文档类型 (Document Manager)

Document types allow you to prepare, review and make corrections to datasets required for Training and Evaluation of Document Understanding^TM Machine Learning models. It enables multiple users to perform a variety of operations:

• 定义并配置要由 ML 模型提取的字段。
• 导入要标注的文档。
• 使用已有的 ML 模型（例如 UiPath 提供的开箱即用型“发票提取”或“收据提取”模型）或使用 AI Center 训练的模型来预标记文档。
• 标注文档。
• 以 AI Center 训练管道所需的格式导出文档。

创建文档类型

Once a project is created and opened, you can create a new Document Type by selecting the New button and selecting the Using Semi-Structured AI option. A new window opens requesting additional information.

以下是创建新“文档类型”会话时可用的选项。通过填写这些值，您可以获得作为起点的建议数据集大小的估计值。

此外，通过选择“开箱即用”文档类型，系统会自动填充和配置您的架构，以使您能够从 AI Center 中提供的预训练模型中受益。这样您就无需手动导入预定义架构，并可加快工作速度，并减少代价高昂的错误。

此外，预加标签端点会自动填充适当的端点，使您可以更方便地在打开新的文档类型后立即使用预标签。

选项	描述
Name (Mandatory)	给新文档类型命名。
Out-of-the-box document type (Mandatory)	从下拉列表中选择一种可用的预训练开箱即用文档类型。
Out-of-the-box regular fields (Optional)	选择要为架构创建的预定义常规字段
Out-of-the-box column fields (Optional)	选择要从文档中提取的预定义列字段。
Out-of-the-box classification fields (Optional)	选择要从文档中提取的预定义分类字段。
Custom regular fields (Optional)	Enter the number of additional regular fields you would like to extract from your documents.
Custom column fields (Optional)	Enter the number of additionalcolumn fields you would like to extract from your documents.
Number of languages (Optional)	输入需要提取的文档的语言数量。
Number of layouts (Optional)	输入需要提取的文档的布局数量。

备注：

Selecting a document type generates a recommended number of pages that need to be used for the dataset.

用户界面

Document Manager 界面包含以下面板：

• 管理栏
• 列字段
• 常规字段
• 分类字段
• 文档视图

Management bar

显示在 Document Manager 中的页面顶部。

使您能够执行多项操作：在文档之间导航、删除/还原文档、搜索/筛选文档、运行 AI 模型预测以及导入和导出文档。

以下是管理栏中的可用项目：

项目	图标	描述
导航		在与活动筛选器匹配的文档之间导航。在两个箭头之间显示一个计数器。它显示了与活动搜索/筛选器匹配的文档总数和当前文档的编号。
搜索		有两项搜索功能： Built-in filters: filters the documents based on the batch/category available options from the drop-down menu. Using keywords: filters the documents based on a text input.
删除/还原		删除或还原文档。已删除的文档位于“已删除”筛选器下。
导入		打开“导入数据”对话框。
导出		打开“导出文件”对话框。
文档名称和类型	不适用	The name of the currently active document and its type. There are three type of documents: 训练文档 验证文档 评估文档 Training and Validation documents are part of training datasets used by Training Pipelines. Evaluation documents are ignored by Training Pipelines and are intended to only be used by Evaluation pipelines in AI Center. These documents are the ones that were marked as evaluation by selecting the Mark this an evaluation set checkbox in the Import data dialog box.
下载		The option is available in the drop-down next to the document name. Select the icon to download a Zip file containing the original document. Besides the original document, all pages converted internally by Document Manager to .jpeg images are downloaded as well.
永久删除		The option is available in the drop-down next to the document name. Permanently deletes individual files. The .pdf and all its .jpeg images are deleted from the AI Center dataset and all the metadata is deleted from the database. When selecting the button, a pop-up message appears asking you if you are sure you want to permanently delete the document. Choose OK to continue or Cancel to revert to the previous screen.
批次名称	不适用	当前批次的名称。
会话名称	不适用	当前会话的名称。
预测 Note: The Predict feature relies on the UiPath Helix Extractor, but only for tenants based in the Europe region. If your tenant is located in a region outside of Europe, this functionality uses the previous-generation model architecture.		Run AI model predictions and display the results. After configuring Prelabelling, the button is enabled in the management bar. Select it to prelabel the current document. The button has three options: 预测：合并预加标签端点（在“预加标签”设置中配置）和生成式预测的结果。如果未配置预加标签端点，则仅使用生成式预测来预测所有字段。 生成式预测：使用生成式预测功能预测所有字段。 模型预测：使用“预加标签”设置中配置的预加标签端点模型预测字段。 The Generative Annotation (prelabeling) functionality discards all manually edited field values for all field types and deletes all tags from the document. At the moment, using the Predict option with Public endpoints for Automation Cloud and Test Cloud prelabels only the first 10 pages of a document. This is a known issue and a fix is in the working. Using the Predict option with ML Skills in AI Center, however, does not impose such a limitation. Generative prelabeling does not consume AI Units when using public endpoints or skills deployed in Automation Cloud™ from Document Manager sessions hosted on Automation Cloud. Generative prelabeling consumes AI Units when calling a public endpoint from a session hosted on Automation Suite, or when calling a skill deployed in Automation Suite from a Document Manager session hosted on Automation Cloud.
设置		Configure OCR and Prelabelling settings or access the How to... panel.

“删除”和“永久删除”选项

Let's go a little bit deeper in understanding the difference between Delete and Permanently Delete options.

• The Delete option deletes the files, but not removing them entirely from your project. The deleted files can still be found under the deleted filter from the Search bar and restored by using the Restore option.
• The Permanently Delete option deletes the selected files without any possibility of restoring them.

搜索选项

Three search options are available in total, two are present in the management bar from the top of the page, and one is using the icon from the bottom left side of the page.

备注：

Please note that for Forms AI only the following built-in filters are available: deleted, labelled, unlabelled.

公开的搜索功能包含两个部分：

• Search using the built-in filters: filters the documents based on the batch/category available options from the drop-down menu.

  备注：

  Selecting more options makes the search more restrictive. For example selecting Batch import1 and Deleted is bringing up only the documents imported in the Batch import1 which are deleted. Take note of combinations that will always return an empty list: selecting Batch import1 and Batch import2 would never return a document since the selection is restrictive and no document can be in two batches at a time.

• Search in documents using keywords: this search bar filters the information based on a text input. You have to enter the keyword(s) as free text in the Search bar. The search looks for the keyword(s) in a document's content or the document name. Multiple words search returns results when the words are adjacent, excluding any punctuation in between them.

  备注：

  Alongside the two bar searches, there is also an inside the document search, identifiable by this sign.

• Search inside the document: allows you to search for instances of text solely in your current document. The search bar, , can be found at the bottom left hand side of the screen.

“设置”菜单

设置按钮有两个可用选项：

• Settings where you can configure the OCR service or Prelabelling
• How to... which has the purpose of a help menu

OCR

为了将文档导入 Document Manager，必须配置 OCR 服务。

OCR 方法

This setting is available only for Document Types (Data Manager sessions) created in AI Center. When created in Document Understanding^TM, this setting is inherited from the Project Settings. If you want to modify these settings in Document Understanding^TM, go back to the Project view and open Project Settings from the bottom left.

基于云的选项包括：

• UiPath® Document OCR - https://du.uipath.com/ocr;
• 中文、日语和韩语 OCR
• 具有最广语言覆盖范围的 Google Cloud Vision OCR；
• 适用于读取日语文档的 Google Cloud Vision OCR（日语版）；
• Microsoft Read OCR。

OCR URL

配置 OCR 时，需要 OCR 服务具有 URL。您可能会用到以下 URL：

• 公共 URL，例如 https://du.uipath.com/ocr，或来自 Google Vision OCR 或 Microsoft Read OCR 的第三方 URL

OCR 密钥

所选 OCR 引擎的相应 API 密钥。例如，对于 UiPath Document OCR，您需要使用 Document Understanding API 密钥。对于 Document Manager Cloud 和 Document Manager On-Prem Online，它是必需的。对于 Document Manager On-Prem Air-gapped，它不是必需的。

For more information, check the Cloud and on-premises usage page.

将 OCR 应用于 PDF

Establishes if the OCR process should be applied or not to PDF documents. If set to True, the OCR is applied to all PDF pages of the document. If set to False, only digitally typed text is extracted. When set to Auto evaluates if the document requires to apply the OCR algorithm depending on the input document. The default value is Auto.

预测/预加标签

备注：

The Predict feature relies on the UiPath Helix Extractor, but only for tenants based in the Europe region. If your tenant is located in a region outside of Europe, this functionality uses the previous-generation model architecture.

如果您已经有一个模型可以提取一些需要加标签的字段，并且只有少量额外的字段需要手动加标签，则可以使用 Document Manager 的预加标签功能来节省时间。

预加标签合并预加标签端点（在“预加标签”设置中配置）和生成式预测的结果。

如果未配置预加标签端点，则仅使用生成式预测来预测所有字段。

可用的选项如下：

预加标签 URL

预加标签要求 ML 模型具有 URL。查看您可以使用的以下 URL：

• Public URLs such as https://du.uipath.com/ie/invoices or https://du.uipath.com/ie/purchase_orders. Visit Public endpoints for Automation Cloud and Test Cloud to check the full list of endpoints.
• 已在本地部署的 AI Center 或 Cloud AI Center 中公开的 ML 技能 URL。

预加标签密钥

AI Unit/Document Understanding API 密钥。预加标签 API 密钥是技能所在组织的 Document Understanding 密钥。

Cloud Document Manager 和本地部署的 Online Document Manager 必须使用此密钥。对于本地部署的离线 Document Manager，此密钥是可选的。

如何…

The How to... option opens the Document Manager help menu.

在帮助菜单，您可以找到以下信息：

• The Document Manager version.
• 指向此文档页面的“文档”链接。
• “标签控件”部分显示处理数据时要使用的控件。
• “文档快捷方式”部分显示用于执行各种操作（例如导航和用户界面缩放）的快捷方式。
• “配置”部分显示有关在安装过程中执行的实例配置的详细信息。

标签控件

命令	描述
左键单击	选择复选框。如果与字段的热键一起使用，则会将所选信息分配给字段。
退格/删除	删除字段的标记值。
右击	显示 OCR 文本和当前标签。
Enter 或反斜杠	对跨越多行文本的表格行进行分组或取消分组。

文档快捷方式

快捷方式	描述
Alt + 向左箭头/向右箭头	在文档之间导航。
Alt + Delete	删除或恢复文档。
Ctrl + 滚动	通过放大或缩小更改文档缩放比例。

列字段

列字段具有以下选项：

• 新建列字段
• 编辑字段
• 展开/折叠列字段值

For more details on column fields, visit this section.

常规字段

常规字段具有以下选项：

• 新建常规字段
• 编辑字段

For more details on regular fields, visit this section.

分类字段

分类字段具有以下选项：

• 新建分类字段
• 编辑字段

For more details on classification fields, visit this section.

文档视图

对于多页文档，您可以像在任何 PDF 查看器中一样自然地滚动页面。要放大或缩小，请使用 Ctrl 并滚动鼠标滚轮。

您可以通过选择文字框并按下相应的按键，将文档分配给某个字段，从而标记文档。您也可以右键单击文字框并验证提取的信息。

For more details on how to label documents, visit this page.

打开新的 Document Manager 会话或筛选器为空时，某些准则将显示在文档视图中：

此外，文档视图中还会显示加载失败：