Document history for Amazon Comprehend
The following table describes the documentation for this release of Amazon Comprehend.
| Change | Description | Date |
|---|---|---|
Amazon Comprehend now supports custom classifier training with native documents. For more information, see Training classification models in Amazon Comprehend. | April 19, 2023 | |
Amazon Comprehend now supports flywheels to help you manage the training and tracking of model versions for custom models. For more information, see Flywheels in Amazon Comprehend. | February 28, 2023 | |
Updated the IAM security topics to include federated identities. For more information, see Identity and Access Management for Amazon Comprehend. | December 22, 2022 | |
Amazon Comprehend now automatically performs the text extraction for image, PDF, or Word input documents prior to running custom classification or custom entity recognition. For more information, see Document processing in Amazon Comprehend. | December 1, 2022 | |
Amazon Comprehend now supports synchronous APIs and console real-time analysis for targeted sentiment. Targeted sentiment determines the sentiment associated with specific entities in a document. For more information, see Targeted sentiment in Amazon Comprehend. | September 21, 2022 | |
Amazon Comprehend has reduced the minimum requirements for training a recognizer with plaintext CSV annotation files. You can now build a custom entity recognition model with as few as three annotated documents and at least 25 annotations per entity type. For more information, see Preparing the training data. | August 3, 2022 | |
Amazon Comprehend now supports up to a 100KB input document for most real-time APIs. For more information, see Guidelines and quotas. | July 18, 2022 | |
Additional PII entity types now detected by Amazon Comprehend. For more information, see Detecting PII entities in Amazon Comprehend. | May 20, 2022 | |
Restructured the Amazon Comprehend Developer Guide table of contents for easier navigation. For more information, see What is Amazon Comprehend. | April 7, 2022 | |
Amazon Comprehend now supports targeted sentiment analysis, which determines the sentiment associated with specific entities in a document. For more information, see Targeted sentiment in Amazon Comprehend. | March 9, 2022 | |
Amazon Comprehend now allows you to analyze images for custom entitiy recognition. For more information, see Detecting custom entities in Amazon Comprehend. | February 28, 2022 | |
You can now copy trained custom models between AWS accounts. For more information, see Copying custom models between accounts in Amazon Comprehend. | February 2, 2022 | |
You can now use AWS Trusted Advisor to view recommendations that can help you optimize the cost and security of your Amazon Comprehend endpoints. For more information, see Using Trusted Advisor with Amazon Comprehend. | September 29, 2021 | |
Amazon Comprehend has launched a suite of features for Comprehend Custom which enable continuous model improvements by giving you the ability to create new model versions, continuously test on specific test sets, and perform live migration to new model endpoints. | September 21, 2021 | |
Amazon Comprehend now allows you to analyze PDF and Word documents for custom entitiy recognition. With PDF and Word formats, you can extract information from documents containing headers, lists and tables. | September 14, 2021 | |
Amazon Comprehend has launched a new endpoints overview feature which provides you a global view of your endpoints. From the endpoints overview page, you can view all of your endpoints in one place to understand your endpoint usage versus your actual resource usage. | August 24, 2021 | |
Amazon Comprehend Medical now allows you to establish a private connection with your Virtual Private Cloud (VPC) by creating an interface VPC endpoint. For more information, see VPC endpoints(PrivateLink). | June 13, 2021 | |
Amazon Comprehend has added four additional languages for the dominant language feature: Hausa (ha), Lao (lo), Maltese (mt), and Oromo (om). For more information, see Supported languages in Amazon Comprehend. | May 10, 2021 | |
With Amazon Comprehend, you can now encrypt custom models using a customer managed key (CMK). For more information, see KMS encryption in Amazon Comprehend. | March 31, 2021 | |
You can now use Amazon S3 Object Lambda Access Points to configure how documents that contain personally identifiable information (PII) are retrieved from your Amazon S3 bucket. You can control access of documents that contain PII and redact PII from documents. For more information, see Using Amazon S3 object Lambda access points for personally identifiable information (PII). | March 18, 2021 | |
You can now label a document with personally identifiable information (PII). Amazon Comprehend can analyze your document for the presence of PII and return the labels of identified PII entity types such as name, address, bank account number, or phone number. For more information, see Label document with PII. | March 11, 2021 | |
With Amazon Comprehend, you can now detect events in a set of documents. When you create an asynchronous events detection job, Amazon Comprehend can detect supported types of financial events. For more information, see Detect events. | November 24, 2020 | |
Amazon Comprehend now allows you to use auto scaling for custom entity recognizer endpoints. With auto scaling, you can automatically set endpoint provisioning to fit your capacity needs. For more information, see Auto scaling with endpoints. | September 28, 2020 | |
To train custom classifiers or entity recognizers, you can now provide augmented manifest files, which are labeled datasets that are produced by Amazon SageMaker Ground Truth. For more information about these files, and for examples, see Multi-class mode, Multi-label mode, and Annotations. | September 22, 2020 | |
Amazon Comprehend now has a tutorial that walks you through a multi-service workflow of analyzing customer reviews and visualizing the analysis results. For more information, see Tutorial: Analyzing insights from reviews. | September 17, 2020 | |
With Amazon Comprehend, you can now detect entities in your text that contain personally identifiable information (PII), such as addresses, bank account numbers, or phone numbers. Amazon Comprehend can provide the location of each PII entity in your text, or it can provide a copy of your text in which the PII is redacted. For more information, see Detect personally identifiable information (PII). | September 17, 2020 | |
Previously, you could only train a model on up to 12 custom entities. Now Amazon Comprehend allows you to train a model on up to 25 custom entities at a time. For more information, see Custom entity recognition. | August 12, 2020 | |
Amazon Comprehend has added five additional languages for the custom entity recognition feature: German (de), Spanish (es), French (fr), Italian (it), and Portuguese (pt). For more information, see Supported languages in Amazon Comprehend. | August 12, 2020 | |
Amazon Comprehend now allows you to establish a private connection with your Virtual Private Cloud (VPC) by creating an interface VPC endpoint. For more information, see VPC endpoints (AWS PrivateLink). | August 11, 2020 | |
With Amazon Comprehend, you can now quickly detect custom entities in individual text documents by running real-time analysis. For more information, see Detecting custom entities in real time with amazon comprehend. | July 9, 2020 | |
Amazon Comprehend now provides support for a second mode in asynchronous Custom Classification for documents that provides greater flexibility when applying custom classes to documents. While multi-class mode associates only a single class with each document, the new multi-label mode can associate more than one. For example, a movie can be classified as both science fiction and action at the same time. For more information, see Multi-class and multi-label modes in custom classification. | December 19, 2019 | |
Amazon Comprehend now provides support for real-time Custom Classification for documents with unstructured text. Customers can use real-time custom classification to understand, label and route information based on their own business rules, synchronously. For more information, see Real-time analysis with custom classification. | November 25, 2019 | |
Amazon Comprehend has added six additional languages for several of its features: Arabic (ar), Hindi (hi), Japanese (ja), Korean (ko), simplified Chinese (zh), and traditional Chinese (zh-TW). These new languages are supported only for Determine Sentiment, Detect Key Phrases, and non-custom Detect Entities operations. For more information, see Supported languages. | November 6, 2019 | |
Previously, you could only train a model on a single custom entity. As a result, you could only search for that one entity with an entity recognition operation. Amazon Comprehend has changed this and you can now train a model on up to 12 custom entities at a time. For more information, see Custom entity recognition | July 9, 2019 | |
Amazon Comprehend now provides a multi-class confusion matrix for added ability to analyze metrics when training a Custom Classifier. This is currently supported using the APIs only. For more information, see Tagging resources in Amazon Comprehend | April 5, 2019 | |
Amazon Comprehend provides tags for Custom Classifiers and Custom Entity Recognizers, which can be used as metadata that enables you to organize, filter, and control access to your resources with a finer level of control than ever. For more information, see Tagging resources in Amazon Comprehend | April 3, 2019 | |
Amazon S3 already enables you to encrypt your input documents, and Amazon Comprehend extends this even farther. By using your own KMS key, you can not only encrypt the output results of your job, but also the data on the storage volume attached to the compute instance that processes the analysis job. The result is end-to-end security. For more information, see KMS encryption in Amazon Comprehend | March 28, 2019 | |
Custom entity recognition extends the capability of Amazon Comprehend by enabling you to identify new entity types not supported as one of the preset generic entity types. This means you can analyze documents and extract entities like product codes or business-specific entities that fit your particular needs. For more information, see Custom entity recognition | November 16, 2018 | |
You can use Amazon Comprehend to build your own models for custom classification, assigning a document to a class or a category. For more information, see Document classification. | November 15, 2018 | |
Amazon Comprehend is now available in Europe (Frankfurt) (eu-central-1). | October 10, 2018 | |
In addition to English and Spanish Amazon Comprehend can now also examine documents in French, German, Italian, and Portuguese. For more information, see Supported languages in Amazon Comprehend. | October 10, 2018 | |
Amazon Comprehend is now available in Asia Pacific (Sydney) (ap-southeast-2). | August 15, 2018 | |
Amazon Comprehend now parses documents to discover the syntax of a document and the part of speech for each word. For more information, see Syntax. | July 17, 2018 | |
Amazon Comprehend now supports asynchronous batch processing for language, key phrase, entity, and sentiment detection. For more information, see Asynchronous batch processing . | June 27, 2018 | |
New guide | This is the first release of the Amazon Comprehend Developer Guide. | November 29, 2017 |
