Table of Contents

Constructor NamedEntityRecognition

Namespace
LMKit.TextAnalysis
Assembly
LM-Kit.NET.dll

NamedEntityRecognition(LM, bool)

Initializes a new instance of NamedEntityRecognition with a default set of entity definitions (all built-in NamedEntityRecognition.NamedEntityType values except Custom and, by default, Other).

public NamedEntityRecognition(LM model, bool includeOtherType = false)

Parameters

model LM

A valid LM model that is not an embedding-only model and has tensor weights loaded.

includeOtherType bool

When true, also includes the catch-all Other in the default definitions. Use this to discover unexpected entities that do not fit any predefined type and for which you have not created a specific Custom label. Defaults to false.

Examples

// Precision-oriented (default): do not include the catch-all bucket
var ner = new NamedEntityRecognition(model);

// Discovery mode (higher recall, potentially more false positives)
var nerWithOther = new NamedEntityRecognition(model, includeOtherType: true);

Remarks

About Other: This is a catch-all bucket intended for discovery of entities that are **not** covered by predefined types (e.g., Person, Location, Organization) and for which you have **not** provided a dedicated custom label. Examples might include domain-specific identifiers or uncommon categories that appear in your data.

Because it is intentionally broad, enabling Other can increase recall at the cost of a higher false-positive rate. When you know the category name you want, prefer defining a Custom label instead of relying on Other.

Exceptions

ArgumentNullException

Thrown if model is null.

InvalidModelException

Thrown if model is an embedding model or was loaded without weights.

NamedEntityRecognition(LM, List<EntityDefinition>)

Initializes a new instance of NamedEntityRecognition with a custom list of entity definitions.

public NamedEntityRecognition(LM model, List<NamedEntityRecognition.EntityDefinition> definitions)

Parameters

model LM

A valid LM model that is not an embedding‐only model and has tensor weights loaded.

definitions List<NamedEntityRecognition.EntityDefinition>

The list of NamedEntityRecognition.EntityDefinition items (built‐in or custom) that this recognizer will extract. Must be non-null and contain at least one item.

Exceptions

ArgumentNullException

Thrown if model is null or definitions is null/empty.

InvalidModelException

Thrown if model is an embedding model or was loaded without weights.