Self-Learning
Это содержимое пока не доступно на вашем языке.
Self-Learning (ONDA)
Section titled “Self-Learning (ONDA)”ONDA (Observer-Native Dialog Architecture) is Plugy’s self-learning system. It continuously improves your bot’s performance by analyzing conversations, proposing optimizations, and safely applying changes.
How Self-Learning Works
Section titled “How Self-Learning Works”The ONDA system runs automatically on a schedule (default: daily). Each iteration follows this pipeline:
- Benchmark — Run test dialogs against your current bot configuration
- Analyze — Identify the weakest quality dimension (Focus, Empathy, Consistency, or Experience)
- Propose — Use AI to generate an improvement targeting the weak dimension
- Re-benchmark — Test the proposed change against the same dialogs
- Safety check — A guardian system verifies the change is safe and improves quality
- Apply or revert — Only apply changes that pass all safety checks
ONDA Levels
Section titled “ONDA Levels”The self-learning system operates across multiple levels, each optimizing a different aspect of your bot:
| Level | What It Optimizes | Description |
|---|---|---|
| L1 | Persona | Your bot’s personality, tone, and communication style |
| L2 | Knowledge Rules | Individual knowledge base entries and responses |
| L3 | Retrieval | How your bot searches and selects relevant knowledge |
| L4 | Model Routing | Which AI model handles which type of question |
| L5 | Meta-Optimizer | The optimization parameters themselves |
| L9 | Cross-Project | Knowledge sharing between your projects |
When one level converges (no more improvements possible), the system advances to the next level.
Safety Guarantees
Section titled “Safety Guarantees”The ONDA Guardian system ensures changes never degrade your bot:
- Quality floor — Responses must maintain a minimum quality score
- Component check — No individual quality dimension can drop below a threshold
- Rollback — Every change can be instantly reverted
- Auto-halt — After 3 consecutive rejected changes, the system pauses and waits for review
Monitoring Self-Learning
Section titled “Monitoring Self-Learning”Track your bot’s improvement in the dashboard:
- View iteration history with before/after quality scores
- See which changes were approved or rejected and why
- Monitor the current ONDA level and optimization target
- Review the B-score trend over time
B-Score
Section titled “B-Score”The B-score is Plugy’s quality metric, computed as a weighted geometric mean of four components:
| Component | Symbol | What It Measures |
|---|---|---|
| Focus | F | Relevance of the answer to the customer’s question |
| Empathy | E | Alignment with instructions, emotional attunement |
| Consistency | 1-sigma | Non-contradiction with prior answers and knowledge base |
| Experience | Lambda | Use of conversation history and feedback |
Weak-link rule: If any component is zero, the entire B-score collapses to zero.