FACTS Alternative Labels

concept anchor label

• At least one concept anchor label is required for every Anchors

^Alternative label ^Domain ^Range ^

concept fact label

• At least one concept fact label is required for every Concept Fact

^Alternative label ^Domain ^Range ^

forced preclusion label

Preclusion stops the firing of any label which is wholly and exactly a part of another label. By default, preclusion is globally ‘on’ in FACTS.

This Alt Label type has the same behaviour as “Forced Preclusion”, which is the intentional addition of labels to a model that then generates desired preclusion without generating its own score. For a full description of preclusion, including “Forced Preclusion”, see this blog post: Preclusion Description

^Alternative label ^Domain ^Range ^

skip over repeat-breaking labels

This Alt Label type is only used for Repeated Fact contexts. It is designed to allow you to ignore intervening semantic units when the extractor is looking for successive firing sequences. One use case for this is when you want to extract from a bulleted list that may have some semantic units (e.g. sentences) between the bullets, and you don’t want those to stop the extractor.

^Alternative label ^Domain ^Range ^

wildcard anchor label

• At least one wildcard anchor label is required for every Wildcard Anchor

^Alternative label ^Domain ^Range ^

wildcard fact label

• At least one wildcard fact label is required for every Wildcard Fact

^Alternative label ^Domain ^Range ^