Add Hybrid Search: Combine Vector similarity with keyword/tag matching

[kornelrabczak]

Problem

Currently, the only search mechanism available is vector-based semantic search using the content of memory notes stored in LadybugDB. While this works well for semantic similarity, it has limitations:

• Exact keyword matches may be missed if the semantic meaning differs
• Tags and keywords stored on notes are not utilized for search ranking
• Results may miss relevant notes that contain specific terms the user is looking for

Proposed Solution

Implement a hybrid search approach that combines:

1. Vector search (semantic similarity via embeddings)
2. Keyword/tag fuzzy search (text matching on keywords[] and tags[] arrays)

Score Fusion Algorithm

Use weighted score combination:
finalScore = (vectorScore * vectorWeight) + (keywordScore * keywordWeight)
Default weights: vectorWeight = 0.7, keywordWeight = 0.3

Matching Strategy

Use contains match (case-insensitive partial matching) via Cypher CONTAINS operator:

• keywords array: match if any element contains the search term
• tags array: match if any element contains the search term