Build the profile and item data graphs
Personalization recommenders read two kinds of data graph: a profile data graph for the person being personalized to, and an item data graph for the catalog (products, content, or offers) being recommended. Create one of each before you build a recommender.
Profile vs. item data graph
| Profile data graph | Item data graph | |
|---|---|---|
| Centered on | One person (the unified profile) | One catalog item (product, content, or offer) |
| Root object | The unified profile from Identity Resolution | Your item/product DMO (Data Model Object — Data Cloud’s canonical object type) |
| Required mode | Real-time only — batch is not eligible | Can be batch or real-time |
| Used by a recommender for | Who to personalize to: their activity, attributes, and calculated insights | What to recommend: the catalog a recommender ranks and returns |
Create the graphs
- From Data 360, open Data Graphs for your data space.
- Create a profile data graph rooted on the unified profile and set it to real-time.
- Create an item data graph rooted on your item/product DMO.
- Confirm both graphs build successfully and return records.
- Open a recommender and point it at the item data graph, with the profile data graph supplying the person context.
Discussion
Comments are not configured for local development. They appear once the site owner connects GitHub Discussions (Giscus).