Please also read Part I of this blog 'Leverage the power of Low Code and AI to automate your document processing' written by Allen Chan, CTO of IBM Digital Business Automation.
IBM's Automation Document Processing is a set of AI-powered services that can read unstructured documents, refine data and apply the data to business applications and workflow. The overall application building flow can be described in 2 phases; design phase and build & run phase.
In the design phase, you can use our low-code designer geared towards business users to set up the document model for machine learning based document classification and deep learning data extraction without being a data scientist. You would be able to easily configure and use pre-defined deep learning model for data extraction for common document types such as invoice, utility bills and bill of lading which can be used as a base to train with your own set of samples.
When it comes to categorizing and classifying documents, there is no need to provide manual input anymore. All you need to do is upload your sample set of documents to use system recommended ML-models. How easy is that!
Additionally, you can also add more context to the important business data by creating composite objects to group important sets of business data to bring more meaning to the extracted data. You can also standardize your data by defining data formatters and converters so data appears the way you have defined because we understand having clean and standardized data is crucial in business processes.
From the build and run standpoint, we provide a low-code tooling and out-of-the-box templates that allow business users to quickly build, customize and deploy their applications without having to build something from the ground up. We also provide AI-powered data correction where data is automatically refined based on the configuration that is set during design time. Based on the user's preferred confidence threshold, data validation is needed only when the field level threshold is below a certain point so that knowledge workers can spend less time doing manual validation and more time focusing on meaningful tasks.
Automation Document Processing can help companies automate various high value applications such as automated utility bill payment by enabling a new way of paying utility bills where banking customers can simply email their utility bill to their bank and have the bank automatically make the payment for them. In this process flow shown in the video, Automation Document Processing plays a crucial role in accurately classifying document types and extracting important bill payment data such as due date, total amount due and service address. Using validation logic set at design time, data clerks only need to look at fields that fail the validation instead of all fields. Using our pre-defined deep learning model already trained with various types of utility bills, organizations would be able to see significant cost and time savings where they can use the provided model as a base for their own model.
For additional information, please visit us at https://www.ibm.com/automation/data-capture
#ArtificialIntelligence(AI)#AutomationDocumentProcessing#CloudPakforBusinessAutomation#DeepLearning#FileNet