A Picture’s worth a Thousand Words

A Picture is Worth a Thousand Words

Group Ten

Brad Wright, Jason Tipton, Brandon Day, Nathan Kinard, Nick Ponder

Key Words: Riya, Algorithms, Likeness Search, Visual Signature, Text Search.

Everyone has heard the old saying, “A picture is worth a thousand words.” Well a new software company called Riya has put this adage to good use. They have succeeded in producing the very first truly visual search engine that is available at their website like.com. In short, like.com allows people to search for purses, jewelry, shoes, and other apparel products based on searches using pictures rather than text. This visual search engine is a novel idea that provides the customer with an easier path to finding specific products than what is available through traditional search engines. Riya is the first company to harness this idea, however, it can easily be inferred that other companies will follow suit, and those that do will have an advantage over companies who solely offer text searches.

Originally, Riya’s course of action for the use of their visual search was in the area of facial recognition. However, the facial recognition had one major flaw; it was not profitable. Riya conducted consumer studies about the facial recognition software on sites such as MySpace.com, and they were less than enthusiastic about the results. Almost as an afterthought, they provided a product search, and were greatly surprised by the popularity of such an application. Riya decided to reroute their efforts to create a visual product search that they named like.com, which specializes in aesthetic products such purses, jewelry, and clothing. This allowed them to enter into a $15 billion online industry that solved their path to profitability problems.

There are other image search engines on the web, but none of them have the capabilities of like.com. Image searches such as those at google.com and pixsy.com still search based on text. A text is entered into their engine, and the site searches for the same text that has a photo assigned to it. Hence, they are still text searches only the text files have pictures attached. The images attached often have no relationship to their text. Like.com has alleviated this problem though their visual signatures.

The visual signature is merely a mathematical representation of the image. An image is defined using 1,000 variables that reside in an algorithm, and if the picture in the search engine and an image in like.com’s data files have enough identical variables, the image is considered to be “like” the one being searched. Subsequently, it will then be offered as a search result. These “likeness searches” provide higher quality results than other search engines.

Another interesting feature of like.com is the search criteria. Like.com allows you to define the importance of some of the variables in your search such as color, size, price, etc. It will then order the results accordingly. Also, even if you enter words into the search criteria, the search is not conducted on the basis of finding the same word. Rather, the text entered into the search engine is converted to a visual signature, and the search is then conducted. For example, if a person is searching for a pair of shoes, they can set the search criteria very specifically based on the previously mentioned variables. In addition if the person types the word “red” into the search engine, it does not search for the word “red”, it converts the word “red” into the visual signature for red, and then conducts the search. This allows for a specific search and very specific results.

As it stands now, like.com only allows you to search with the pictures on their site. However, they soon will offer an upload feature that will allow people to upload their own pictures and run likeness searches on them. This will increase mobility, because people will be able to search wherever and whatever they want. No longer will people be confined to shopping at the mall or at home in front of their computer. When they see an item that they like, they can simply run a quick search at like.com. Thus, they will be able to save themselves the hassles of shopping at the mall or conducting long searches online.

They will also offer a tool bar which will allow people to get pictures off of the internet and run likeness searches on them. A feature that is currently offered but will be expanded is the celebrity search. This allows people to see what celebrities are wearing and run likeness searches on those products to find similar ones often at a cheaper price.

Another add on that like.com plans to incorporate is the cross-matching feature. This feature will allow people to assemble outfits that match online. For example, if a person finds a jacket that they like, they can search for the shoes, hat, or anything else to match.

The door that like.com has opened seems to offer infinite possibilities. The aggravation that conventional searches result in can be eased greatly with this technology. As it stands, like.com is the only company that offers this type of search. However, as in any market, their will be copy cats. It is reasonable to believe that this area will greatly expand in the future to products other than those of aesthetic value. Consumers will look to cure their search headaches, and websites that offer visual searches will be able to alleviate them. Software companies that offer visual searches could create great revenue using click-through pay, as is the case with like.com. Merchants, who advertise their products at these sites, will have another avenue for marketing.

Imagine being able to find a jacket that you saw somebody wearing without the hassle of describing it with words or having to know the brand. All that you will have to do is snap a picture with your phone and run a likeness search at like.com. This new innovative software gives consumers the opportunity to find exactly what they want at the price they are willing to pay. The visual search offered by like.com proves the old saying, “A picture is worth a thousand words” to be definitely true.

References

Magid, L. (2006, December 3). A Sales Clerk You May Like. Retrieved April 14, 2007, from CBS NEWS Web site: http://www.cbsnews.com/stories/2006/12/03/scitech/pcanswer/main2223519.shtml

Arrington, M. (2006, November 8). Riya’s Like.com Is First True Visual Image Search. Retrieved April 14, 2007, from TechCrunch Web site: http://www.techcrunch.com/2006/11/08/riyas-likecom-is-first-true-visual-image-search/

Dunn, R. (2006, November 8). Visual Recognition Technology is Applied to Search at Like.com. Retrieved April 14, 2007, from StepForth Web site: http://news.stepforth.com/blog/2006/11/visual-recognition-technology-is.php

Farber, D., Dignan, L., & Berlind, D. (2006, November 8). Riya launches Like.com visual similarity shopping. Retrieved April 14, 2007, from ZDNet Web site: http://blogs.zdnet.com/BTL/?p=3904

Like visual shopping alpha. Retrieved April 14, 2007, from Riya Inc. Web site: http://www.like.com/

Questions about Like.com (Group 10)

1.	What is the only true image-based search engine?

A.	Google.com

Yahoo.com

Like.com

Ask.com

2.	Like.com was created by what company?

Riya

Microsoft

Sony

D.	None of These

1. 3.3. 2.3.

Visual search technology defines images using ___________ variables.

1,000,000

100,000

1,000

100

4.	A visual search is a ____________ representation of an image

A.	mathematical

numerical

C.	geographical

D.	scientific

5.	Like.com has a specific feature that allows you to shop for apparel that _________ wear.

teachers

B.	Dr. McGaughey

C.	celebrities

D.	medical doctors

6.	Like.com’s main revenue comes from ________________.

A.	banner ads

pop-ups

pop-under

D.	click-through

7.	You can shop at like.com _____________.

A.	via cataolog

online

by mail

by phone