Back in early 2012, our Emerging Channels/BBY Open group set off on a journey to explore our product data in a unique way utilizing semantic web and linked data technologies to extract insight from the large amounts of product data we had available to us. We named this highly experimental effort Metis, after the Greek Titaness of wisdom and deep thought, which we figured was an appropriate moniker given the nature of what we were trying to accomplish. Our hypothesis was there were vast untapped, unstructured and semi-structured data resources right underneath our noses that would yield a wealth of rich relationships and insight when RDF modeling and ontologies like Good Relations were applied to give the data meaning to machines.
Over the next couple of years, the Metis core grew to power a number of sub-projects aimed at solving various data and insight-related problems we had identified internally. Metis expanded to contain a number of core features that all leveraged semantic web technologies:
- RDF/XML serialization process for all product data in the catalog
- Initial publicly-facing alpha SPARQL endpoint
- First stab at a Like for Like or product endpoint (codename: Maia, an offspring of Metis)
- Internal SPARQL endpoint — used for serendipitous long tail product discovery
- Relationship builders — establishing ties between primary products and consumables or accessories from internal and external data sources
As with many experimental ventures, Metis’s roadmap was finite. In July of 2014, the team disbanded to work on other initiatives, and the product was retired. However, the information around the company and customer benefits of semantic technology from the POC and production tests was a valuable outcome of the work. The Metis story has been told a number of times, most recently at the Semantic Technology & Business Conference in August of 2014.
So given the positive outcomes we found, what’s next for Metis, or a Metis-like equivalent? Luckily, there is an ongoing effort from our Product Development team looking at revitalizing some of the Metis work in another POC effort. This would include expanding RDF serialization to include Turtle, JSON-LD, and potentially others. Also high on the list of wants is a beta version SPARQL endpoint open to public consumption.
We appreciate all the feedback over the life of the Metis project, including the recent queries around its status and additional feature requests. We look forward to bringing new and exciting insights and status updates as future initiatives come into play.
RDF (Resource Description Framework): a metadata model used for data exchange on the web. Can take the form of a handful of different serializations, including Turtle, JSON-LD and RDF/XML.
Ontology: formal definitions of metadata that allow the author to define classes of data resources, data properties, and relationships between classes of things, for the purpose of knowledge sharing and reuse.
SPARQL (SPARQL Protocol and RDF Query Language): standards-based query language and protocol used to query RDF resources
Originally published by Ira Brooker on 02/06/2015 at developer.bestbuy.com.