video is such a chaotic process that almost any advance on current technology is likely to be beneficial. In conjunction with existing text-based search, visual search will make finding what you're looking for a more intuitive, successful and less frustrating experience. Now search tools can literally, "see what you mean."

Further Reading:

Berkeley on blobworld (http://elib.cs.berkeley.edu/photos/blobworld/) and USSB on Netra (http://maya.ece.ucsb.edu/Netra/index2.html)

# # # #

About the author: Tony DeYoung is a Web developer and technophile living in San Francisco. As an independent contractor, he makes his living finding new media technologies and developing applications or consulting services around them. His latest venture, sonify.org (http://www.sonify.org) is a developer- oriented Web site devoted to interactive audio for the Web and mobile devices. He holds advanced degrees in psychology and statistics from Yale and Stanford. He can be reached at tony@sonify.org

/-------------------------------------------------------------------\ **JOB SECURITY -- CAREER GROWTH -- CHALLENGING POSITIONS** The internet.com Careers Channel is powered by dice.com, the leading online Information Technology (IT) job board. Whether you need to start your new job today, are searching for your dream job, or are just wondering what your skills are worth, you'll find the tools you need to land your next great job. Don't wait any longer! http://www.internet.com/sections/careers.html


~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 2. OTHER VOICES: The Net Effect: May the Best Interface Win!, The next phases of Web development, GPL, Legally Speaking, Grow Your Site, Keep Your Users

>The Net Effect: May the Best Interface Win!

Most companies still have tremendous difficulty building Web sites that are easy to use - despite years and billions of dollars spent trying. http://www.techreview.com/magazine/jun01/garfinkel.asp MIT Technology Review, June 2001

>The next phases of Web development

Despite the downturn in the Internet economy, Web technology is alive and well. What will be the next phase of Web development? http://www.canadacomputes.com/v3/story/1,1017,6734,00.html CanadaComputes.com, May 31, 2001

>Grow Your Site, Keep Your Users

How does eBay scale and retain its massive user base? http://www.usability.gov/ http://www.computerworld.com/cwi/story/0,1199,NAV47_STO60997,00.html Computer World, June 4, 2001

>GPL, Legally Speaking

An attorney's cure for corporate's unreasonable fear of Open Source infection: the truth. By Lawrence Rosen. http://www.webmasterbase.com/article.php/415 http://www.openmagazine.net/featured/01/04/18/1529243.shtml Webmasterbase.com

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 3. NET NEWS: Failure to act fast takes people out of their domain, Landmark case of Internet publishing, Save As XML Plug-In for Macintosh, Mighty morphin' Internet billboards are coming to a mall near you

>Failure to act fast takes people out of their domain

Neglect to pay the mortgage and you lose your home. But what if you never got the eviction notice? What if you just came home from work one night and found the new owners sitting in the dining room and eating your food? http://www.usatoday.com/life/cyber/tech/2001-06-04-domain-name-woes.htm USA TODAY, June 5, 2001

>Landmark case of Internet publishing

A 19th century law suit involving a British nobleman, his servant and a London newspaper was cited this week as a precedent in a landmark case on Internet publishing. http://www.smh.com.au/news/0106/06/pageone/pageone6.html Sydney Morning Herald, June 5, 2001

>Save As XML Plug-In for Macintosh

Adobe Save As XML is a free plug-in for Adobe Acrobat designed to help publishers conserve time by repurposing content. The plug-in allows content in tagged Adobe PDF files to be saved in Web friendly formats. http://www.adobe.com/support/downloads/89a6.htm http://www.xml.com/pub/r/1121 XML.com, June 4, 2001

>Mighty morphin' Internet billboards are coming to a mall near you

I-Open, a Pennsylvania company, has developed a technology that allows retailers such as Saks Fifth Avenue and Nine West to use the Internet to deliver precisely targeted messages to consumers via high-definition, flat-screen displays at their retail stores. http://www.zdnet.com/ecommerce/stories/main/0,10475,2767992,00.html Interactive Week, June 4, 2001

That's it for this week, see you next time.

Scott Clark Managing Editor, WebReference.com sclark@internet.com

Dan Ragle Assistant Editor, WebReference.com dragle@internet.com

Andrew King Newsletter Editor, WebReference.com aking@internet.com

webreference-update-text Is Powered By http://By.SparkLIST.com/ In-house vs. Outsourced Email List Hosting: Are you paying too much for your email list management? Download the FREE REPORT: http://SparkLIST.com/free-report/

webreference-update-text Is Powered By http://By.SparkLIST.com/ SparkLIST delivers high volume fast and reliable email newsletter outsourcing which allows you to focus on your core business. Contact us: mailto:free-quote@sparklist.com

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Advertising: If you are interested in advertising in our newsletters, call Frank Fazio on 1-203-662-2997 or send email to mailto:ffazio@internet.com ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ For contact information on sales offices worldwide visit http://www.internet.com/mediakit/salescontacts.html ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ For details on becoming a Commerce Partner, contact David Arganbright on 1-203-662-2858 or mailto:commerce-licensing@internet.com ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ To learn about other free newsletters offered by internet.com or to change your subscription visit http://e-newsletters.internet.com ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ internet.com's network of more than 150 Web sites are organized into 16 channels: Internet Technology http://internet.com/it E-Commerce/Marketing http://internet.com/marketing Web Developer http://internet.com/webdev Windows Internet Technology http://internet.com/win Linux/Open Source http://internet.com/linux Internet Resources http://internet.com/resources ISP Resources http://internet.com/isp Internet Lists http://internet.com/lists Download http://internet.com/downloads International http://internet.com/international Internet News http://internet.com/news Internet Investing http://internet.com/stocks ASP Resources http://internet.com/asp Wireless Internet http://internet.com/wireless Career Resources http://internet.com/careers EarthWeb http://www.earthweb.com ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ To find an answer - http://search.internet.com ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Looking for a job? Filling an opening? - http://jobs.internet.com ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ This newsletter is published by Jupitermedia Corp. http://internet.com - The Internet & IT Network Copyright (c) 2001 Jupitermedia Corp. All rights reserved. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ For information on reprinting or linking to internet.com content: http://internet.com/corporate/permissions.html ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~