Hybrid Search with Amazon OpenSearch Service

This article provides a solution for building a hybrid search system using Amazon OpenSearch Service. It combines lexical and semantic search to improve search relevance and user experience.

Specifically, the article covers:

• Hybrid search workflow in OpenSearch Service using search pipelines and normalization processors
• Integrating OpenSearch Service with Amazon SageMaker for generating vector embeddings
• Deploying a sample web application for searching a retail product dataset
• Experiments comparing lexical, semantic, and hybrid search results
• Fine-tuning hybrid search configuration using normalization and score combination techniques