Helpful Terms - Web Verifier - Foundation 23.1 - Foundation 23.1 - Brainware - external

Brainware Intelligent Capture Web Verifier

Web Verifier
Foundation 23.1


Accumulated Learnset
A common learnset is known as a Accumulated Learnset.
Automatic Supervised Learning
This process uses the Associated Search Engine to process, classify, and extract information.
Base class
The highest level of a classification.
A batch is a stack of electronic documents. The electronic documents may be paper-based or files created using applications.
Brainware Table Extraction
An extraction method that facilitates interactive table training.
Set of possible values for a field.
Child class
A class spawned by a parent class. Also called a sub-class.
A set of documents that are grouped by common content. Each class usually has a mnemonic name that describes its contents.
Common Learnset
An accumulation of local learnsets.
A parent document class.
Any electronic file mainly consisting of ASCII text. If this is initially the case, OCR or filtering must be applied to create the text representation classified, have fields used for extraction, and have one or more images attached.
The system no longer manages documents released by the export function.
A logical structure inside a batch for coherent documents. For example, a folder may consist of all pages of a correspondence with many folders inside one batch.
Global Learnset
A general learnset that encompasses similar classes or projects. See also Local Learnset.
Given a view with a set of documents in vector representation and their class assignment, a neural network is created, so that the defined classes can be reproduced without error. This neural network is the used in all subsequent classification tasks.
In classification, a learnset is a set of documents whose class assignments are specified by the user. For each view and each class, the user must provide a sufficient number of representative documents. Similarly, in extraction, a learnset is a set of documents whose field contents are selected by the user from a set of candidates.
Local Learnset
A learnset specific to a document class.
Neural network
An artificial neural network is an application that in some ways works like a human brain. This includes the ability to learn. It consists of artificial neurons that are linked into a network of layers. The neural network can receive signals through an input layer, process it within the internal layers, and send signals through the output layer. During learning, a specified input (called a teacher signal, such as documents from a Learnset) and the desired output (such as the corresponding classes) are presented to the network together. Processing is then adjusted until the desired output can be produced from the teacher signal.
Optical Character Recognition (OCR)
The reading and recognition of symbols of text from a piece of paper or a scanned image. OCR detects the symbols and converts them into characters and words that can be read electronically.
Parent class
A class with derived class, called children.
A collection of customer settings for applications. Projects are created in Verifier and sent to Runtime Server for productive operation.
Smart Indexing
Smart indexing uses a database lookup to determine document attributes. It can be used for automatic indexing and to support manual indexing.
A derivative class. Also called a child class.
An internal structure representing the logical structure of a document. The Workdoc represents the data created during processing of a single document and contains all OCR and analysis results.


Analysis means the document content is analyzed and a set of possible values for a field are generated called candidates.
Classification means assigning one or more classes and corresponding confidence values to one or more unknown documents.
Evaluation means determining a class or the contents of a field from confidence levels, weights, or distances for classes or candidates.
Extraction means automatic document indexing and its information that helps to sort, file and search documents.
Importing means bringing documents into the system for management and processing.
Indexing means assigning attributes to a document. This can either be done manually, semi-automatically (Smart Indexing), or entirely automatically (Extraction).
Validation involves confirming whether a processing result is correct.
Verification is related to quality assurance. It involves taking a processed document, checking the processing results, and correcting any errors.