Keywords to Context: Semantic Search and Retrieval-Augmented Generation with OpenSearch
Time: Wednesday 13th May 11:00-11:30
Room: Te Aro
Track: Drupal Development
Keyword-based search has served us well for years, but it often falls short when users ask natural language or exploratory questions. Semantic search improves relevance by understanding intent, meaning and context, while Retrieval-Augmented Generation (RAG) goes a step further by transforming search results into clear, human friendly answers.
This talk will explore how OpenSearch and Skpr enable semantic search by understanding the meaning behind content and queries, rather than just matching keywords, with the RAG processor layering generative AI on top of retrieved results. We’ll walk through a practical end-to-end flow: indexing and chunking content, generating the embeddings for semantic search, retrieving the most relevant search results, and using RAG to produce concise summaries and answers grounded directly in search results.
Along with answering questions, RAG can be also be used to summarise multiple documents, highlight key themes, and reduce information overload, while staying grounded in the indexed content without AI hallucinations.
By the end of the session, you’ll understand when semantic search is a better fit than traditional search, how OpenSearch’s RAG processor enables summarised, clear and contextual responses, and how these approaches can deliver faster access to accurate information, increase user engagement, and create search experiences that go beyond a simple list of links.
This talk will explore how OpenSearch and Skpr enable semantic search by understanding the meaning behind content and queries, rather than just matching keywords, with the RAG processor layering generative AI on top of retrieved results. We’ll walk through a practical end-to-end flow: indexing and chunking content, generating the embeddings for semantic search, retrieving the most relevant search results, and using RAG to produce concise summaries and answers grounded directly in search results.
Along with answering questions, RAG can be also be used to summarise multiple documents, highlight key themes, and reduce information overload, while staying grounded in the indexed content without AI hallucinations.
By the end of the session, you’ll understand when semantic search is a better fit than traditional search, how OpenSearch’s RAG processor enables summarised, clear and contextual responses, and how these approaches can deliver faster access to accurate information, increase user engagement, and create search experiences that go beyond a simple list of links.
Speakers
Daniel Veza
Daniel Veza joined PreviousNext as a Drupal Developer in late 2021, having previously worked at Sparks Interactive in New Zealand as a Drupal Developer for 6 years.
Daniel has been an active part of the Drupal community since 2015 and in 2024 he became a core maintainer for Layout Builder. He has led or been involved in building countless Drupal sites across the public and private sectors in both New Zealand and Australia.
Daniel has been an active part of the Drupal community since 2015 and in 2024 he became a core maintainer for Layout Builder. He has led or been involved in building countless Drupal sites across the public and private sectors in both New Zealand and Australia.