najia-afk/aria_skills/pattern_recognition

---
name: aria-pattern-recognition
description: "🔍 Behavioral pattern detection in memory streams"
metadata: {"aria": {"emoji": "🔍"}}
---

# aria-pattern-recognition

Behavioral pattern detection engine. Analyzes memory streams to find
recurring topics, temporal habits, sentiment drift, emerging interests,
and knowledge gaps. Stores detected patterns in semantic memory.

## Architecture

Memory stream (list of memory dicts) ↓ TopicExtractor (9 keyword domains + entity regex + tech regex) ↓ FrequencyTracker (sliding window, default 30 days) ↓ PatternRecognizer ├── Topic recurrence (repeated subjects) ├── Interest emergence (growth rate analysis) ├── Temporal patterns (peak hours, active days) ├── Sentiment drift (valence trend over time) └── Knowledge gaps (repeated questions) ↓ SemanticMemory (category: pattern_detection)

## Pattern Types

| Type | Description | Detection Method |
|------|-------------|-----------------|
| `topic_recurrence` | Same topic appears repeatedly | Frequency ≥ min threshold |
| `interest_emergence` | New topic growing rapidly | Recent/historical ratio ≥ growth rate |
| `temporal` | Usage patterns by hour/day | Peak hour and active day analysis |
| `sentiment_drift` | Emotional trend over time | Linear regression on valence |
| `knowledge_gap` | Same question asked repeatedly | Question mark + recurrence |

## Usage

```bash
# Run full pattern detection (auto-fetches memories if empty)
exec python3 /app/skills/run_skill.py pattern_recognition detect_patterns '{}'

# With explicit memories and confidence threshold
exec python3 /app/skills/run_skill.py pattern_recognition detect_patterns '{"min_confidence": 0.5}'

# Get recurring topics
exec python3 /app/skills/run_skill.py pattern_recognition get_recurring '{"min_frequency": 0.3}'

# Get emerging interests
exec python3 /app/skills/run_skill.py pattern_recognition get_emerging '{"min_growth_rate": 2.0}'

# Check detection stats
exec python3 /app/skills/run_skill.py pattern_recognition get_pattern_stats '{}'

Functions

detect_patterns

Run full pattern detection on a list of memories. If no memories provided, auto-fetches from semantic memory via api_client. Stores top 20 detected patterns back into semantic memory with confidence scores.

get_recurring

Get topics that appear with frequency above a threshold (events per day). Uses the internal frequency tracker's sliding window.

get_emerging

Get topics that are growing rapidly (recent frequency ÷ historical frequency ≥ growth rate multiplier).

get_pattern_stats

Get statistics from the last detection run — pattern counts by type, new vs persistent patterns, analysis window.

Web Dashboard

Available at /patterns in the Aria web UI. Features:

• Chart.js doughnut chart (patterns by type)
• Confidence distribution bar chart
• Pattern list with color-coded type indicators
• Type filter dropdown
• "Run Detection" button

Dependencies

• api_client (semantic memory for storage and retrieval)