Implement text analysis solutions 4.1 Apply language model text analysis
A support center must remove names, phone numbers and account IDs from chat transcripts…
Study guide skill: Configure detection of sentiment, tone, safety issues, and sensitive content
Question B1-037
A support center must remove names, phone numbers and account IDs from chat transcripts and documents before storing them for analytics. Which capability should you use?
Select one answer.
Answer: B
PII detection identifies, classifies and redacts sensitive data across text, conversations and native documents, returning entity categories, confidence scores and redacted output. Sentiment analysis and translation don't redact, and Content Safety moderation targets harmful content rather than personal data.
Deep dive
How it works
PII detection is a core Azure Language capability that identifies, classifies and redacts sensitive data in text, conversations and native documents. You get back structured output with entity categories, confidence scores and redacted results that depend on your API configuration. You can customize entity types, masking styles and exclusions.
PII detection comes in three feature types, so pick the one that matches your data:
- Text PII handles string payloads and returns detection and redaction results synchronously, for messages, prompts, logs and other text fields.
- Conversation PII handles multi-turn exchanges and transcripts, where turn boundaries and conversation context affect detection and masking. It runs asynchronously and suits contact centers, meetings and voice transcripts, which makes it the fit for this center's chat transcripts.
- Document-based PII handles native
.pdf,.docxand.txtfiles in asynchronous, storage-based pipelines, and returns redacted output that preserves the document structure. That covers the documents.
The other options, and when you'd use them
- Azure Translator: Translator is a neural machine translation service. Use it when content needs to be in another language. It changes the language of the text, and it isn't a redaction feature.
- Azure AI Content Safety text moderation: Content Safety detects harmful user-generated and AI-generated content. Use it to moderate abusive or unsafe text, not to remove personal data.
- Azure Language sentiment analysis: Sentiment analysis labels text as negative, neutral or positive, with confidence scores at sentence and document level. Use it to learn how customers feel, not to redact their details. Its Learn page currently carries a retirement notice, so check it before you start new work on it.
Exam tip
"Remove names, phone numbers and IDs before storing" means PII detection with redaction. Choose Text, Conversation or Document-based PII depending on whether the input is strings, transcripts or files.
References
Last reviewed: