Amazon Comprehend

Learn how Amazon Comprehend helps to analyze and extract meaningful insights from a document together with its use cases.

Amazon Comprehend, a natural language processing (NLP) serverless service, is trained to identify important components in data, such as language, people, places, and others. It is designed to help developers analyze and extract meaningful insights from the content of documents.

Press + to interact

The service can show the confidence score of the text, assess the text’s positivity or negativity, and automatically group a number of text files into different categories according to their topics.

Press + to interact
Working of Amazon Comprehend
Working of Amazon Comprehend

Amazon Comprehend can be used to gather insights, build custom models, provide topic modeling, and for document processing. The most common insights that Amazon Comprehend can extract from the text are entities, key phrases, personally identifiable information (PII), sentiments, and language. It can also be customized to recognize specific terms or phrases within a document.

Amazon Comprehend uses a learning model to process a cluster of documents, understand their topics, and determine the context of each document. Amazon Comprehend works in three modes: single-document, multi-document, and batch processing, depending on the number of documents it processes.

Use case examples

Amazon Comprehend has distinct features and limitations. Let’s explore which features can be utilized for specific scenarios.

Extracting key topics from customer reviews

Consider a use case where a large tech company wants to understand the key topics or themes that customers are discussing in their reviews of a newly released smartphone.

In this scenario, the company can use Amazon Comprehend to extract these themes automatically from the reviews. Amazon Comprehend’s topic modeling capabilities will let the company analyze latent topics or themes within the reviews. The company would also be able to get insights into the discovered topics along with the keywords associated with each topic. 

Press + to interact
Topic modeling by Amazon Comprehend
Topic modeling by Amazon Comprehend

Hiding personally identifiable information

Consider another scenario where a telecommunications company manages a large volume of customer data. It has to protect customers’ personal information to maintain customer trust and comply with privacy regulations. In this scenario, Amazon Comprehend's PII entity detection emerges as an ideal solution, capable of scanning customer interaction logs, emails, and chat transcripts to identify and redact sensitive PII effectively.

Press + to interact
Safeguarding customer privacy with Amazon Comprehend
Safeguarding customer privacy with Amazon Comprehend

Get hands-on with 1300+ tech skills courses.