Namespace NOpenNLP.Tools.Ml.Model

Classes

AbstractDataIndexer

Abstract class for collecting event and context counts used in training.

AbstractModel
AbstractModelReader
AbstractModelWriter
BinaryFileDataReader
ComparableEvent

A maxent event representation which we can use to sort based on the predicates indexes contained in the events.

ComparablePredicate

A maxent predicate representation which we can use to sort based on the outcomes. This allows us to make the mapping of features to their parameters much more compact.

Context

Class which associates a real valued parameter or expected value with a particular contextual predicate or feature. This is used to store maxent model parameters as well as model and empirical expected values.

DataIndexerFactory
DynamicEvalParameters
EvalParameters

This class encapsulates the varibales used in producing probabilities from a model and facilitaes passing these variables to the eval method.

Event

The context of a decision point during training. This includes contextual predicates and an outcome.

FileEventStream

Class for using a file of events as an event stream. The format of the file is one event per line with each line consisting of outcome followed by contexts (space delimited).

GenericModelReader
GenericModelWriter
HashSumEventStream
MutableContext

Class used to store parameters or expected values associated with this context which can be updated or assigned.

ObjectDataReader
OnePassDataIndexer

An indexer for maxent model data which handles cutoffs for uncommon contextual predicates and provides a unique integer index for each of the predicates.

OnePassRealValueDataIndexer

An indexer for maxent model data which handles cutoffs for uncommon contextual predicates and provides a unique integer index for each of the predicates and maintains event values.

PlainTextFileDataReader
RealValueFileEventStream
SequenceStreamEventStream<T>

Class which turns a sequence stream into an event stream.

Sequence<T>

Class which models a sequence.

TwoPassDataIndexer

Collecting event and context counts by making two passes over the events. The first pass determines which contexts will be used by the model, and the second pass creates the events in memory containing only the contexts which will be used. This greatly reduces the amount of memory required for storing the events. During the first pass a temporary event file is created which is read during the second pass.

UniformPrior

Provide a maximum entropy model with a uniform prior.

Interfaces

IDataIndexer

Object which compresses events in memory and performs feature selection.

IDataReader
IMaxentModel

Interface for maximum entropy models.

IPrior

This interface allows one to implement a prior distribution for use in maximum entropy model training.

ISequenceClassificationModel<T>

A classification model that can label an input sequence.

ISequenceStream<T>

Interface for streams of sequences used to train sequence models.

Enums

AbstractModel.ModelType