notice

This is documentation for Rasa Open Source Documentation v2.0.x, which is no longer actively maintained.
For up-to-date documentation, see the latest version (2.1.x).

Version: 2.0.x

rasa.nlu.test

log_evaluation_table

log_evaluation_table(report: Text, precision: float, f1: float, accuracy: float) -> None

Log the sklearn evaluation metrics.

remove_empty_intent_examples

remove_empty_intent_examples(intent_results: List[IntentEvaluationResult]) -> List[IntentEvaluationResult]

Remove those examples without an intent.

Arguments:

  • intent_results - intent evaluation results
  • Returns - intent evaluation results

remove_empty_response_examples

remove_empty_response_examples(response_results: List[ResponseSelectionEvaluationResult]) -> List[ResponseSelectionEvaluationResult]

Remove those examples without a response.

Arguments:

  • response_results - response selection evaluation results
  • Returns - response selection evaluation results

drop_intents_below_freq

drop_intents_below_freq(training_data: TrainingData, cutoff: int = 5) -> TrainingData

Remove intent groups with less than cutoff instances.

Arguments:

  • training_data - training data
  • cutoff - threshold
  • Returns - updated training data

write_intent_successes

write_intent_successes(intent_results: List[IntentEvaluationResult], successes_filename: Text) -> None

Write successful intent predictions to a file.

Arguments:

  • intent_results - intent evaluation result
  • successes_filename - filename of file to save successful predictions to

write_intent_errors

write_intent_errors(intent_results: List[IntentEvaluationResult], errors_filename: Text) -> None

Write incorrect intent predictions to a file.

Arguments:

  • intent_results - intent evaluation result
  • errors_filename - filename of file to save incorrect predictions to

write_response_successes

write_response_successes(response_results: List[ResponseSelectionEvaluationResult], successes_filename: Text) -> None

Write successful response selection predictions to a file.

Arguments:

  • response_results - response selection evaluation result
  • successes_filename - filename of file to save successful predictions to

write_response_errors

write_response_errors(response_results: List[ResponseSelectionEvaluationResult], errors_filename: Text) -> None

Write incorrect response selection predictions to a file.

Arguments:

  • response_results - response selection evaluation result
  • errors_filename - filename of file to save incorrect predictions to

plot_attribute_confidences

plot_attribute_confidences(results: Union[
List[IntentEvaluationResult], List[ResponseSelectionEvaluationResult]
], hist_filename: Optional[Text], target_key: Text, prediction_key: Text, title: Text) -> None

Create histogram of confidence distribution.

Arguments:

  • results - evaluation results
  • hist_filename - filename to save plot to
  • target_key - key of target in results
  • prediction_key - key of predictions in results
  • title - title of plot

plot_entity_confidences

plot_entity_confidences(merged_targets: List[Text], merged_predictions: List[Text], merged_confidences: List[float], hist_filename: Text, title: Text) -> None

Create histogram of confidence distribution.

Arguments:

  • results - evaluation results
  • hist_filename - filename to save plot to
  • target_key - key of target in results
  • prediction_key - key of predictions in results
  • title - title of plot

evaluate_response_selections

evaluate_response_selections(response_selection_results: List[ResponseSelectionEvaluationResult], output_directory: Optional[Text], successes: bool, errors: bool, disable_plotting: bool) -> Dict

Creates summary statistics for response selection.

Only considers those examples with a set response. Others are filtered out. Returns a dictionary of containing the evaluation result.

Arguments:

  • response_selection_results - response selection evaluation results
  • output_directory - directory to store files to
  • successes - if True success are written down to disk
  • errors - if True errors are written down to disk
  • disable_plotting - if True no plots are created
  • Returns - dictionary with evaluation results

evaluate_intents

evaluate_intents(intent_results: List[IntentEvaluationResult], output_directory: Optional[Text], successes: bool, errors: bool, disable_plotting: bool) -> Dict

Creates summary statistics for intents.

Only considers those examples with a set intent. Others are filtered out. Returns a dictionary of containing the evaluation result.

Arguments:

  • intent_results - intent evaluation results
  • output_directory - directory to store files to
  • successes - if True correct predictions are written to disk
  • errors - if True incorrect predictions are written to disk
  • disable_plotting - if True no plots are created
  • Returns - dictionary with evaluation results

merge_labels

merge_labels(aligned_predictions: List[Dict], extractor: Optional[Text] = None) -> List[Text]

Concatenates all labels of the aligned predictions.

Takes the aligned prediction labels which are grouped for each message and concatenates them.

Arguments:

  • aligned_predictions - aligned predictions
  • extractor - entity extractor name
  • Returns - concatenated predictions

merge_confidences

merge_confidences(aligned_predictions: List[Dict], extractor: Optional[Text] = None) -> List[float]

Concatenates all confidences of the aligned predictions.

Takes the aligned prediction confidences which are grouped for each message and concatenates them.

Arguments:

  • aligned_predictions - aligned predictions
  • extractor - entity extractor name
  • Returns - concatenated confidences

substitute_labels

substitute_labels(labels: List[Text], old: Text, new: Text) -> List[Text]

Replaces label names in a list of labels.

Arguments:

  • labels - list of labels
  • old - old label name that should be replaced
  • new - new label name
  • Returns - updated labels

write_incorrect_entity_predictions

write_incorrect_entity_predictions(entity_results: List[EntityEvaluationResult], merged_targets: List[Text], merged_predictions: List[Text], error_filename: Text) -> None

Write incorrect entity predictions to a file.

Arguments:

  • entity_results - response selection evaluation result
  • merged_predictions - list of predicted entity labels
  • merged_targets - list of true entity labels
  • error_filename - filename of file to save incorrect predictions to

collect_incorrect_entity_predictions

collect_incorrect_entity_predictions(entity_results: List[EntityEvaluationResult], merged_predictions: List[Text], merged_targets: List[Text])

Get incorrect entity predictions.

Arguments:

  • entity_results - entity evaluation results
  • merged_predictions - list of predicted entity labels
  • merged_targets - list of true entity labels
  • Returns - list of incorrect predictions

write_successful_entity_predictions

write_successful_entity_predictions(entity_results: List[EntityEvaluationResult], merged_targets: List[Text], merged_predictions: List[Text], successes_filename: Text) -> None

Write correct entity predictions to a file.

Arguments:

  • entity_results - response selection evaluation result
  • merged_predictions - list of predicted entity labels
  • merged_targets - list of true entity labels
  • successes_filename - filename of file to save correct predictions to

collect_successful_entity_predictions

collect_successful_entity_predictions(entity_results: List[EntityEvaluationResult], merged_predictions: List[Text], merged_targets: List[Text])

Get correct entity predictions.

Arguments:

  • entity_results - entity evaluation results
  • merged_predictions - list of predicted entity labels
  • merged_targets - list of true entity labels
  • Returns - list of correct predictions

evaluate_entities

evaluate_entities(entity_results: List[EntityEvaluationResult], extractors: Set[Text], output_directory: Optional[Text], successes: bool, errors: bool, disable_plotting: bool) -> Dict

Creates summary statistics for each entity extractor.

Logs precision, recall, and F1 per entity type for each extractor.

Arguments:

  • entity_results - entity evaluation results
  • extractors - entity extractors to consider
  • output_directory - directory to store files to
  • successes - if True correct predictions are written to disk
  • errors - if True incorrect predictions are written to disk
  • disable_plotting - if True no plots are created
  • Returns - dictionary with evaluation results

is_token_within_entity

is_token_within_entity(token: Token, entity: Dict) -> bool

Checks if a token is within the boundaries of an entity.

does_token_cross_borders

does_token_cross_borders(token: Token, entity: Dict) -> bool

Checks if a token crosses the boundaries of an entity.

determine_intersection

determine_intersection(token: Token, entity: Dict) -> int

Calculates how many characters a given token and entity share.

do_entities_overlap

do_entities_overlap(entities: List[Dict]) -> bool

Checks if entities overlap.

I.e. cross each others start and end boundaries.

Arguments:

  • entities - list of entities
  • Returns - true if entities overlap, false otherwise.

find_intersecting_entities

find_intersecting_entities(token: Token, entities: List[Dict]) -> List[Dict]

Finds the entities that intersect with a token.

Arguments:

  • token - a single token
  • entities - entities found by a single extractor
  • Returns - list of entities

pick_best_entity_fit

pick_best_entity_fit(token: Token, candidates: List[Dict[Text, Any]]) -> Optional[Dict[Text, Any]]

Determines the best fitting entity given intersecting entities.

Arguments:

  • token - a single token
  • candidates - entities found by a single extractor
  • attribute_key - the attribute key of interest

Returns:

the value of the attribute key of the best fitting entity

determine_token_labels

determine_token_labels(token: Token, entities: List[Dict], extractors: Optional[Set[Text]] = None, attribute_key: Text = ENTITY_ATTRIBUTE_TYPE) -> Text

Determines the token label for the provided attribute key given entities that do not overlap.

Arguments:

  • token - a single token
  • entities - entities found by a single extractor
  • extractors - list of extractors
  • attribute_key - the attribute key for which the entity type should be returned

Returns:

entity type

determine_entity_for_token

determine_entity_for_token(token: Token, entities: List[Dict[Text, Any]], extractors: Optional[Set[Text]] = None) -> Optional[Dict[Text, Any]]

Determines the best fitting entity for the given token, given entities that do not overlap.

Arguments:

  • token - a single token
  • entities - entities found by a single extractor
  • extractors - list of extractors

Returns:

entity type

do_extractors_support_overlap

do_extractors_support_overlap(extractors: Optional[Set[Text]]) -> bool

Checks if extractors support overlapping entities

align_entity_predictions

align_entity_predictions(result: EntityEvaluationResult, extractors: Set[Text]) -> Dict

Aligns entity predictions to the message tokens.

Determines for every token the true label based on the prediction targets and the label assigned by each single extractor.

Arguments:

  • result - entity evaluation result
  • extractors - the entity extractors that should be considered
  • Returns - dictionary containing the true token labels and token labels from the extractors

align_all_entity_predictions

align_all_entity_predictions(entity_results: List[EntityEvaluationResult], extractors: Set[Text]) -> List[Dict]

Aligns entity predictions to the message tokens for the whole dataset using align_entity_predictions.

Arguments:

  • entity_results - list of entity prediction results
  • extractors - the entity extractors that should be considered
  • Returns - list of dictionaries containing the true token labels and token labels from the extractors

get_eval_data

get_eval_data(interpreter: Interpreter, test_data: TrainingData) -> Tuple[
List[IntentEvaluationResult],
List[ResponseSelectionEvaluationResult],
List[EntityEvaluationResult],
]

Runs the model for the test set and extracts targets and predictions.

Returns intent results (intent targets and predictions, the original messages and the confidences of the predictions), response results ( response targets and predictions) as well as entity results (entity_targets, entity_predictions, and tokens).

Arguments:

  • interpreter - the interpreter
  • test_data - test data
  • Returns - intent, response, and entity evaluation results

get_entity_extractors

get_entity_extractors(interpreter: Interpreter) -> Set[Text]

Finds the names of entity extractors used by the interpreter.

Processors are removed since they do not detect the boundaries themselves.

Arguments:

  • interpreter - the interpreter
  • Returns - entity extractor names

is_entity_extractor_present

is_entity_extractor_present(interpreter: Interpreter) -> bool

Checks whether entity extractor is present.

is_intent_classifier_present

is_intent_classifier_present(interpreter: Interpreter) -> bool

Checks whether intent classifier is present.

is_response_selector_present

is_response_selector_present(interpreter: Interpreter) -> bool

Checks whether response selector is present.

get_available_response_selector_types

get_available_response_selector_types(interpreter: Interpreter) -> List[Text]

Gets all available response selector types.

remove_pretrained_extractors

remove_pretrained_extractors(pipeline: List[Component]) -> List[Component]

Remove pre-trained extractors from the pipeline.

Remove pre-trained extractors so that entities from pre-trained extractors are not predicted upon parsing.

Arguments:

  • pipeline - the pipeline

Returns:

Updated pipeline

run_evaluation

run_evaluation(data_path: Text, model_path: Text, output_directory: Optional[Text] = None, successes: bool = False, errors: bool = False, component_builder: Optional[ComponentBuilder] = None, disable_plotting: bool = False) -> Dict

Evaluate intent classification, response selection and entity extraction.

Arguments:

  • data_path - path to the test data
  • model_path - path to the model
  • output_directory - path to folder where all output will be stored
  • successes - if true successful predictions are written to a file
  • errors - if true incorrect predictions are written to a file
  • component_builder - component builder
  • disable_plotting - if true confusion matrix and histogram will not be rendered
  • Returns - dictionary containing evaluation results

generate_folds

generate_folds(n: int, training_data: TrainingData) -> Iterator[Tuple[TrainingData, TrainingData]]

Generates n cross validation folds for given training data.

combine_result

combine_result(intent_metrics: IntentMetrics, entity_metrics: EntityMetrics, response_selection_metrics: ResponseSelectionMetrics, interpreter: Interpreter, data: TrainingData, intent_results: Optional[List[IntentEvaluationResult]] = None, entity_results: Optional[List[EntityEvaluationResult]] = None, response_selection_results: Optional[
List[ResponseSelectionEvaluationResult]
] = None) -> Tuple[IntentMetrics, EntityMetrics, ResponseSelectionMetrics]

Collects intent, response selection and entity metrics for cross validation folds.

If intent_results, response_selection_results or entity_results is provided as a list, prediction results are also collected.

Arguments:

  • intent_metrics - intent metrics
  • entity_metrics - entity metrics
  • response_selection_metrics - response selection metrics
  • interpreter - the interpreter
  • data - training data
  • intent_results - intent evaluation results
  • entity_results - entity evaluation results
  • response_selection_results - reponse selection evaluation results
  • Returns - intent, entity, and response selection metrics

cross_validate

cross_validate(data: TrainingData, n_folds: int, nlu_config: Union[RasaNLUModelConfig, Text], output: Optional[Text] = None, successes: bool = False, errors: bool = False, disable_plotting: bool = False) -> Tuple[CVEvaluationResult, CVEvaluationResult, CVEvaluationResult]

Stratified cross validation on data.

Arguments:

  • data - Training Data
  • n_folds - integer, number of cv folds
  • nlu_config - nlu config file
  • output - path to folder where reports are stored
  • successes - if true successful predictions are written to a file
  • errors - if true incorrect predictions are written to a file
  • disable_plotting - if true no confusion matrix and historgram plates are created

Returns:

dictionary with key, list structure, where each entry in list corresponds to the relevant result for one fold

compute_metrics

compute_metrics(interpreter: Interpreter, training_data: TrainingData) -> Tuple[
IntentMetrics,
EntityMetrics,
ResponseSelectionMetrics,
List[IntentEvaluationResult],
List[EntityEvaluationResult],
List[ResponseSelectionEvaluationResult],
]

Computes metrics for intent classification, response selection and entity extraction.

Arguments:

  • interpreter - the interpreter
  • training_data - training data
  • Returns - intent, response selection and entity metrics, and prediction results.

compare_nlu

compare_nlu(configs: List[Text], data: TrainingData, exclusion_percentages: List[int], f_score_results: Dict[Text, Any], model_names: List[Text], output: Text, runs: int) -> List[int]

Trains and compares multiple NLU models. For each run and exclusion percentage a model per config file is trained. Thereby, the model is trained only on the current percentage of training data. Afterwards, the model is tested on the complete test data of that run. All results are stored in the provided output directory.

Arguments:

  • configs - config files needed for training
  • data - training data
  • exclusion_percentages - percentages of training data to exclude during comparison
  • f_score_results - dictionary of model name to f-score results per run
  • model_names - names of the models to train
  • output - the output directory
  • runs - number of comparison runs
  • Returns - training examples per run

log_results

log_results(results: IntentMetrics, dataset_name: Text) -> None

Logs results of cross validation.

Arguments:

  • results - dictionary of results returned from cross validation
  • dataset_name - string of which dataset the results are from, e.g. test/train

log_entity_results

log_entity_results(results: EntityMetrics, dataset_name: Text) -> None

Logs entity results of cross validation.

Arguments:

  • results - dictionary of dictionaries of results returned from cross validation
  • dataset_name - string of which dataset the results are from, e.g. test/train