Describe features of Natural Language Processing (NLP) workloads on Azure for Microsoft Azure AI Fundamentals (AI-900)

Sample Practice Questions

1. Question 1

   A company employs a team of customer service agents to provide telephone and email support to customers. The company develops a webchat bot to provide automated answers to common customer queries. Which business benefit should the company expect as a result of creating the webchat bot solution?
   1. A. increased sales
   2. B. a reduced workload for the customer service agents
   3. C. improved product reliability
   Explanation

   The correct answer is: B. a reduced workload for the customer service agents.

   When a webchat bot fields the routine, frequently asked questions that customers send, the human agents are freed from repetitive triage and can spend their time on harder cases that genuinely need expertise. That shift translates directly into a reduced workload for the customer service team, which is the primary and most easily measured business outcome of deploying a Q&A bot. Increased sales is not a logical consequence of automating support answers; the bot is reactive and does not actively cross-sell or generate leads. Improved product reliability would come from engineering or quality changes to the product itself, not from a chat interface that explains existing features. The realistic, direct benefit of the automation is therefore reduced agent workload.

2. Question 2

   You are developing a solution that uses the Text Analytics service. You need to identify the main talking points in a collection of documents. Which type of natural language processing should you use?
   1. A. entity recognition
   2. B. key phrase extraction
   3. C. sentiment analysis
   4. D. language detection
   Explanation

   The correct answer is: B. key phrase extraction.

   Identifying the main talking points in a collection of documents is exactly what key phrase extraction is designed for. The feature, part of the Azure AI Language service (formerly Text Analytics), scans each document and returns the most salient noun phrases, which serve as a concise topic summary you can use to compare or cluster documents by subject. Entity recognition tags named items such as people, places, organisations, and dates with category labels; it is great for pulling structured facts out of text but does not surface the broader themes of a document. Sentiment analysis scores polarity (positive, neutral, negative) and tells you how the writer feels rather than what they are writing about. Language detection only identifies which language a document is written in by returning an ISO code, contributing nothing about subject matter. So when the goal is to surface the central topics across many documents, key phrase extraction is the right NLP workload.

3. Question 3

   In which two scenarios can you use speech recognition? Each correct answer presents a complete solution. NOTE: Each correct selection is worth one point.
   1. A. an in-car system that reads text messages aloud
   2. B. providing closed captions for recorded or live videos
   3. C. creating an automated public address system for a train station
   4. D. creating a transcript of a telephone call or meeting
   Explanation

   The correct answers are: B. providing closed captions for recorded or live videos, D. creating a transcript of a telephone call or meeting.

   Speech recognition (the Speech-to-Text capability of Azure AI Speech) takes spoken audio as input and produces text as output, so any scenario that turns spoken words into written text is a fit. Providing closed captions for recorded or live videos is a classic speech-to-text use case because the soundtrack must be transcribed in near real time and overlaid on the video. Creating a transcript of a telephone call or meeting is the same pattern at a longer time scale; speech recognition converts each speaker's voice into a written record that can later be searched or summarised. An in-car system that reads text messages aloud is the opposite direction: it takes written text and synthesises spoken audio, which is text-to-speech (speech synthesis), not recognition. An automated public address system for a train station likewise broadcasts spoken announcements generated from text, which is again speech synthesis rather than recognition.

4. Question 4

   HOTSPOT - To complete the sentence, select the appropriate option in the answer area. Hot Area:
   Explanation

   Reference: https://azure.microsoft.com/en-gb/services/cognitive-services/speech-to-text/#features

5. Question 5

   You need to build an app that will read recipe instructions aloud to support users who have reduced vision. Which version service should you use?
   1. A. Text Analytics
   2. B. Translator
   3. C. Speech
   4. D. Language Understanding (LUIS)
   Explanation

   The correct answer is: C. Speech.

   Reading recipe steps aloud requires text-to-speech (also called speech synthesis), which is delivered by Azure AI Speech. Speech synthesis converts written content into natural-sounding spoken audio using neural voices, exactly what an accessibility app for low-vision users needs. Text Analytics, a subset of the Language service, handles sentiment, key phrases, and entity recognition over text and produces no audio output. The Translator service moves text between languages and again has no speech generation capability. Language Understanding (LUIS, now CLU) classifies free-text into intents and entities; it can help a bot understand voice commands paired with speech-to-text but cannot read text aloud. Only the Speech service natively produces spoken output, making it the correct service for this scenario.

Other Microsoft Azure AI Fundamentals (AI-900) domains

• Describe Artificial Intelligence workloads and considerations (76 questions)
• Describe features of computer vision workloads on Azure (46 questions)
• Describe fundamental principles of machine learning on Azure (58 questions)