What's changed: Added per-section figures (cert-figure-retrofit). Added a speech section (s3) to AI-103 Chapter 5 (the official text-analysis domain’s "speech" sub-area = STT/TTS, speech as agent modality, custom speech / custom neural voice, multimodal reasoning from audio, speech translation). Existing s1 (Azure AI Language) and s2 (translation/custom language models) preserved non-destructively
5.2Translation and Custom Language Models
Understand text translation (Azure AI Translator), custom options when prebuilt is insufficient (conversational language understanding CLU, custom classification, custom NER), and when to use flexible generative-prompt analysis.
To convert text to another language, use Azure AI Translator. When prebuilt features fall short, consider custom. For chatbots, extract user intent and entities with conversational language understanding (CLU); for classification into your categories, use custom classification; for org-specific extraction, custom NER—all trained on your data. Increasingly, generative prompting ("extract key points, topics, and sentiment from this text") offers flexible analysis not bound to predefined categories.
Common: (1) "convert to another language" = Azure AI Translator (vs language detection). (2) chat intent and entities = CLU. (3) classify into your categories = custom classification; org-specific extraction = custom NER (train on your data). (4) flexible extraction of key points/topics/sentiment without predefined categories = generative prompting.
Watch out: (1) language detection (which language) vs Translator (convert to another) differ. (2) CLU (intent + entities) vs plain sentiment/NER—do not confuse. (3) Custom needs training/evaluation on your data—avoid over-investing when prebuilt or generative prompting suffices. (4) Generative prompting is flexible but watch hallucination (pair with grounding/evaluation).
5.2.1Section summary
- Translation uses Azure AI Translator (distinct from language detection)
- Custom: intent/entities = CLU, your categories = custom classification, org extraction = custom NER (train on your data)
- Flexible analysis beyond predefined categories = generative prompting (watch hallucination)
Sign in to track progress — Log in.
Quick check
(just a quick review)Q1. Which Azure service converts text from one language to another?
Q2. For a chatbot, which Azure AI Language capability extracts user intent and entities?
Q3. How do you classify text into org-specific categories that prebuilt cannot?
Q4. Which is a strong approach to flexibly extract key points/topics/sentiment without predefined categories?
Keep track of your progress
The full study guide is free to read. Sign up free to practice with the question bank, track what you have read, review your mistakes, and highlight passages.