Individual

Combines the results from all extractors into a single result set.

Remarks

This collation provider acts as an aggregator, and produces a result set containing all results from all extractors.

One of the primary uses for this collation provider is to capture data values which appear in multiple presentation formats. For example, in a Data Type designed to capture dates, each extractor might match a different date/time format. (i.e. '01/01/2000', 'January 1, 2000', '01-JAN-2000', etc.) In a Data Type designed to match the label for an Invoice Date field, each extractor might match a different variation of the field label.

Another common use for this collation provider is to attempt extraction of the same value using multiple extraction methods. As an example, consider a structured form where the Effective Date appears to the right of a text label "Effective Date:", and just below the section heading "Contract Information". In this case, Extractor A would attempt to find a date to the right of "Effective Date:", while Extractor B searches for dates just below "Contract Information". This way, if Extractor A fails for some reason, Extractor B may succeed. In cases where both succeed, the Deduplication property found on Data Type objects can be configured to select 1 result as the winner.

Other known uses include the following:

• When extracting features for Lexical classification, it is useful for aggregating multiple feature types into a single list of features.
• When creating classification rules for a Document Type, it is useful as an "OR" operator, where each extractor defines a different rule.

Properties

Name	Type		Description
Output Feature Name	Boolean	►	If this value is true, a feature name will be output in place of the matched value. DefaultFalse This option is useful when extracting non-word classification features. As an example, consider the scenario where a Vehicle Fleet Inventory document contains no text other than a list of VIN numbers. With no keywords present on the document, the only way to properly classify the document is to recognize that it contains a high frequency of VIN numbers. Since it would be impractical to train Grooper with an example of every possible VIN number, what is needed is an extractor which will match VIN numbers on the document, and output a token such as "feature_vin_number" to the classification engine. In this way, a document containing 25 VIN numbers will be treated as though it contained 25 separate occurrences of the phrase "feature_vin_number".
Head Adjust	Int32	►	Specifies an adjustment to the starting position of each match. Default0 Values can be positive or negative. Positive values move forward in the document content. Negative values move backward through the document content.
Tail Adjust	Int32	►	Specifies an adjustment to the ending position of each match. Default0 Values can be positive or negative. Positive values move forward in the document content. Negative values move backward through the document content.

Used By

Data Type