Microsoft Azure AI Fundamentals (AI-900)Describe features of Natural Language Processing (NLP) workloads on AzureHard
A customer support center transcribes all incoming calls for quality assurance. The management team wants to quickly understand the main subjects and recurring issues discussed across thousands of call transcripts without having to manually read through them. They need to identify high-level themes, not specific keywords or sentiments. Which Azure AI NLP capability would be most effective for this analysis?
- AText Classification
- BNamed Entity Recognition (NER)
- CTopic Modeling
- DText Summarization
Show answer & explanationAnswer & explanation
Correct answer: C. Topic Modeling
Topic Modeling is an unsupervised technique that automatically identifies abstract 'topics' (clusters of related words) within a large collection of texts, making it ideal for discovering high-level themes and recurring issues in call transcripts without prior categorization.
Why the other options are wrong
- A. Text Classification assigns documents to pre-defined categories, but the requirement is to 'understand the main subjects... without having to manually read through them', implying discovery, not pre-definition.
- B. NER extracts specific named entities (people, organizations), not overarching themes.
- D. Text Summarization condenses individual transcripts, it doesn't find common themes across a large dataset.
Topic Modeling
An unsupervised machine learning technique that identifies abstract 'topics' or latent semantic structures within a collection of text documents by analyzing word co-occurrence patterns.
- Unsupervised discovery of themes.
- Reveals underlying structure of a corpus.
- Useful for large datasets where themes are not predefined.
Memory trick: Topics float up from calls like clouds from vapor.