A developer is building a system to assist legal professionals by automatically identifying and extracting specific legal clauses, contract dates, and party names from scanned PDF documents. Which combination of Azure AI services would be most effective for this scenario?
- AAzure AI Translator and Azure AI Language (Language Understanding).
- BAzure AI Language (Sentiment Analysis) and Azure AI Vision for Object Detection.
- CAzure AI Language (Text Analytics) for Key Phrase Extraction and Azure AI Vision for OCR.
- DAzure AI Vision for OCR and Azure AI Language (Custom Named Entity Recognition).
Show answer & explanationAnswer & explanation
Correct answer: D. Azure AI Vision for OCR and Azure AI Language (Custom Named Entity Recognition).
To process scanned PDF documents, Optical Character Recognition (OCR) from Azure AI Vision is essential to convert images of text into machine-readable text. Once text is extracted, Custom Named Entity Recognition (Custom NER) from Azure AI Language is needed to specifically identify and extract unique entities like legal clauses, contract dates, and party names that are not standard general entities.
Why the other options are wrong
- A. Azure AI Translator is for language translation, and Language Understanding (LUIS) is for interpreting user intent in conversational AI, neither addresses entity extraction from scanned documents.
- B. Sentiment Analysis determines emotional tone, not specific entities, and Object Detection is for identifying objects in images, not text.
- C. Key Phrase Extraction is too general; it won't extract specific legal clauses or dates. OCR is correct but the NLP part is insufficient.
OCR + Custom NER
A powerful combination for extracting domain-specific information from scanned or image-based documents, where OCR converts the image to text and Custom NER identifies specialized entities.
- OCR is crucial for non-digital text sources.
- Custom NER allows training for unique entity types.
- Ideal for legal, medical, or industry-specific document processing.
Memory trick: Scanned words need seeing, then custom finding.