Skip to main content

Amazon Comprehend

What is it​

A natural language processing (NLP) service that uses machine learning to discover information and relationships in unstructured text.

What is it for​

Analyze text to identify entities (people, places, organizations), sentiments (positive, negative, neutral), key phrases, languages, and topics.

Use cases​

  • Customer feedback analysis (reviews, comments, emails)
  • Document and article classification
  • Information extraction from legal or medical documents
  • Social media analysis for trends and sentiments
  • Building text-based search and recommendation applications

Key points​

  • Managed NLP: No prior machine learning knowledge required to use
  • Multiple features: Language detection, sentiment analysis, entity extraction, key phrase extraction, topic detection
  • Customization: Allows training custom models for entities and text classification
  • Amazon Comprehend Medical: A specialized version for extracting information from clinical text
  • Integration: Integrates with other AWS services like S3, Lambda, and Kinesis

Comparison​

  • Amazon Comprehend: Offers a managed and scalable NLP solution, eliminating the need to build and train ML models from scratch. Ideal for those who need quick results and don't want to manage ML infrastructure.
  • Manual NLP/libraries (e.g., NLTK, SpaCy): Offers greater control and flexibility for customization but requires knowledge in ML and programming, plus infrastructure management. Suitable for very specific use cases or research.