DocuWare Intelligent Document Processing (IDP) extends existing document-management solutions with AI-powered features. This enables a high level of automation in office processes.
Read more about DocuWare IDP in general
Training the artificial intelligence for IDP
As an initial step in setting up Intelligent Document Processing for your company, the artificial intelligence is trained to meet your needs.
This training includes the classification of documents and the extraction of relevant data from them:
Splitting: IDP recognizes the individual documents merged into one file, and separates them so they can be treated individually.
Classification: IDP recognizes specific document types, or document classes, such as invoices and delivery notes.
Extraction: IDP extracts relevant data from the documents and automatically adds them as index values.
The training of the AI behind IDP can be carried out in two ways:
DocuWare IDP Platform: Training is performed on the standalone IDP platform. This option supports any documents, whether or not they are archived in DocuWare. Training on the IDP platform is usually carried out by your DocuWare contact.
DocuWare Configurations: You can train splitting and classification workflows directly from the DocuWare Configurations, using documents that are already archived in your DocuWare file cabinets. This option is described in the section below.
You can connect DocuWare with these AI classification and extraction agents through an IDP configuration.
Training splitting and classification from DocuWare
You can train new splitting and classification models directly from the DocuWare Configurations, using documents already stored in your file cabinets. Because the training uses your actual documents, the resulting AI models are tailored to the specific formats, layouts, and content of your files.
To start a training:
In DocuWare Configurations, go to the DocuWare IDP section.
Click the button for training a new splitter or classifier. A dialog opens that guides you through the setup and displays the file cabinets available for training.
Select the file cabinets that contain the documents you want to use.
For splitting, select at least one file cabinet.
For classification, select at least two file cabinets. The dialog indicates whether each file cabinet contains enough documents for training.
Start the training. Training may take up to 24 hours to complete. You do not need to wait for the training to finish before creating and configuring the workflow.
Training for extraction models is not yet available through the DocuWare Configurations.
Processing email and documents with IDP
Intelligent Document Processing (IDP) can process documents as they enter the DocuWare system, whether they arrive by email or through the DocuWare Desktop Scan or Import plug-ins.
Importing email
To import emails automatically into DocuWare, you create an email import configuration in which you define where the messages will be stored and how their attachments are handled—for example, whether the attachments are archived with the email or saved as separate documents.
If you add an IDP configuration to this setup, DocuWare uses Intelligent Document Processing to classify every attached PDF and extract its key data before the files are archived, providing AI-driven automation for your email workflow.
Read more about configuring IDP for email import
Importing documents via DocuWare Desktop Apps
Documents added to DocuWare with the Desktop apps, whether via the Scan or Import plug-ins, can now be processed by IDP. For example, paper invoices captured with DocuWare Scan and existing PDF files brought in through the Import plug-in are automatically split, classified, indexed, and archived.
Read more about configuring IDP for DocuWare Desktop Apps
Validating IDP results
IDP returns a confidence value for every result. Results with a low confidence value can be sent to a user for confirmation instead of being processed automatically. This is called validation.
Validation is optional and is set separately for each processing step. Documents that need confirmation appear in the IDP Validations tab of the DocuWare Web Client. Only users with the Configure IDP right see this tab.
Validating splitting and classification
In DocuWare Configurations, go to the DocuWare IDP section. The setting applies to the whole organization, not to a single IDP configuration.
Validate if confidence is lower than: Sends a result to the IDP Validations tab if its confidence value stays below the percentage you enter. Results at or above the value are processed automatically. Clear the checkbox to process all results without manual validation.
Values from 1 to 99.999 are allowed. Set the threshold separately for splitting and for classification.
There is no recommended default value. A high value such as 95 is a useful starting point, because IDP tends to return either a high confidence value above 90 or a very low one. To find a suitable value for your documents, process a batch of about 100 typical documents and compare the confidence values that IDP returns.
Validating extraction
Validation for extraction is set per IDP configuration, because the field mapping is defined in each configuration. You switch it on while creating an IDP configuration, after an extraction model has been trained.
User validation required for AI results: Sends every extraction result to the IDP Validations tab. A confidence threshold is not available for extraction, so the setting is either on or off.
Read more about validating IDP results in the Web Client