mindmeld.components.role_classifier module

This module contains the role classifier component of the MindMeld natural language processor.

class mindmeld.components.role_classifier.RoleClassifier(resource_loader, domain, intent, entity_type)

Bases: mindmeld.components.classifier.Classifier

A role classifier is used to determine the target role for entities in a given query. It is trained using all the labeled queries for a particular intent. The labels are the role names associated with each entity within each query.

domain

str -- The domain that this role classifier belongs to

intent

str -- The intent that this role classifier belongs to

entity_type

str -- The entity type that this role classifier is for

roles

set -- A set containing the roles which can be classified

dump(model_path, incremental_model_path=None)

Persists the trained role classification model to disk.

Parameters:

• model_path (str) -- The model path.
• incremental_model_path (str, Optional) -- The timestamped folder where the cached models are stored.

fit(queries=None, label_set=None, incremental_timestamp=None, load_cached=True, **kwargs)

Trains a statistical model for role classification using the provided training examples.

Parameters:

• queries (list of ProcessedQuery) -- The labeled queries to use as training data
• label_set (list, optional) -- A label set to load. If not specified, the default training set will be loaded.
• incremental_timestamp (str, optional) -- The timestamp folder to cache models in

inspect(query, gold_label=None, dynamic_resource=None)

load(model_path)

Loads the trained role classification model from disk.

Parameters:model_path (str) -- The location on disk where the model is stored

predict(query, entities, entity_index)

Predicts a role for the given entity using the trained role classification model.

Parameters:

• query (Query) -- The input query
• entities (list) -- The entities in the query
• entity_index (int) -- The index of the entity whose role should be classified

Returns:

The predicted role for the provided entity

Return type:

str

predict_proba(query, entities, entity_index)

Runs prediction on a given entity and generates multiple role hypotheses and associated probabilities using the trained role classification model.

Parameters:

• query (Query) -- The input query
• entities (list) -- The entities in the query
• entity_index (int) -- The index of the entity whose role should be classified

Returns:

a list of tuples of the form (str, float) grouping roles and their probabilities

Return type:

list

unload()

Unloads the model from memory. This helps reduce memory requirements while training other models.

view_extracted_features(query, entities, entity_index)

Extracts features for a given entity for role classification.

Parameters:

• query (Query or str) -- The input query
• entities (list) -- The entities in the query
• entity_index (int) -- The index of the entity whose role should be classified

Returns:

The extracted features from the given input

Return type:

dict

CLF_TYPE = 'role'