Amazon Comprehend

Teams can train custom text classification models and custom entity recognition models by uploading labeled training data in CSV or augmented manifest format. Comprehend handles all ML pipeline steps automatically — feature engineering, training, hyperparameter optimization, and model evaluation. Models can be deployed to real-time endpoints for synchronous inference or used in asynchronous batch jobs. Built-in metrics (precision, recall, F1) help evaluate model quality before deployment.

Identifies over 30 types of personally identifiable information — including names, addresses, Social Security numbers, credit card numbers, phone numbers, email addresses, dates of birth, bank account numbers, and driver's license numbers. Supports both detection mode (returns entity locations and types with confidence scores) and redaction mode (replaces PII with placeholder tags in the output text), enabling compliance with GDPR, CCPA, and other privacy regulations.

A specialized variant that extracts medical entities such as conditions, medications, dosages, procedures, test results, and anatomical terms from clinical text. Links extracted entities to standard medical ontologies including ICD-10-CM (diagnoses), RxNorm (medications), and SNOMED CT (clinical terms). Operates under AWS BAA for HIPAA-eligible workloads and supports processing of clinical notes, discharge summaries, pathology reports, and clinical trial documentation.

Goes beyond document-level sentiment to identify sentiment expressed toward specific entities mentioned in the text. For example, in a product review mentioning both battery life and screen quality, targeted sentiment can separately classify the sentiment toward each attribute. Returns entity-level sentiment scores (positive, negative, neutral, mixed) with confidence values, enabling granular aspect-based sentiment analysis without custom model training.

Processes large collections of documents stored in Amazon S3 via asynchronous batch jobs, supporting up to 5 GB of input data per job with individual documents up to 100 KB. Results are written back to S3 in JSON format. Supports all standard NLP APIs in batch mode, enabling cost-effective processing of millions of documents without managing infrastructure. Jobs can be monitored via CloudWatch and orchestrated using Step Functions or Lambda triggers.