VueTag Preview / Trial
Vue.ai offers a free trial or a proof of concept (POC) for the its automated product tagging solution - VueTag. This note explains the scope of this preview and how you can make use of it.
How many products?
- For the purpose of the POC we usually work with a small subset of your product catalog.
- In most cases, a few 100s or 1000s of products gives us sufficient data to showcase the capabilities of the VueTag solution.
- Ideally, you can export a subset of products across various categories (product types) from your catalog
What is needed?
- The product catalog can be provided to us as a CSV file or any exportable format from your e-commerce database
- We require you to send us the product images (multiple angle images, if available), any text metadata available with the products (title, description, unstructured text blob etc)
- Once the data for the POC is made available to us, based on the provided data we will do an analysis on the methodology that can be adopted for tagging. This could be one (or more) of the following:
- Using the images available
- Using the images + text metadata and a reconciliation of the two to arrive at an aggregated set of tags for a product
What will the output be
- The output of the tagged data will be available as a CSV file similar to the format of the input file, with the tags appended as columns. For example,
Product ID Title Category Description
Product ID Title Category Description Vue_category Vue_subcategory Vue_BaseColor Vue_Pattern etc…
Additionally, the tagged output could also be made available in the VueTag Dashboard for your review
- As a part of the POC, an internal review of the tagged data is done by our team from which accuracy numbers are presented for the various tags. Typically our accuracy of networks is
- Category prediction: >90%-95%
- Attributes prediction: >80%-85%
A sample set review on the POC sample data gives both sides an understanding of how well the solution can work for your business and potentially, and where network improvements may be required on our side.
- The taxonomy of the output will be in the default Vue.ai taxonomy for the POC. It is possible for us to work with your own business taxonomy as well by doing a mapping of the taxonomy based on the definition of the tags. This is not a part of the POC scope and will be taken up in a paid (contracted) scope.
- We would encourage and require you to think of the POC evaluation criteria, which could be one or more of the following.
- Accuracy - How accurate the tags are
- Coverage - For the subset of your product catalog provided how extensive the tag output is
- Additionally, Vue.ai also offers custom network development to extend the visual prediction capabilities. This is not a part of the POC scope and will be taken up in a paid (contracted) scope
- VueTag API documentation is available here for your review. With this you can automate parts of the product onboarding process on your end as you see fit. We can suggest ways of doing this based on your needs.