TAXONOMY AND FASHION
Taxonomy might sound complicated, however since becoming the science of naming, describing and classifying items, has since being transformed and made much easier by data science. Artificial intelligence has also made a big difference in recent years, it has helped the retail industry create a lot more comprehensive descriptions for products than it previously did, it has also increased the speed and scale of items retailers can sell, as well as being able to work with more complex products. Something like this could never be cheap, so for retailers to access these services they have had to invest a lot of money for this to work.
Companies have had to adapt to make room for these new advancements in technology, as well as reaching out to help themselves grow even bigger. Psykhe is a relatively new e-commerce website who recommend products based on personality traits; by identifying a user and the desired object, it creates a ‘personality profile’ based on a set criteria, two of which are scores on openness and neuroticism, however there is no human input in the process. Alternatively, there is Rebag, a platform specifically for reselling items, who created a universal taxonomy specifically for designer bags which helps sellers choose a fair resell price.
If you want to try this approach, you will need to hire machine learning engineers alongside fashion taxonomists. These people will be able to take very specific details and working with a data specialist, train different algorithms to detect and identify these specific characteristics. It needs to use a combination of natural language processing so that it can read the product descriptions, which allows the computers to see the image of that product. Think of this as mapping a products DNA.
Sometimes your system will flag a product that has a low level of confidence, all this means is that it doesn’t know if the attributes it detected are the perfect match; but the taxonomists know how to supplement the machine. New products will be added later on, or there might be changes here and there, what will happen is it will just be added as an additional attribute.
The reason this is becoming so important over time is because in fashion, a lot of people will have a different name for the exact same thing. So, when you take a product and describe it, and then photograph it before it is used in an editorial, that specific products taxonomy gets enriched using a fashion knowledge graph that was created by data scientists specifically for the fashion industry. This will store any fashion terminology that you can think of off the top of your head, with the relationships associated with that specific type of garment and then help later on in the recommendation process. This helps search results but having a taxonomy that is enhanced using artificial intelligence, it means more products can be identified on a much wider spectrum.
Over time, taxonomies will get more sophisticated and advances in technology will further enhance this. This will make life for consumers much easier, so would you consider using this?