Domain 4: Natural Language Processing Workloads on Azure

<- Previous Domain | Overview | Next Domain ->

Weight: 15-20%

This domain tests whether you can identify common NLP scenarios and match them to Azure AI Language and Azure AI Speech capabilities.

---

Common NLP Workload Scenarios

Scenario	What It Does	Example
Key phrase extraction	Finds main concepts in text	Extract "delivery delay" from support feedback
Entity recognition	Finds named entities	Person, organization, location, date, quantity
Sentiment analysis	Detects sentiment in text	Positive, negative, neutral, or mixed feedback
Language modeling	Understands or generates language patterns	Summarization, completion, classification, Q&A
Speech recognition	Converts speech to text	Transcribe a call recording
Speech synthesis	Converts text to speech	Read an answer aloud
Translation	Converts text or speech between languages	Translate English to French

Key Phrase Extraction

Key phrase extraction identifies important terms and phrases from text. Use it to summarize themes from reviews, tickets, survey responses, or documents.

Entity Recognition

Entity recognition extracts structured items from text. Named entity recognition can identify entities such as people, organizations, locations, dates, and quantities.

Sentiment Analysis

Sentiment analysis classifies opinion in text. Use it when the scenario asks whether customer feedback is positive, negative, neutral, or mixed.

Language Modeling

Language modeling is broader than a single NLP feature. It supports language understanding and generation tasks such as summarization, question answering, classification, and content generation.

Speech Recognition and Synthesis

• Speech recognition: speech to text.
• Speech synthesis: text to speech.

If a scenario starts with audio and needs text, choose speech recognition. If it starts with text and needs audio, choose speech synthesis.

Translation

Translation converts content between languages. Some scenarios involve text translation; others involve speech translation.

---

Azure Tools and Services

Azure AI Language

Azure AI Language supports text analytics and language understanding features such as:

• Key phrase extraction.
• Entity recognition.
• Sentiment analysis.
• Language detection.
• Question answering and conversation-oriented language understanding scenarios.

Azure AI Speech

Azure AI Speech supports speech workloads such as:

• Speech to text.
• Text to speech.
• Speech translation.
• Custom speech scenarios.

---

Decision Rules

Find important phrases in feedback       -> Key phrase extraction
Find people, places, dates, quantities   -> Entity recognition
Detect positive/negative/neutral tone    -> Sentiment analysis
Audio to written text                    -> Speech recognition
Written text to audio                    -> Speech synthesis
Convert language A to language B         -> Translation
Text analytics and language features     -> Azure AI Language
Speech input/output                      -> Azure AI Speech

---

Exam Traps

• Speech is not the same as Language: Audio input/output maps to Azure AI Speech; text analytics maps to Azure AI Language.
• Entity recognition vs key phrases: Entities are typed items; key phrases are important concepts.
• Sentiment is opinion: Sentiment analysis is not entity extraction or summarization.
• Translation can be text or speech: Read the input and output format before choosing a service.

Flashcards

Flashcards

1 / 4

❓

Which NLP feature identifies people, locations, and dates in text?

(Click to reveal)

💡

Entity recognition.

---

<- Previous Domain | Overview | Next Domain ->