Knowledge Processing uses a separate database that contains layout information about different kinds of documents. When the ReadSoft Knowledge Processing Service is running, documents entering the system are matched against existing layouts.
Thus, Knowledge Processing is used to learn classification and interpretation. While processing documents in XBOUND, the system continuously learns how to classify and interpret information.
If a matching layout is found in the Knowledge Processing database, that information is used to classify documents and/or to find and interpret fields on the documents. Any corrections made while operators verify the fields result in adjustments in the Knowledge Processing database.
If no matching layout is found, and if a Knowledge Processing step is included in the process, verifying fields results in saving new layout information in the Knowledge Processing database. Documents are then much more likely to be classified automatically, and fields are much more likely to be found and interpreted automatically, the next time that kind of document is processed.
Knowledge Processing results in better interpretation of machine-printed character fields, date fields, and amount fields. It is recommended in solutions that include semi-structured documents.
Some kinds of documents are not suitable for Knowledge Processing. For example, handwritten correspondence should not be classified by Knowledge Processing. Set up such documents to circumvent Knowledge Processing.
If you want to classify documents such as these automatically, you must add classifiers using the Capture Components Administration module (see Capture Components Administration Help for instructions). Then reimport the solution.
In a solution that is set up to use both Knowledge Processing and those classifiers, the system will first try to identify documents using classifiers from the imported solution. If classification succeeds with at least one Certain match, then Knowledge Processing is not used.
In the Capture Components Administration module, enable Knowledge Processing (Capture Components Administration Help topic).
Optional: In the Capture Components Administration module, "learn" classification offline.
In the XBOUND Process Designer, create an XBOUND process that uses Knowledge Processing.
Optional: Route specific kinds of document to avoid Knowledge Processing.
You can:
Verify fields using Knowledge Processing in RCF Verification (Capture Framework Help topic)
Check whether a document was interpreted using Knowledge Processing in RCF Verification (Capture Framework Help topic)
Troubleshooting: Connection with the Knowledge Processing Service is lost
Finding and deactivating LDPs that are corrupt or learned the wrong document class
Troubleshooting: Knowledge Processing mixes up two documents
Under what circumstances are document classes, field values, and field positions learned?