An A.I. (and Taxonomy) Problem: One Wedding Dress
“I want people to see the dress, but focus on the woman.” Vera Wang.
On Thursday, July 29 (Asia and Oceania) and Friday, July 30 (Europe and Americas), I participated as a panellist at the Omnilytics Fashion Summit discussing the future of data-driven taxonomy with Artificial Intelligence.
What is Fashion Taxonomy?
For reference, fashion taxonomy is defined as the science of naming, describing, and classifying items into categories.
There are so many ways of describing the same thing, which is quite nuanced and proprietary to the fashion industry but is a big problem for customers. For example, what someone might call a “duvet coat” might be a “down jacket” at Balenciaga and a “puffer coat” for Prada.
Fashion companies usually hire staff to maintain its taxonomy, which includes every element of a garment, including physical attributes such as neckline, body style, colour, pattern, cut, product dimensions, and material. Yet, it doesn’t stop there. The taxonomy also includes details such as occasion, season, price, and brand.
Then fashion experts provide multiple examples of varied attributes such as “French girl style” or “date night”, in addition to details on construction and fit, such as “twill” or “billowy”. Those that provide these attributes are often called taxonomists.
The goal is to direct and point customers to specific garments and/or accessories based on the words typed into search on the fashion brand website or via a fashion e-commerce website such as Matches Fashion, Moda Operandi, and many, many others.
The process is never-ending as fashion and social trends are constantly evolving. The vernacular for clothing styles and brands are often dictated by pop culture or language broadcasted on social media.
Usually, when a fashion taxonomy is completed by people, it is already obsolete.
What Does A Data-Driven Fashion Taxonomy with Artificial Intelligence Mean?
Using data, data scientists can train algorithms to identify characteristics in new products, using a combination of natural language processing to “read” product descriptions (supplied by the brands or the designers) and computer vision to “see” product images. Then the responses and/or purchases during queries can be measured and aggregated to confidence scores to see what categories are more successful than others.
If the system flags a product with a low level of confidence, meaning that it is unsure if the attributes it attached are a perfect match, the taxonomists can supplement what the machine provides.
If a new type of product such as “Cross Wrap Crop Tank Tops”, or an attribute is added, taxonomists or A.I. can add these additional attributes.
A product’s taxonomy is repeatedly enriched automatically using a fashion knowledge graph created by data scientists working with fashion experts. This knowledge graph stores thousands of descriptive fashion terms that have relationships associated with them, which helps in the product recommendation process.
Taxonomy never stops tagging. It’s a living, breathing set of information.
Taxonomies also drive sourcing, supply, and seasonal trend forecasting.
But what happens when customer search intent crosses with deep emotional need and human psychology?
Is Artificial Intelligence a match for the intensity of purpose a woman has when searching for – not a wedding dress – but – the wedding dress?
This is when clothing moves beyond just covering one’s naked body to personifying one’s existence of love to the world. Impossible? Possible? Let’s find out. Here’s an example:
Customer Search Intent: One Wedding Dress
My fiance and I are getting married this October.
Her wedding dress search began the day after I proposed which was Christmas Day 2020, December 25th. My fiance’s first search began immediately online.
Wedding dress websites start by asking if you want a new or used dress.
Then they ask you if you have your measurements. If you do have your measurements and know as a bride you will not gain or lose weight between the time you search on the website and the wedding, congratulations! The bride must be a robot!
Other websites will give you an option between measurements and street size.
If you are human, you will choose street size, which also doesn’t have a locked sizing chart. Street sizes vary between each United States state, territory, or region. Why? Wedding dress designers and fashion brands want every bride to feel beautiful regardless of shape. So a street size of 6 might really be a 12.
My fiance tried on multiple wedding dresses that were street size 2, 4, 6, 8, and all of them fit. All the while hearing shop bells ring out when a bride said “Yes!” to the dress. My fiance would stand there in silence and wonder, “Am I a bridezilla and never find the ‘one’?”
The deep investigation for the finding of the one often devolved into, “Maybe I should buy two just in case I don’t like the ‘one’ on my wedding day.”
My fiance sometimes drove hundreds of miles to visit women’s homes who had never worn their wedding dresses and put them up for sale.
Artificial intelligence has no idea the heartbreak of a wannabe bride seeing her wedding dress she is selling being tried on by a stranger – because her wedding plans were cancelled.
Because buying a wedding dress, as my fiance educated me, is not about finding a physical dress, it’s about finding a silhouette.
Don’t ask me what the difference between a Trumpet and a Mermaid silhouette is. The formal definition states:
Trumpet wedding dresses flare at the thigh to elongate the legs.
Mermaid wedding dresses flare at or below the knee, making for an even more dramatic silhouette and highlighting the torso with a long bodice.
To me, both look nearly identical.
My fiance finally decided on her dress on Friday, 14 May 2021.
That’s one hundred and forty days (140) since she began searching. One wedding dress.
She has shown everyone including strangers what her wedding dress looks like via pictures on her phone – except me. I have to wait until October.
I only found out the color of her dress when I picked out my tuxedo and she told me that my shirt should be off-white.
“Why?” I asked.
“Because the wedding dress is off-white.”
Go search for off-white wedding dresses. 223,000,000 results.
So I have no idea what the fashion taxonomy of my future wife’s dress is. Maybe A.I. does.
But it doesn’t matter because now the search has begun for the undergarments. Each wedding dress needs specific undergarments to keep its shape.
Next will be the headpiece – to compliment the dress.
Then the shoes.
Then the hair.
Fingers crossed it doesn’t rain.
Then one Thursday in October, we will be married. They say that’s when the hard part begins.
“Marriage is like a 5,000-piece jigsaw puzzle – all sky.” Cathy Ladman
This article was originally published by Gary Jackson on LinkedIn.
You might also like
Omnilytics Acquires Data Labelling Platform Supahands In $20m Deal
Omnilytics, a fashion analytics and insights software, announced its intention to acquire and partner with data labelling platform, Supahands, in a deal valued at US$20 million. Omnilytics will be embarking on an ongoing strategic acquisition drive to bolster its core functions and technological capabilities to supercharge its long-term growth. Kendrick Wong, Omnilytics CEO and Co-Founder […]
An A.I. (and Taxonomy) Problem: One Wedding Dress
“I want people to see the dress, but focus on the woman.” Vera Wang. On Thursday, July 29 (Asia and Oceania) and Friday, July 30 (Europe and Americas), I participated as a panellist at the Omnilytics Fashion Summit discussing the future of data-driven taxonomy with Artificial Intelligence. What is Fashion Taxonomy? For reference, fashion taxonomy […]
What Will Women’s Workwear Be Like Post-Pandemic?
Employers in the U.S. can now start planning a return to the office as, following a successful vaccine rollout, the country eases many of its Covid-19 restrictions. Data from the Centers for Disease Control and Prevention (CDC) shows that 67% of American adults aged 18 or older have received at least one dose of the […]
Market Insights Made Easy for Speed and Precision
With the recent mass digitisation of fashion retail due to the pandemic, merchandisers, buyers, and designers are increasingly relying on insights to power decision-making. The need to move away from gut feel to actionable insights is becoming more urgent given the high level of competition in online trading. For many brands and retailers new to […]
Top 5 Menswear Trends for Spring/Summer 2021
Like most things in this pandemic, the Spring/Summer 2021 menswear trends are not of the usual variety. The runway season itself was atypical, with major brands like Balenciaga and Gucci opting for virtual fashion shows to eliminate any risk of spreading Covid-19. Designers are embracing the new normal not just in the way they present […]
Product Matching: The Wings for the E-Commerce Boom
For the average customer, the entire retail world now fits into a 15” laptop or 5” smartphone. It is true you can find your preferred product for purchase across different e-commerce sites and portals. However have you ever stopped to wonder about the magic that happens behind the screen that allows you to be presented with different options for the same product, same model, and same variant?