Artificial intelligence is used to train Optical Character Recognition (OCR) systems for improved accuracy and the identification of common data elements based on context.
Artificial Intelligence allows data capture applications to quickly determine what type of document is being processed before extracting data from the OCR text.
AI algorithms use text matching, page layouts, and artificial intelligence to train models that are able to identify documents by type even when the formatting and quality varies significantly.
A good example of Artificial Intelligence based automatic classification is the LoanStacker application, which takes a complete residential mortgage loan file and identifies the more than 500 forms, disclosures, tax records, and contracts they contain. Once identified these documents can sent to the appropriate workflows for approval, data entry, etc.
While most data capture applications are able to identify document types based on recognition templates, AI classification algorithms are much faster and significantly improve throughput when there are many different types of documents being processed. Trained Artificial Intelligence classification models can also seem to “understand” the common traits of different document types and sort them correctly even when presented with new formats.
Simple Software’s SimpleIndex application provides keyword and pattern matching based document classification at a much lower cost than enterprise solutions.
Our collection of OCR Data Capture applications all have built-in automatic document classification capabilities, including machine learning and script based manual overrides.
Artificial Intelligence OCR training refers to one of two things:
- Tuning the OCR Engine to improve recognition of new fonts, languages, or handwritten text.
- Training data capture software to identify the correct location of fields on various related documents.
The first type of Artificial Intelligence OCR training was a commonly used feature in early desktop OCR applications. However, modern OCR engines are trained on huge sample sets during development and manual training is more likely to decrease accuracy than anything. It is rarely used outside of the development environment.
The second type of OCR training is used by enterprise data capture applications to automate the creation of recognition templates. The most common application is accounts payable invoices, where every vendor has their own layout and formatting but share the same data fields. These systems can “learn” from user feedback, improving the recognition accuracy until all fields are consistently read.
Field position training usually starts with a generic template that can identify the fields using the most common labels. Whenever a field is missed or read in the wrong position, the user highlights the correct field position on that document during a manual review. The new position is used by the machine learning algorithm to generate an updated template that correctly identifies the fields on that sample. If the document has consistent layout and decent image quality the template can be trained after just 2-3 samples.
More complex documents and documents that have a lot of layout variation can take many samples to train, and sometimes they can fail to train altogether. AI OCR training is not magic, and there will always be some cases where it is unable to consistently read a document correctly. If 100% accuracy is needed for these documents then it is important to choose a data capture platform that offers the ability to manually override the OCR training.
Many newer OCR systems no longer offer the ability to manually create templates and rely fully on the machine learning function. While these systems can be easier to configure, they will never reach the level of accuracy that can be achieved by one that offers a manual override.
There are also many kinds of documents that can be easily parsed with simple pattern matching, or where an experienced user can create a template that works perfectly in just a few hours. This can save a lot of time, user frustration, and licensing costs compared to machine learning. It is important to know when AI OCR training is really needed, and the experts at Simple Software can help.