Flexible product data management
Are you constrained by inflexible product data models?
Is your ecommerce platform full of product data that you can’t make use of?
Do you struggle to make intelligent decisions about products because essential data is spread across multiple systems?
Most product databases being used today require you to define upfront a data model, and to limit content to either unstructured descriptions, or key value pairs.
The Hullabalook platform enables you to store your product data in a blended normalised and document structure which is supported by a horizontally scalable data loading process which is able to fetch, and fuse data from your file servers and APIs. When combined with our data enrichment platform, you can mine images and text for additional product tags and use them within your front-end without immediately.
The Hullabalook approach to data management
Data snapshots & provenance
We start by ingesting your product data in real-time by polling your servers to find any updates – whether on feeds or APIs. We calculate the deltas from the previous product snapshot and use this information to instantly decide in which cases we need to re-run our computer vision and product classification models – at all times tracking the source of each attribute so we can understand its data provenance.
Performance & latency
The Hullabalook platform has built in resource prioritisation to ensure that the attributes which need to be delivered with the lowest latency are updated first – like stock and price. While more demanding computations (like our computer vision modules) run in parallel.
3. The Hullabalook platform enables agile in-flight tagging; so if a pair of trousers is getting more clicks when classified as ‘formal’ than ‘casual’ then we can adapt the classification accordingly. The platform continuously evaluates the self-consistency of the fused products to ensure the highest levels of accuracy at all times.