Integrated data enrichment framework

We’ve processed over 1 million products through the Hullabalook platform, across a huge range of categories. So our technology has a detailed understanding of everything from dresses to lightbulbs and our platform can undertake a rich feature analysis across all of these attributes to deliver to you enriched and accurate product tags.

Having enriched product data enables you to offer your customers a better filtering and search experience – getting them to the item they want to purchase faster and increasing the chance that they will convert.

The core beliefs built into our data enrichment framework:

1. We believe…

…that almost all Attributes are continuous not discreet. For example rather than asking “is this dress red?” we ask “how much of this dress is red, and which shade of red?”

1. Benefits of this approach

By creating new enriched Attributes or Tags on a spectrum we can achieve much higher specificity when ranking products – there is a huge difference between ‘pink’ and ‘burgundy’ – interpreting it with our ranking algorithm gets the best results at the top of the page

2. We believe…

… that the best data enrichment results will come from analysing all data sources, so we process unstructured text using NLP techniques alongside all images using our computer vision models

2. Benefits of this approach

By using all available data sources we can make sure that Tags like colour are derived from images, whereas styles like “mid-rise” or “high-rise”, which are impossible to detect from images, are derived from unstructured text.

3. We believe …

… that having a wide range of enriched product tags should be balanced with having accurate product tags. Quality is extremely important to ensure a good customer experience, and this should be driven by deep routed domain expertise.

3. Benefits of this approach

Having enriched data which is incorrect or misleading can result in higher returns rates and a poor customer experience. Our data enrichment framework has accuracy at its core so that you know that the data on your site is always correct.