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?
- ANamed Entity Recognition (NER) with pre-built models
- BKey Phrase Extraction
- CSentiment Analysis
- DCustom Named Entity Recognition (NER)
Show answer & explanationAnswer & 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.