Curriculum Overview: Unit 4 - Natural Language Processing (NLP) on Azure
Unit 4: Describe features of Natural Language Processing (NLP) workloads on Azure (15–20%)
Curriculum Overview: Natural Language Processing (NLP) Workloads on Azure
This unit covers the fundamental concepts and Azure-specific services related to Natural Language Processing (NLP). NLP is the branch of AI that enables computers to understand, interpret, and generate human language in both written and spoken forms. This module accounts for 15–20% of the AI-900 exam content.
Prerequisites
Before starting this unit, learners should have a basic understanding of the following:
- Cloud Computing Basics: Familiarity with Microsoft Azure (resource groups, regions, and the Azure Portal).
- AI Fundamentals: Understanding of general AI workloads as covered in Unit 1 (e.g., how AI differs from traditional software).
- Machine Learning Concepts: A high-level grasp of training and validation datasets (Unit 2).
Module Breakdown
| Module | Focus Area | Difficulty |
|---|---|---|
| 1. NLP Core Scenarios | Key phrase extraction, sentiment analysis, and entity recognition. | Beginner |
| 2. Azure AI Language | Features and capabilities of the unified Language service. | Intermediate |
| 3. Azure AI Speech | Speech-to-text, text-to-speech, and translation. | Intermediate |
| 4. Conversational AI | Conversational Language Understanding (CLU) and Chatbots. | Intermediate |
Learning Objectives per Module
Module 1: NLP Core Scenarios
- Define Tokenization & N-grams: Understand how text is broken down into smaller units for processing.
- Sentiment Analysis: Identify how machines assign sentiment scores (positive, negative, neutral) to text.
- Entity Recognition: Differentiate between Named Entity Recognition (NER) and Named Entity Linking (NEL).
Module 2: Azure AI Language Capabilities
- Key Phrase Extraction: Identify the main talking points in a document.
- Language Detection: Automatically identify the language in which text is written.
- Translation: Understand the mechanics of machine translation for text.
Module 3: Azure AI Speech Capabilities
- Speech Recognition: Convert spoken audio into text (Speech-to-Text).
- Speech Synthesis: Convert text into life-like spoken audio (Text-to-Speech).
- Real-time Translation: Understand how speech can be translated into other languages or text formats.
Visual Overview of NLP Services
Success Metrics
You will have mastered this unit when you can:
- Select the Right Tool: Correctly choose between Azure AI Language and Azure AI Speech for a given business scenario.
- Explain Sentiment Scoring: Describe how a score of 0.9 vs 0.1 impacts business decision-making.
- Identify Entities: Look at a sentence (e.g., "I visited Microsoft in Seattle") and identify the entities (Organization: Microsoft, Location: Seattle).
- Differentiate Synthesis vs. Recognition: Clearly explain which process is used for a Voice Assistant (Synthesis) vs. a Meeting Transcription tool (Recognition).
Real-World Application
NLP is one of the most visible forms of AI in modern industry. Understanding these Azure services allows you to build:
- Customer Support Bots: Using Sentiment Analysis to escalate angry customers to human agents immediately.
- Accessibility Tools: Using Speech-to-Text to provide real-time captioning for the hearing impaired.
- Global Communication: Using Translation services to localize documentation or support international business meetings in real-time.
- Content Insights: Using Key Phrase Extraction to automatically tag thousands of news articles or research papers for easier searching.
[!TIP] In the AI-900 exam, remember that Azure AI Language handles the meaning and structure of text, while Azure AI Speech handles the conversion between audio and text.