> ## Documentation Index
> Fetch the complete documentation index at: https://docs.hiroshios.xyz/llms.txt
> Use this file to discover all available pages before exploring further.

# Reciprocal Rank Fusion Memory

> Under-the-hood details of Hiroshi's 70/30 concurrent keyword and semantic vector memory.

Hiroshi utilizes a concurrent **Reciprocal Rank Fusion (RRF)** memory retrieval pipeline to achieve robust context matching across both semantic concepts and exact code symbols.

### 📊 The Fusion Formula

When a memory retrieval sweep is triggered, Hiroshi queries the vector similarity indexing and FTS5 keyword indexes concurrently, scoring matching records using reciprocal ranks:

$\text{RRF Score}(d) = \sum_{m \in M} \frac{w_m}{k + r_m(d)}$

Where:

* (M) represents the retrieval models (Vector Cosine Similarity & SQLite FTS5).
* (w\_m) represents the model weight: **0.70** for Vector and **0.30** for FTS5.
* (r\_m(d)) is the 1-based rank index of document (d) within model (m).
* (k) is a smoothing constant set to **60.0**.

```text theme={null}
                     ┌──> Concurrent Vector Search (70% Weight) ──┐
[ Memory Query ] ────┤                                            ├──> [ Reciprocal Rank Fusion ] ──> Top-N Context
                     └──> Concurrent SQLite FTS5 (30% Weight) ────┘
```

This hybrid model guarantees that if your embedding server goes offline or has connection drops, the FTS5 keyword engine acts as an automatic fallback, keeping your memory layer functional.