Amazon Comprehend

What is Amazon Comprehend?

Amazon Comprehend is an Amazon Web Services (AWS) service that uses natural language processing (NLP) and machine learning to extract insights from unstructured text. It analyzes sentence structure, identifies entities such as names, places, and organizations, and detects sentiment (positive, negative, or neutral). It supports content sorting, trend spotting, and processing customer feedback at scale. Because it runs on AWS infrastructure, it handles large volumes of text efficiently and reduces manual analysis work.

Who created Amazon Comprehend?

Amazon Web Services (AWS), the cloud computing arm of Amazon, developed Amazon Comprehend and launched it on January 29, 2017. The service applies machine learning to sentiment analysis, entity recognition, and key phrase extraction.

What is Amazon Comprehend used for?

• Business and call center insights
• Customer sentiment analysis and automated categorization of support requests
• Product improvement through customer survey analysis
• Search and indexing of product reviews by key phrases, entities, and sentiment
• Legal document analysis (contracts, court records)
• Financial document classification and entity extraction
• Employee feedback analysis, including Targeted Sentiment
• Content moderation and toxic content detection
• Generative AI safety via unsafe input prompt detection
• Topic modeling across document collections
• PII detection and redaction
• Relationship discovery between financial events
• Document categorization

Who is Amazon Comprehend for?

Legal professionals managing briefs, financial analysts processing financial documents, and customer support professionals analyzing call center analytics and categorizing inbound requests. In general, it suits anyone who needs to extract information from large volumes of text.

Pricing and usage

• Core NLP tasks: The APIs handle entity recognition, sentiment analysis, key phrase extraction, and language detection. Requests are measured per 100 characters, with a minimum charge of 3 units per request.
• PII handling: The service finds and redacts personally identifiable information, priced on the same 100-character unit structure.
• Custom models: Train custom models for text categorization or entity extraction. Asynchronous inference is charged per character, training has an hourly fee, and model management carries a monthly cost.
• Topic modeling: Priced based on the total size of documents processed in a job.
• Trust and safety features: Toxic content and prompt safety detection are priced per 100-character units with a minimum charge.
• Cost estimation: Use the AWS Pricing Calculator to estimate costs for your tasks and data volume.
• Free tier: AWS offers a free tier covering certain text and document sizes for eligible APIs, available to new and existing AWS customers.

Factor in training, inference, model management, and endpoint costs when planning real-time classification.