Draft: Adds OpenSearch Embedding Search via Bedrock to suggest.py and adds OpenSearch service module

[syqs]

This PR integrates semantic vector search into our suggest.py flow by creating and using new service Amazon OpenSearch Serverless (AOSS) and AWS Bedrock embeddings.

⚠️ Note: this PR is not ready for merge, it is a proof of concept

flowchart  TD
A[User types a query] --> V[Vector Embedding via Bedrock Titan] --> C[OpenSearch k-NN] -->R[Normalize and sort results] --> D[Display in suggestion results dropdown]

Loading

With this change, we now have a hybrid search engine that combines:
Semantic vector similarity via Bedrock → OpenSearch
Traditional keyword search (boosted by title/abstract relevance)
This gives us the best of both worlds: meaning-aware search and precision keyword matching.

🛠 Implementation Notes

Index Mapping Updates

Added embedding field (knn_vector, 1024-dim) for Titan embeddings
Authors stored as keyword for precise aggregation + display

Authentication

Integrated AWS4Auth with AOSS (aoss service target)
Supports IAM users or roles with data access policy

This PR introduces new pydantic schemas for payload validations

opensearch_service includes lazy initialization for better performance and graceful degradation
Three new dependencies added to the pyproject.toml ("pydantic>=2.0.0,<3.0.0", "requests-aws4auth>=1.2.3,<2.0.0", iso8601==1.0.2)

Future implementation:

flowchart TD
    A[User Query] --> K[Keywords] --> D[BM25 Search]
    A --> V[Vector Embedding] --> C[OpenSearch k-NN]
    C --> E[Hybrid Scoring]
    D --> E
    E --> F[Final Results]

Loading

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud