The Problem – Born in the mailroom scanning paper
Most legacy capture vendors began focused on the electronic capture of paper. Focused on the automation of paper capture in the mailroom, legacy scanning vendors focused on:
- Scanning and Recognition – Components that are unique for mailroom scanning include scanning batches of documents, separator pages, bar-code reading, Optical Character Recognition as well as hand writing recognition.
- Indexing – Screens for indexing of documents. Based for paper, results are around recognition of characters and fields and leveraging confidence levels for keying of data.
- Bulk Ingestion into ECM and Data platforms – Growing up as a separate infrastructure, vendors would have specific adapters for where the content would flow to after it was indexed. For example, Captiva has adapters for Documentum and Application Extender along with many others before being acquired by Documentum.
Next generation capture solutions need to do all of the above and more embracing the affordability and accessibility of limitless computing power with technology like Machine Learning/Artificial Intelligence as well as cloud capabilities.
The Solution: OpenCapture – intelligent capture with machine learning
Legacy capture tools generally rely on two approaches to data capture:
- Location Template Approach (example – DataCap, Kofax, InputAccel)– a template defines where data is located in a given document. A zone is given to denote where a piece of data resides. For example, the tool could be told to look in a given box in the top right corner of the header to pull the “Report Number” value. This approach only works well when the positional data is known and very consistent across all documents. Templates need to be created for every type of captured document.
- Key/Value pair Template Approach (example – Ephesoft– A second approach is to provide a Key/Value pair template. In this approach, instead of defining the zonal position of the data, the tool is told to look for a given key, for example: “Invoice Number”, and then the tool will look at surrounding text to pull the value – for example, preferring text to the left or underneath the key. This approach works well when the target data may be anywhere within the document, but runs into problems when the Key text is inconsistent. Using our invoice example, some vendors may display Invoice Number as Invoice Num, Invoice Nbr, Invoice #, etc. Existing Capture tools have approaches for minimizing this problem, but it is still an issue for many clients.
Both approaches are typically augmented with additional processing to look up and verify sources against other systems (example PO number, account number….). This processing can include both configuration and customization depending on requirements.
OpenCapture combines the above approaches while adding Machine Learning to address handling incorrect data extraction that is corrected by the user during indexing. For legacy tools, an error that is manually corrected on one document will continue to be an error on the next, similar document unless the algorithm or template is changed. OpenCapture leverages machine learning to correct the template/approach to gradually reduce the indexing effort for subsequent documents. Current capture tools require a manual administrative update to the template or an entirely new template. In reality, this means that templates aren’t updated for most corrected extraction mistakes leading to user frustration.
Focusing on modern technologies, OpenCapture does more than just intelligently extract content, OpenCapture and Capture 2.0 will take into account machine learning to allow the indexing components to learn over time to achieve better results. View the video below or look at our blog for other information about OpenCapture