[Test] Add Comprehensive Test Coverage for Embedding-Based Signals

[Xunzhuo]

We need comprehensive test coverage for all embedding-based signals to ensure they function properly and to prevent regressions.

Background

Currently, these signals lack proper test coverage, leading to undetected failures in production.

Severity

High (Testing/Quality)

Scope

This issue focuses on adding integration tests specifically for embedding-based signals, separate from the E2E tests that revealed the problems.

General Embedding Signal Tests

Integration Tests:

• Test embedding model loading
• Test embedding model performance
• Test embedding model memory usage
• Test concurrent embedding requests

Acceptance Criteria

• All unit tests for PII detection signal pass (minimum 20 test cases)
• All unit tests for keyword matching signal pass (minimum 15 test cases)
• All unit tests for domain classification signal pass (minimum 20 test cases)
• All integration tests for embedding-based signals pass
• Test coverage for embedding signal code reaches >90%
• Tests run in CI/CD pipeline
• Tests are documented with clear descriptions
• Performance benchmarks are established for embedding operations

Success Metrics

• PII Detection: 100% accuracy for all PII types in unit tests
• Keyword Matching: 100% accuracy for all configured keywords in unit tests
• Domain Classification: >95% accuracy for obvious domain queries in unit tests
• Test Execution Time: All signal tests complete in <5 minutes
• Code Coverage: >90% coverage for all embedding signal code

[yehudit1987]

/assign