Near Ordered Contexts

Save PDF

Last Updated: May 13, 2026
4 minute read

Semaphore
Documentation

This is an abstract Concept Class (no concept can be made an instance of it in the model).

It represents all ordered near Context extractors. Ordered Near Contexts are extractors that look for facts (and anchors) that are close to each other and in a specific order in some context. How close they are can be specified using the count metadata.

Hierarchical relationships

Label	Range	Mandatory or Optional	Constraints
fact	Context Elements	Required	one or more allowed
anchor	Context Elements	Optional	one or more allowed
not anchor	Context Elements	Optional	one or more allowed
not fact	Context Elements	Optional	one or more allowed
optional anchor	Context Elements	Optional	one or more allowed
optional fact	Context Elements	Optional	one or more allowed

Associative relationships

Label	Range	Mandatory or Optional	Constraints
depended on by context	Contexts	Optional	only one allowed
depends on context	Contexts	Optional	only one allowed
extract fact as	Fact Name	Optional	only one allowed
group fact as	Fact Name	Optional	only one allowed
location	Locations	Optional	only one allowed
precluded by context	Contexts	Optional	only one allowed
precludes context	Contexts	Optional	only one allowed
Punctuation	Punctuation rules	Optional	only one allowed

Alternative labels

None

Metadata

Label	Range	Mandatory or Optional	Constraints
context position from document end	Integer	Optional	only one allowed
context position from document start	Integer	Optional	only one allowed
fact position from context end	Integer	Optional	only one allowed
fact position from context start	Integer	Optional	only one allowed
field	String	Optional	only one allowed
width for context	Integer	Optional	only one allowed
return group fact ID	Boolean	Optional	only one allowed
return raw text also	Boolean	Optional	only one allowed
return value using regex	String	Optional	only one allowed
{1-10} greedy repeat	Integer	Optional	one or more allowed
{1-10} non-greedy repeat	Integer	Optional	one or more allowed
debug	Boolean	Optional	only one allowed

Example model and tests

In our example, we will look for a taxonomy fact (link) of food types found between two anchors (link) of ‘Ingredient’ and ‘Preparation’ respectively. The context looks for near ordered across paragraphs. The convention is the same for other Near ordered contexts, the difference is the skip will apply to the range specified. A non greedy skip of 3 in a Near ordered across phrases in sentence(s) will skip up to 3 sentences, in a Near ordered across phrases in document, the skip will apply to words.

Our near ordered across paragraphs is ordered in this way:

Anchor of Ingredients
Taxonomy fact
Anchor of Procedure

With the completed concept looking thus:

With a near count of 15 set via the metadata property of ‘Near count’:

Thus, between 15 paragraphs, starting with ‘Ingredient’ and ending with ‘Preparation’, look for any taxonomy fact.

Positive test content

Using the example model, we can look for any item in the FoodTypes hierarchy from the Open Food Standards model, after the word ‘Ingredients’ and before ‘Preparation’. Therefore, where ‘Gin’ is a child of FoodTypes, for input text:

Cookbook

Ingredients

Tonic

Procedure

(Tonic is not returned as it is not in the FoodType hierarchy) Will return thus:

Negative test content

All of these texts below should not return anything:

Cookbook

Procedure

Drink the Gin

Ingredients

Tonic

The context looks for an order of 1) Ingredient 2) A taxonomy fact 3) Anchor of Procedure. Nothing is returned as the found text does not match the specified order. Although ‘gin’ is found it is not after ‘Ingredient’ and before ‘Procedure’

The Semaphore Fact Extraction Framework (FACTS)