AI-900 Practice Question: Describe features of Natural Language Processing workloads on Azure

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

• ✗

  Summarising how many messages were exchanged in a conversation

  Why it's wrong here

  Summarising how many messages were exchanged in a conversation is a purely quantitative, metadata-level operation that counts turns or utterances without understanding their content. Conversation summarization, in contrast, applies natural language processing to generate a qualitative, semantic synthesis of what was discussed, such as issues raised, decisions made, and follow-up actions. A message count is a simple descriptive statistic that provides no insight into the substance of the dialogue, so it falls outside the definition of conversation summarization.

• ✓

  Generating concise summaries of dialogues capturing key points, decisions, and action items

  Why this is correct

  Generating concise summaries of dialogues capturing key points, decisions, and action items is precisely what conversation summarization does in Azure AI Language. This service takes a conversation transcript—from calls, chats, or meetings—and uses abstractive and extractive techniques to produce a distilled narrative that allows reviewers to grasp the essence without reading the full transcript. It identifies speaker-specific contributions, highlights action items, and preserves the overall context, making it the correct and most complete description of the capability.

• ✗

  A tool for moderators to summarise flagged content for compliance review

  Why it's wrong here

  A tool for moderators to summarise flagged content for compliance review is a specialized scenario that might use conversation summarization, but it is not the fundamental definition. Conversation summarization is a general-purpose capability applicable to any dialogue—support calls, sales meetings, or telehealth visits—and is not restricted to flagged or moderated content. Compliance review often also involves additional services like content moderation and sensitive-data analysis to handle policy matters. Thus, limiting it to moderator workflows mischaracterises the broad nature of conversation summarisation.

• ✗

  Automatically creating FAQ articles from the most common chatbot conversations

  Why it's wrong here

  Automatically creating FAQ articles from the most common chatbot conversations is a downstream use case that relies on conversation summarization, but it is not the definition of that capability. FAQ generation requires extracting and clustering question-answer pairs from repeated interactions, which involves intent detection and answer extraction rather than producing a holistic distillation of a single dialogue. Conversation summarization condenses the entire dialogue into key points, decisions, and action items, not specifically Q&A pairs. Therefore, it is an application of summarization, not a description of the core capability.

Go deeper

Related to this question