Curriculum Overview: Natural Language Processing (NLP) Workloads

This document provides a structured roadmap for mastering Natural Language Processing (NLP) as part of the Azure AI Fundamentals (AI-900) curriculum. NLP is a critical AI workload that empowers machines to understand, interpret, and generate human language.

Prerequisites

Before beginning this module, learners should have a foundational understanding of the following:

• General AI Concepts: Understanding that AI is a broad field including machine learning and deep learning.
• Cloud Fundamentals: Familiarity with the basic concept of "Software as a Service" (SaaS) and how Azure provides AI resources.
• Basic Data Literacy: Understanding the difference between structured data (tables) and unstructured data (text, audio).

Module Breakdown

Learning Objectives per Module

Module 1: Foundations of NLP

• Define NLP as the workload dedicated to human language analysis.
• Distinguish NLP from other AI workloads like Computer Vision or Document Intelligence.

Module 2: Text Analysis Scenarios

• Identify how Key Phrase Extraction identifies main talking points.
• Understand Entity Recognition (NER) and Entity Linking (NEL) for identifying people, places, and dates.
• Determine the emotional tone of text using Sentiment Analysis.

Module 3: Speech & Translation

• Describe the process of converting audio to text (Speech Recognition) and text to audio (Speech Synthesis).
• Explain how Machine Translation enables multilingual communication.

Module 4: Conversational AI

• Describe the role of Conversational Language Understanding (CLU) in identifying user intent.
• Explain how NLP powers automated customer support agents and bots.

Visual Overview of NLP Scenarios

Success Metrics

To demonstrate mastery of this curriculum, the learner should be able to:

1. Categorize Scenarios: Given a business problem (e.g., "We need to know if customers are happy"), identify that Sentiment Analysis is the correct NLP tool.
2. Differentiate Services: Correct identify when to use Azure AI Language (for text) versus Azure AI Speech (for audio).
3. Understand Intent: Explain how CLU helps a bot understand that "Book a flight" and "I want to fly" represent the same user intention.
4. Identify Entities: Point out that in the sentence "I live in New York," "New York" is a Location entity identified through Named Entity Recognition (NER).

Real-World Application

NLP is not just theoretical; it is used across industries to solve complex communication problems:

[!TIP] Social Media Monitoring: Companies use sentiment analysis to track brand reputation in real-time by analyzing millions of tweets and posts.

Industry Examples

\begin{tikzpicture}[node distance=2cm] \node (center) [draw, rectangle, fill=blue!10, inner sep=10pt] {\textbf{NLP in Industry}}; \node (health) [above of=center, xshift=-3cm, draw, rounded corners, fill=green!5] {Healthcare: Medical Record Coding}; \node (retail) [above of=center, xshift=3cm, draw, rounded corners, fill=green!5] {Retail: Global Customer Support}; \node (finance) [below of=center, xshift=-3cm, draw, rounded corners, fill=green!5] {Finance: Document Summarization}; \node (tech) [below of=center, xshift=3cm, draw, rounded corners, fill=green!5] {Tech: Personal Voice Assistants}; \draw [->] (center) -- (health); \draw [->] (center) -- (retail); \draw [->] (center) -- (finance); \draw [->] (center) -- (tech); \end{tikzpicture}

• Global Support: Using Machine Translation, a support center in London can assist a customer in Tokyo in real-time.
• Accessibility: Speech Synthesis (Text-to-Speech) allows visually impaired users to consume written digital content.
• Efficiency: Key Phrase Extraction allows legal firms to summarize thousands of pages of documents quickly to find relevant case laws.