Retailers understand their products well, but much of that knowledge remains in the heads of buyers, merchandisers and stylists rather than in a format machines can use.
The same gap exists on the shopper side. Someone may search for “burgundy” when the catalogue says “wine,” while a request such as “a linen dress for a beach wedding” combines fabric, occasion, climate and dress code. Traditional product data does not connect these meanings.
That worked when a person did the interpreting. It is not workable now. Retailers cannot scale meaning that has to be reapplied by hand every season. Nor can they build the agentic experiences customers increasingly expect. Those experiences depend on relationships and interpretations nobody ever recorded. The market is already moving: Google added conversational attributes to its product feed specification in May.
What the session covers
James Brooke will show GEMA, Mapp’s graph enrichment platform, running live on a catalogue. The platform is initially focused on fashion but extends to other lifestyle categories, including beauty and homewares. Standard product data in, a governed graph out, and a shopper experience built on top of it. Attendees will learn
- How a fashion ontology defines the physics of fashion at feature level, using outfit, fit and silhouette knowledge
- How consumer language maps onto that vocabulary, and how a compositional query gets decomposed and resolved against the graph
- Why the scaffolding has to cover content, campaigns and category pages, not just products
- What this changes for shoppers on the retailer’s own site and for agents reading the catalogue from outside
Bring questions on graph modelling, governance, agentic architecture, or how a meaning layer sits inside a composable stack.
