You go to the store and find the perfect pair of shoes, except that they don’t have them in your size, or in the color that you want.
You shop online and try to find a pair of pants and a shirt that match, except that it’s hard to tell how they will look together.
Can a search engine help you make shopping easier?
Imagine a different scenario, now…
You go to a facility to create a 3-dimensional scan of your body which you can then upload to a clothing search engine and recommendation system. Or you enter detailed body measurements into that search and recommendation system.
You create a user profile that details things such as your skin color, eye color, hairstyle/color, whether or not you wear glasses, and other details about your appearance.
In addition to this personal information, you can also add non-personal information, such as the type or color of clothing you are presently looking for. Other details can be entered, such as your location, the season of the year, your hairstyle, where you are planning to go while wearing the clothes you are looking for, such as a business meeting or a restaurant or night club.
Your clothing preferences could be added to this profile, including themes such as formal or casual or workout clothes. If you like specific brands of clothes, those could also be added. Your preferences for certain materials and color palette preferences could be entered.
This fashion search engine could then be used to enable clothing experts and “friends” to recommend clothes for you, to let clothing advertisers target you with ads that match what you might want to find and to let a search engine provide recommendations based upon aggregated user-profiles and user recommendations.
The clothes could be shown on the 3D model that you submitted, or that was built based upon your submitted measurements.
A patent filing from Yahoo describes a Fashion Search Engine that can use a 3-dimensional model to present clothing, a way for others to provide clothing recommendations, and a matching algorithm that can help you find clothes that match.
Why Build a Fashion Search Engine/Recommendation System?
The patent application provides a laundry list (sorry for the pun – couldn’t help it) of costs and limitations associated with shopping for clothes at a store or online.
Both can take a considerable amount of time, and provide the risk that you might miss out on something that you might want, but haven’t come across.
When buying online, there’s a risk that the clothes purchased might be the wrong color or size or style or fabric. There’s also no way to ensure that those clothes match and that they can be worn together.
You also can’t always easily bring along others with you when you shop to make suggestions or recommendations.
The patent application is:
Method for Search and Marketing Fashion Garments Online
Invented by David Burgess, Shyam Kapur, and Caitlin Smallwood
Assigned to Yahoo
US Patent Application 20090019053
Published January 15, 2009
Filed July 13, 2007
A method of generating clothing recommendations for a potential purchaser uses user-specific information to generate a list of suggested garments for the user and displays a graphical representation of the garments.
The user-specific information may include recommendations from one or more friends, fashion experts, or other purchasers, optionally including information based on purchasing history or shopping history of other purchasers. The graphical representation may include a three-dimensional representation. Recommendations may be prioritized based on ranking.
The user profile values that you enter could be updated at any time so that if you lose or gain weight, it can allow for those changes. Or if you are searching for a bathing suit one day, and a business suit the next, you receive the appropriate results for your search. It also allows for recommendations from others…
Wisdom of the Shopping Crowds
The fashion search results you see could be based upon a mixture of factors, such as:
Expert Recommendations – clothes could be recommended by an “expert,” who has been hired to make these kinds of recommendations
Friend Recommendations – Other users of this system might be given permission by the person searching to access parts of his or her user profile that are needed to make a recommendation, and may make recommendations for clothes.
These “friends” of the searcher could be added to the user profile for future use, and this recommendation/searching system might send them an alert via email or instant messenger when the searcher is looking for new clothes.
The searcher could rate or rank the recommendations of their friends, other users, and the experts, and that information might be stored in the user profile as well as being aggregated with other ratings or rankings, in a “wisdom of the crowds” approach to help other searchers by bringing the:
“best” users and experts to the forefront so that users of the system can continuously find other people’s recommendations to help them with their clothing searches.
Historic Shopping Data
This fashion search might collect data about the searcher’s purchasing and searching habits over time to help them find “relevant” clothing. Examples include
how many times the user has searched for a particular brand of shirt in the last 12 months, or requested recommendations from a particular expert, or purchased jeans that cost less than $150, etc.
Feedback on Clothes
A shopper might provide information on clothes that they purchased, flagging brands and sizes as favorites, or rating or ranking the recommendations of experts, and requesting recommendations from them in the future or suggesting them to friends.
The shopper might also add feedback to their profile about clothes that they already own, even if they didn’t purchase those clothes through the system, to inform it about their likes and dislikes, and to let friends comment on, and rate those clothes.
A matching algorithm could also be used to help someone find clothes, by taking information from the user profile and gathering clothes or complete outfits that match what the searcher is looking for.
The searcher can add rules to the system, such as “all t-shirts go with all jeans”, or “these shoes go with all jeans,” or “this color “goes with” that color.”
Recommendations could be made to the searcher on things such as matching shoes or pants or shirts, etc.
Presenting Clothes to a Searcher
Once a search has been completed, the clothes found could be presented to the searcher on the 3D model, if that feature is available for all clothes suggested and the searcher wants to see them on the model.
The model could be rotated so that the searcher can see the clothes from different angles.
Different sizes, colors, and styles of the clothes shown could be changed on the model as desired, as long as the designer or manufacturer or distributor offers those options.
Made-to-measure clothes that are custom-tailored could also be shown, and alterations for those could be made on the screen, such as adjusting a hem on a pair of pants or bringing in the waist on a shirt.
Advertising Along Side Suggested Clothes
In addition to showing the searcher suggested clothing, this system might present ads based upon what the searcher is looking for, as well as information found in the searcher’s profile. These ads might also be shown while the user of this fashion search engine is using other applications, such as email.
This fashion search and recommendation system would enable searchers to add the clothes that they’ve found to a shopping cart to be purchased.
It may also make further recommendations based upon selections for purchase, such as shoes to go with the clothes chosen.
Last June, I wrote about a Yahoo patent filing for a Meal Search that would try to help make it easier for people trying to find a restaurant or recipes or nutrition information to locate that information. Seeing Yahoo come up with a very focused and limited type of search that would solve a specific set of tasks appeared to be an interesting exercise from the search engine/portal.
This Fashion search also provides a very focused and limited kind of approach to solving a problem that they believe exists when people try to shop for clothes online or offline.
Many people commented on my post about the meal search that they didn’t believe that restaurants would provide information about the meals that they offered, including nutritional information, to make such a system work. Would clothing manufacturers and distributors provide the 3-dimensional models to Yahoo that would make a fashion search work the way that they envision?
Would people use a fashion search engine like this, and provide detailed information about themselves to the search engine, including a 3-dimensional model of themselves? Would you?