Microsoft Certified: Azure AI Engineer AssociateImplement natural language processing solutionsHard

A legal firm needs to process thousands of legal documents to automatically extract specific clauses, dates, and party names. These documents follow a highly structured, yet varying, format. The firm wants to build a solution that can learn to recognize these specific data points without extensive manual rule creation for each document type. Which Azure AI Language feature is best suited for this scenario?

  1. ANamed Entity Recognition (NER) with pre-built models
  2. BKey Phrase Extraction
  3. CSentiment Analysis
  4. DCustom Named Entity Recognition (NER)
Show answer & explanation

Correct answer: D. Custom Named Entity Recognition (NER)

While pre-built NER models can detect common entities, legal documents often contain highly specific entities and clause structures that generic models won't recognize. Custom NER allows the legal firm to train a model with their own annotated legal documents to accurately extract specific clauses, dates (in legal contexts), and party names, adapting to their varying formats without hard-coding rules.

Why the other options are wrong

  • A. Pre-built NER might detect some common entities (like generic dates or person names), but it won't be accurate or specific enough for unique legal clauses and party roles.
  • B. Key Phrase Extraction identifies general topics, not specific structured data like clauses or party names.
  • C. Sentiment Analysis determines emotional tone, which is irrelevant for extracting structured data from legal documents.

Custom Named Entity Recognition (NER)

A feature within Azure AI Language that allows users to train their own NER models to identify and extract specific, domain-relevant entities from unstructured text, going beyond the capabilities of pre-built models.

  • Extracts domain-specific entities.
  • Requires labeled training data.
  • Ideal for specialized industries like legal or medical.

Memory trick: Custom NER precisely picks out specialized details.

More Implement natural language processing solutions questions