ABOUT THIS DEMO

With this project, born as a joint experiment between WordLift and Jina AI, we want to achieve the following:

1. Create an impact on E-Commerce stores by leveraging on structured linked data (WordLift) and Neural Search (Jina)
2. Build a demo to explore the advantages of using a Product Knowledge Graph and Neural Search applied to a fictitious fashion store
   1. Can we create a search experience capable of improving the user experience while (potentially) increasing the sales?
   2. Are we able to scale from an ultra basic solution to an high-end sophisticated implementation that can work well for larger brands (ie more products, more features, etc.)?
   3. Can we leverage on entity tagging when building such an experience?
   4. What is the real breakthrough?

THE DATASET

We have derived the dataset by slicing and enriching the WDC Schema.org Table Corpus for public download provided by the University of Mannheim. 

The original dataset included 18.521 rows with the following attributes and density:

{‘offers’: 99, ‘image’: 97, ‘productid’: 51, ‘name’: 100, ‘description’: 51, ‘sku’: 100, ‘brand’: 100}

We obtained 9.203 products to which we added: category (accessories, shoes, bags and clothing) and color using NLP.