关于“Document Understanding”活动

Document Understanding 活动允许您对文档进行处理、分类和提取，以便自动解释文档。

Document Understanding 活动是文档处理的关键组件。 它允许系统解释文档的含义并从中提取信息，从而使完整文档处理成为可能。 您可以使用文档分类和数据提取方法从具有不同结构的多个文档中提取和解释信息。

总体而言，文档要经过以下五个基本步骤：

• 定义文档类型和要提取的数据；
• 提供文本及其位置；
• 对指定列表中的文档进行分类；
• 提取信息；
• 由人工确认提取的数据。

然后，可以按照您想要的方式导出和使用提取的数据。

Document Understanding 活动包括以下包：

• Document Processing Contracts: Contains data structures and interfaces to be used for integrating with the Document Understanding framework.
• Document Understanding OCR Local Server: Contains no activities, but once installed in a project, it enables you to use UiPath Document OCR running locally on the robot machine.
• Document Understanding Process: A template that provides logging, exception handling, retry mechanisms, and all the methods that should be used in a Document Understanding workflow, out of the box.
• Document Understanding: Allows you to process documents.
• Intelligent OCR: Contains the infrastructure for enabling document processing workflows.
• ML Services: Allows you to retrieve and call all ML Skills available on your tenant, as well as upload files or create labeling tasks.
• OCR: Contains the UiPath Screen OCR and UiPath Document OCR activities which use UiPath's in-house OCR engines.
• OCR Contracts: Contains the contracts needed to develop new OCR engines.
• OmniPage: Contains the OmniPage OCR activity which extracts data from an indicated UI (User Interface) element or image using the OmniPage OCR engine.
• PDF: Contains activities designed to extract data from PDF and XPS files.

新式项目和活动

Visit Activity or API-based limits to check the current specifications and parameters for optimal utilization of Document Understanding modern projects and Document Understanding activities.

项目兼容性

使用下表确定平台和活动包之间的兼容性：

Table 1. Document Understanding activities project compatibility

	Windows - 旧版 Studio 项目	Windows (新式) Studio 项目	跨平台 Studio 项目
Document Understanding OCR 本地服务器	✅	✅	✅
Document Understanding 活动	❌	✅	✅
智能 OCR	✅	✅	❌
ML 服务	✅	✅	❌
OCR	✅	✅	❌
OCR 合同	✅	✅	✅
omnipage	✅	✅	❌
PDF	✅	✅	❌