Skip to main content
Contexts describe the placement, audience, and intent behind a recommendation request. Properly configuring them keeps results relevant across your product surfaces.

​
When to create a new context

Create a dedicated context whenever:
  • The UI surface has unique layout or business goals (homepage vs. detail page).
  • You need a different blend of retrieval settings, filters, or pipeline behaviour.
  • You want to measure performance separately from other experiences.
Start with a handful of high-impact contexts, then expand as you identify more nuanced placements.

​
Configure contexts in the console

  1. Navigate to Console → Contexts.
  2. Click New context and enter a numeric Context ID (for example 101). The ID is stored as a number and is the value your engineering team will send in API calls.
  3. Fill in the modal to define how the context should behave:
    • Context Name and Description document the placement.
    • Select an Associated Model and Recommendation Type once the underlying model exists.
    • Configure Filters and optional combinations (AND/OR) to enforce inventory and eligibility rules.
    • Use the context to decide what gets retrieved before scoring.
    • Use the Rules Engine for post-scoring controls such as boost, bury, pin, dedupe, reorder, or grouping.
  4. Save the configuration and share the numeric context ID with your engineering team.

​
Reference contexts in API calls

Attach the context ID to every recommendation request so NeuronSearchLab can apply the proper logic. 
curl -X GET "https://api.neuronsearchlab.com/recommendations" \
  -H "Authorization: Bearer <access_token>" \
  -G \
  --data-urlencode "user_id=user-123" \
  --data-urlencode "context_id=101"
The context you choose also appears in analytics, allowing you to compare conversion rates and engagement across placements.

​
Iterate continuously

Monitor performance metrics and adjust context settings as business goals evolve. Combine context changes with explicit Rules Engine updates and event tracking to keep the learning loop tight.