Optimizing Bots

Back to NLP Topics

Optimize your assistant’s NLP by refining names, synonyms, patterns, and engine configurations. Start by defining synonyms at the task level, then manage them at the app level.

---

NLP Optimization Options

Go to Natural Language for your assistant:

Section	What to configure
Training > Machine Learning Utterances	Add and manage utterances for intent recognition.
Training > Synonyms & Concepts	Optimize intent/entity recognition with synonyms and concept groups.
Training > Patterns & Rules	Define patterns for slang, idioms, and complex expressions.
NLU Config	Set recognition confidence thresholds and engine-specific tuning.

---

NLP Version 3

All new VAs use NLP Version 3 by default (since v10.0). Existing VAs were migrated on January 21, 2024.

What’s new in V3:

• Improved Traits Engine accuracy.
• Transformer and KAEN model support for English; Transformer for other languages.
• Improved security and performance.
• Enables LLM & Generative AI features.
• Access to Zero-shot Model and Few-shot Model for intent detection.
• Access to Intent Discovery.

On-premise customers can run both V2 and V3 in parallel, or switch directly to V3.

---

Where to Go Next

Customize each NLP engine for your use case:

• Machine Learning Engine
• Knowledge Graph Training
• Fundamental Meaning Engine
• Traits
• Ranking and Resolver