Synopsis
CPRD.com lets people ask questions about published content and receive answers with source references. Follow one research question through Drupal, Azure AI Search and a Foundry IQ knowledge base to see how the feature retrieves evidence, builds an answer and handles gaps in the available content.
Description
What happens when someone asks CPRD.com, “Which approved studies mention cardiovascular outcomes?” This session follows that question from the Drupal interface to a cited answer.
I’ll show how Drupal content enters Azure AI Search, how a Foundry IQ knowledge base retrieves relevant material, and how Drupal presents the answer and its sources. Using this example, I’ll share three decisions that shaped the feature: limiting which content can support an answer, making source references useful to readers, and responding clearly when the evidence is insufficient.
Attendees will leave with a practical picture of how grounded question answering can build on an existing Drupal search index.
Session outline
- The question and the user’s need.
- The Drupal content and Azure AI Search foundation.
- Follow the question through retrieval, answer generation and citations.
- Three design decisions: content scope, sources and insufficient evidence.
- Takeaways and questions.